Merge pull request #114 from HansKristian-Work/physical-srv-uav-access-chain

Generate more straight forward code for BDA descriptors.

67 files changed, 1593 insertions, 1998 deletions

diff --git a/dxil_converter.cpp b/dxil_converter.cpp
index 5a14eba..cf61fd0 100644
--- a/dxil_converter.cpp
+++ b/dxil_converter.cpp
@@ -700,6 +700,7 @@ spv::Id Converter::Impl::get_physical_pointer_block_type(spv::Id base_type_id, c
 		return entry.meta.coherent == meta.coherent &&
 		       entry.meta.nonreadable == meta.nonreadable &&
 		       entry.meta.nonwritable == meta.nonwritable &&
+		       entry.meta.size == meta.size &&
 		       entry.meta.stride == meta.stride &&
 		       entry.base_type_id == base_type_id;
 	});
@@ -753,9 +754,18 @@ spv::Id Converter::Impl::get_physical_pointer_block_type(spv::Id base_type_id, c
 
 	if (meta.stride > 0)
 	{
-		type_id = builder().makeRuntimeArray(type_id);
+		if (meta.size == 0)
+		{
+			type_id = builder().makeRuntimeArray(type_id);
+			type += "Array";
+		}
+		else
+		{
+			type_id = builder().makeArrayType(type_id, builder().makeUintConstant(meta.size / meta.stride),
+			                                  meta.stride);
+			type += "CBVArray";
+		}
 		builder().addDecoration(type_id, spv::DecorationArrayStride, meta.stride);
-		type += "Array";
 	}
 
 	spv::Id block_type_id = builder().makeStructType({ type_id }, type.c_str());
diff --git a/opcodes/converter_impl.hpp b/opcodes/converter_impl.hpp
index 353efdf..dde1c85 100644
--- a/opcodes/converter_impl.hpp
+++ b/opcodes/converter_impl.hpp
@@ -341,6 +341,7 @@ struct Converter::Impl
 		bool nonreadable;
 		bool coherent;
 		uint8_t stride;
+		uint32_t size;
 	};
 
 	struct ResourceMeta
diff --git a/opcodes/dxil/dxil_buffer.cpp b/opcodes/dxil/dxil_buffer.cpp
index 9bc9c46..442b77d 100644
--- a/opcodes/dxil/dxil_buffer.cpp
+++ b/opcodes/dxil/dxil_buffer.cpp
@@ -53,10 +53,10 @@ bool raw_access_byte_address_can_vectorize(Converter::Impl &impl, const llvm::Ty
 }
 
 bool raw_access_structured_can_vectorize(
-	Converter::Impl &impl, const llvm::Type *type,
-	const llvm::Value *index, const llvm::Value *byte_offset,
-	unsigned stride,
-	unsigned vecsize)
+		Converter::Impl &impl, const llvm::Type *type,
+		const llvm::Value *index, unsigned stride,
+		const llvm::Value *byte_offset,
+		unsigned vecsize)
 {
 	// vec3 vectorization requires scalar block layout always.
 	if (!impl.options.scalar_block_layout && vecsize == 3)
@@ -101,11 +101,11 @@ RawVecSize raw_access_structured_vectorize(
     const llvm::Value *byte_offset,
 	uint32_t mask)
 {
-	if (mask == 0xfu && raw_access_structured_can_vectorize(impl, type, index, byte_offset, stride, 4))
+	if (mask == 0xfu && raw_access_structured_can_vectorize(impl, type, index, stride, byte_offset, 4))
 		return RawVecSize::V4;
-	else if (mask == 0x7u && raw_access_structured_can_vectorize(impl, type, index, byte_offset, stride, 3))
+	else if (mask == 0x7u && raw_access_structured_can_vectorize(impl, type, index, stride, byte_offset, 3))
 		return RawVecSize::V3;
-	else if (mask == 0x3u && raw_access_structured_can_vectorize(impl, type, index, byte_offset, stride, 2))
+	else if (mask == 0x3u && raw_access_structured_can_vectorize(impl, type, index, stride, byte_offset, 2))
 		return RawVecSize::V2;
 	else
 		return RawVecSize::V1;
@@ -381,6 +381,41 @@ static spv::Id build_physical_pointer_address_for_raw_load_store(Converter::Impl
 	return emit_u32x2_u32_add(impl, ptr_id, byte_offset_id);
 }
 
+static spv::Id build_vectorized_physical_load_store_access(Converter::Impl &impl, const llvm::CallInst *instruction,
+                                                           unsigned vecsize, const llvm::Type *element_type)
+{
+	spv::Id ptr_id = impl.get_id_for_value(instruction->getOperand(1));
+	const auto &meta = impl.handle_to_resource_meta[ptr_id];
+	unsigned mask = (1u << vecsize) - 1u;
+
+	// If we can express this as a plain access chain, do so for clarity and ideally better perf.
+	// If we cannot do it trivially, fallback to raw pointer arithmetic.
+	bool can_vectorize = false;
+
+	if (meta.stride)
+	{
+		if (raw_access_structured_can_vectorize(impl, element_type,
+		                                        instruction->getOperand(2), meta.stride,
+		                                        instruction->getOperand(3), vecsize))
+		{
+			can_vectorize = true;
+		}
+	}
+	else if (raw_access_byte_address_can_vectorize(impl, element_type,
+	                                               instruction->getOperand(2), vecsize))
+	{
+		can_vectorize = true;
+	}
+
+	if (can_vectorize)
+	{
+		auto access = build_buffer_access(impl, instruction, 0, 0, element_type, mask);
+		return access.index_id;
+	}
+	else
+		return 0;
+}
+
 static bool emit_physical_buffer_load_instruction(Converter::Impl &impl, const llvm::CallInst *instruction,
                                                   const Converter::Impl::PhysicalPointerMeta &ptr_meta,
                                                   uint32_t mask = 0, uint32_t alignment = 0)
@@ -408,15 +443,26 @@ static bool emit_physical_buffer_load_instruction(Converter::Impl &impl, const l
 	}
 
 	auto *element_type = instruction->getType()->getStructElementType(0);
+	// If we can express this as a plain access chain, do so for clarity and ideally better perf.
+	// If we cannot do it trivially, fallback to raw pointer arithmetic.
+	spv::Id array_id = build_vectorized_physical_load_store_access(impl, instruction, vecsize, element_type);
+
 	spv::Id physical_type_id;
 	spv::Op value_cast_op;
 	get_physical_load_store_cast_info(impl, element_type, physical_type_id, value_cast_op);
 
 	if (vecsize > 1)
 		physical_type_id = builder.makeVectorType(physical_type_id, vecsize);
-	spv::Id ptr_type_id = impl.get_physical_pointer_block_type(physical_type_id, ptr_meta);
 
-	spv::Id u64_ptr_id = build_physical_pointer_address_for_raw_load_store(impl, instruction);
+	auto tmp_ptr_meta = ptr_meta;
+	tmp_ptr_meta.stride = array_id ? vecsize * get_type_scalar_alignment(impl, element_type) : 0;
+	spv::Id ptr_type_id = impl.get_physical_pointer_block_type(physical_type_id, tmp_ptr_meta);
+
+	spv::Id u64_ptr_id;
+	if (array_id)
+		u64_ptr_id = impl.get_id_for_value(instruction->getOperand(1));
+	else
+		u64_ptr_id = build_physical_pointer_address_for_raw_load_store(impl, instruction);
 
 	auto *ptr_bitcast_op = impl.allocate(spv::OpBitcast, ptr_type_id);
 	ptr_bitcast_op->add_id(u64_ptr_id);
@@ -425,6 +471,8 @@ static bool emit_physical_buffer_load_instruction(Converter::Impl &impl, const l
 	auto *chain_op = impl.allocate(spv::OpAccessChain, builder.makePointer(spv::StorageClassPhysicalStorageBuffer, physical_type_id));
 	chain_op->add_id(ptr_bitcast_op->id);
 	chain_op->add_id(builder.makeUintConstant(0));
+	if (array_id)
+		chain_op->add_id(array_id);
 	impl.add(chain_op);
 
 	auto *load_op = impl.allocate(spv::OpLoad, physical_type_id);
@@ -797,6 +845,11 @@ static bool emit_physical_buffer_store_instruction(Converter::Impl &impl, const
 	}
 
 	auto *element_type = instruction->getOperand(4)->getType();
+
+	// If we can express this as a plain access chain, do so for clarity and ideally better perf.
+	// If we cannot do it trivially, fallback to raw pointer arithmetic.
+	spv::Id array_id = build_vectorized_physical_load_store_access(impl, instruction, vecsize, element_type);
+
 	spv::Id physical_type_id;
 	spv::Op value_cast_op;
 	get_physical_load_store_cast_info(impl, element_type, physical_type_id, value_cast_op);
@@ -804,9 +857,16 @@ static bool emit_physical_buffer_store_instruction(Converter::Impl &impl, const
 	spv::Id vec_type_id = physical_type_id;
 	if (vecsize > 1)
 		vec_type_id = builder.makeVectorType(physical_type_id, vecsize);
-	spv::Id ptr_type_id = impl.get_physical_pointer_block_type(vec_type_id, ptr_meta);
 
-	spv::Id u64_ptr_id = build_physical_pointer_address_for_raw_load_store(impl, instruction);
+	auto tmp_ptr_meta = ptr_meta;
+	tmp_ptr_meta.stride = array_id ? vecsize * get_type_scalar_alignment(impl, element_type) : 0;
+	spv::Id ptr_type_id = impl.get_physical_pointer_block_type(vec_type_id, tmp_ptr_meta);
+
+	spv::Id u64_ptr_id;
+	if (array_id)
+		u64_ptr_id = impl.get_id_for_value(instruction->getOperand(1));
+	else
+		u64_ptr_id = build_physical_pointer_address_for_raw_load_store(impl, instruction);
 
 	auto *ptr_bitcast_op = impl.allocate(spv::OpBitcast, ptr_type_id);
 	ptr_bitcast_op->add_id(u64_ptr_id);
@@ -815,6 +875,8 @@ static bool emit_physical_buffer_store_instruction(Converter::Impl &impl, const
 	auto *chain_op = impl.allocate(spv::OpAccessChain, builder.makePointer(spv::StorageClassPhysicalStorageBuffer, vec_type_id));
 	chain_op->add_id(ptr_bitcast_op->id);
 	chain_op->add_id(builder.makeUintConstant(0));
+	if (array_id)
+		chain_op->add_id(array_id);
 	impl.add(chain_op);
 
 	spv::Id elems[4] = {};
diff --git a/opcodes/dxil/dxil_resources.cpp b/opcodes/dxil/dxil_resources.cpp
index bb8be9d..0ae5c98 100644
--- a/opcodes/dxil/dxil_resources.cpp
+++ b/opcodes/dxil/dxil_resources.cpp
@@ -27,6 +27,7 @@
 #include "logging.hpp"
 #include "opcodes/converter_impl.hpp"
 #include "spirv_module.hpp"
+#include "dxil_buffer.hpp"
 
 namespace dxil_spv
 {
@@ -1597,28 +1598,20 @@ static bool emit_cbuffer_load_physical_pointer(Converter::Impl &impl, const llvm
 {
 	auto &builder = impl.builder();
 
-	spv::Id member_index = impl.get_id_for_value(instruction->getOperand(2));
 	bool scalar_load = instruction->getType()->getTypeID() != llvm::Type::TypeID::StructTyID;
 	unsigned scalar_alignment;
-	spv::Id byteaddr_id;
 	uint32_t alignment;
 
 	const llvm::Type *result_component_type;
 
 	if (!scalar_load)
 	{
-		auto *mul_op = impl.allocate(spv::OpIMul, builder.makeUintType(32));
-		mul_op->add_id(member_index);
-		mul_op->add_id(builder.makeUintConstant(16));
-		impl.add(mul_op);
-		byteaddr_id = mul_op->id;
 		result_component_type = instruction->getType()->getStructElementType(0);
 		scalar_alignment = get_type_scalar_alignment(impl, result_component_type);
 		alignment = 16;
 	}
 	else
 	{
-		byteaddr_id = member_index;
 		// DXIL emits the alignment, but we cannot trust it, DXC is completely buggy here and emits
 		// obviously bogus alignment values.
 		// Use scalar alignment.
@@ -1632,7 +1625,17 @@ static bool emit_cbuffer_load_physical_pointer(Converter::Impl &impl, const llvm
 	spv::Id physical_type_id = 0;
 	get_physical_load_store_cast_info(impl, result_component_type, physical_type_id, value_cast_op);
 
-	spv::Id addr_vec = emit_u32x2_u32_add(impl, impl.get_id_for_value(instruction->getOperand(1)), byteaddr_id);
+	spv::Id index_id;
+
+	if (!scalar_load)
+	{
+		index_id = impl.get_id_for_value(instruction->getOperand(2));
+	}
+	else
+	{
+		unsigned addr_shift_log2 = raw_buffer_data_type_to_addr_shift_log2(impl, instruction->getType());
+		index_id = build_index_divider(impl, instruction->getOperand(2), addr_shift_log2, 1);
+	}
 
 	auto *result_type = instruction->getType();
 	unsigned physical_vecsize;
@@ -1659,15 +1662,20 @@ static bool emit_cbuffer_load_physical_pointer(Converter::Impl &impl, const llvm
 
 	Converter::Impl::PhysicalPointerMeta ptr_meta = {};
 	ptr_meta.nonwritable = true;
+	ptr_meta.stride = alignment;
+	ptr_meta.size = 64 * 1024;
 	spv::Id ptr_type_id = impl.get_physical_pointer_block_type(result_type_id, ptr_meta);
 
 	auto *ptr_bitcast_op = impl.allocate(spv::OpBitcast, ptr_type_id);
-	ptr_bitcast_op->add_id(addr_vec);
+	ptr_bitcast_op->add_id(impl.get_id_for_value(instruction->getOperand(1)));
 	impl.add(ptr_bitcast_op);
 
-	auto *chain_op = impl.allocate(spv::OpAccessChain, builder.makePointer(spv::StorageClassPhysicalStorageBuffer, result_type_id));
+	// Out of bounds is undefined behavior for root descriptors.
+	// Allows a compiler to assume that the index is unsigned and multiplying by stride does not overflow 32-bit space.
+	auto *chain_op = impl.allocate(spv::OpInBoundsAccessChain, builder.makePointer(spv::StorageClassPhysicalStorageBuffer, result_type_id));
 	chain_op->add_id(ptr_bitcast_op->id);
 	chain_op->add_id(builder.makeUintConstant(0));
+	chain_op->add_id(index_id);
 	impl.add(chain_op);
 
 	auto *load_op = impl.allocate(spv::OpLoad, instruction, result_type_id);
diff --git a/reference/shaders/descriptor_qa/descriptor_qa.bindless.cbv-as-ssbo.descriptor-qa.comp b/reference/shaders/descriptor_qa/descriptor_qa.bindless.cbv-as-ssbo.descriptor-qa.comp
index 8e95bd3..d31cad4 100644
--- a/reference/shaders/descriptor_qa/descriptor_qa.bindless.cbv-as-ssbo.descriptor-qa.comp
+++ b/reference/shaders/descriptor_qa/descriptor_qa.bindless.cbv-as-ssbo.descriptor-qa.comp
@@ -5,7 +5,7 @@
 layout(local_size_x = 1, local_size_y = 1, local_size_z = 1) in;
 
 layout(buffer_reference) buffer AtomicCounter;
-layout(buffer_reference, std430) buffer AtomicCounter
+layout(buffer_reference, buffer_reference_align = 4, std430) buffer AtomicCounter
 {
     uint _m0;
 };
diff --git a/reference/shaders/descriptor_qa/descriptor_qa.bindless.descriptor-qa.comp b/reference/shaders/descriptor_qa/descriptor_qa.bindless.descriptor-qa.comp
index 7445ea7..fcd47f7 100644
--- a/reference/shaders/descriptor_qa/descriptor_qa.bindless.descriptor-qa.comp
+++ b/reference/shaders/descriptor_qa/descriptor_qa.bindless.descriptor-qa.comp
@@ -5,7 +5,7 @@
 layout(local_size_x = 1, local_size_y = 1, local_size_z = 1) in;
 
 layout(buffer_reference) buffer AtomicCounter;
-layout(buffer_reference, std430) buffer AtomicCounter
+layout(buffer_reference, buffer_reference_align = 4, std430) buffer AtomicCounter
 {
     uint _m0;
 };
diff --git a/reference/shaders/descriptor_qa/descriptor_qa.bindless.descriptor-qa.sm66.comp b/reference/shaders/descriptor_qa/descriptor_qa.bindless.descriptor-qa.sm66.comp
index 0a1d0ab..8157210 100644
--- a/reference/shaders/descriptor_qa/descriptor_qa.bindless.descriptor-qa.sm66.comp
+++ b/reference/shaders/descriptor_qa/descriptor_qa.bindless.descriptor-qa.sm66.comp
@@ -5,7 +5,7 @@
 layout(local_size_x = 1, local_size_y = 1, local_size_z = 1) in;
 
 layout(buffer_reference) buffer AtomicCounter;
-layout(buffer_reference, std430) buffer AtomicCounter
+layout(buffer_reference, buffer_reference_align = 4, std430) buffer AtomicCounter
 {
     uint _m0;
 };
diff --git a/reference/shaders/descriptor_qa/descriptor_qa.bindless.ssbo.descriptor-qa.comp b/reference/shaders/descriptor_qa/descriptor_qa.bindless.ssbo.descriptor-qa.comp
index 049c910..5695290 100644
--- a/reference/shaders/descriptor_qa/descriptor_qa.bindless.ssbo.descriptor-qa.comp
+++ b/reference/shaders/descriptor_qa/descriptor_qa.bindless.ssbo.descriptor-qa.comp
@@ -5,7 +5,7 @@
 layout(local_size_x = 1, local_size_y = 1, local_size_z = 1) in;
 
 layout(buffer_reference) buffer AtomicCounter;
-layout(buffer_reference, std430) buffer AtomicCounter
+layout(buffer_reference, buffer_reference_align = 4, std430) buffer AtomicCounter
 {
     uint _m0;
 };
diff --git a/reference/shaders/dxil-builtin/atomic-bin-op.root-descriptor.comp b/reference/shaders/dxil-builtin/atomic-bin-op.root-descriptor.comp
index 08825b2..a501c13 100644
--- a/reference/shaders/dxil-builtin/atomic-bin-op.root-descriptor.comp
+++ b/reference/shaders/dxil-builtin/atomic-bin-op.root-descriptor.comp
@@ -1,9 +1,10 @@
 #version 460
 #extension GL_EXT_buffer_reference : require
+#extension GL_EXT_buffer_reference_uvec2 : require
 layout(local_size_x = 1, local_size_y = 1, local_size_z = 1) in;
 
 layout(buffer_reference) buffer PhysicalPointerUintArray;
-layout(buffer_reference, std430) buffer PhysicalPointerUintArray
+layout(buffer_reference, buffer_reference_align = 4, std430) buffer PhysicalPointerUintArray
 {
     uint value[];
 };
diff --git a/reference/shaders/dxil-builtin/atomic-compare-exchange.root-descriptor.comp b/reference/shaders/dxil-builtin/atomic-compare-exchange.root-descriptor.comp
index addde1a..4429c57 100644
--- a/reference/shaders/dxil-builtin/atomic-compare-exchange.root-descriptor.comp
+++ b/reference/shaders/dxil-builtin/atomic-compare-exchange.root-descriptor.comp
@@ -1,9 +1,10 @@
 #version 460
 #extension GL_EXT_buffer_reference : require
+#extension GL_EXT_buffer_reference_uvec2 : require
 layout(local_size_x = 1, local_size_y = 1, local_size_z = 1) in;
 
 layout(buffer_reference) buffer PhysicalPointerUintArray;
-layout(buffer_reference, std430) buffer PhysicalPointerUintArray
+layout(buffer_reference, buffer_reference_align = 4, std430) buffer PhysicalPointerUintArray
 {
     uint value[];
 };
diff --git a/reference/shaders/dxil-builtin/barycentrics-2.frag b/reference/shaders/dxil-builtin/barycentrics-2.frag
index 302d676..d846596 100644
--- a/reference/shaders/dxil-builtin/barycentrics-2.frag
+++ b/reference/shaders/dxil-builtin/barycentrics-2.frag
@@ -22,7 +22,7 @@ void main()
 ; Schema: 0
 OpCapability Shader
 OpCapability SampleRateShading
-OpCapability FragmentBarycentricNV
+OpCapability FragmentBarycentricKHR
 OpExtension "SPV_NV_fragment_shader_barycentric"
 OpMemoryModel Logical GLSL450
 OpEntryPoint Fragment %3 "main" %10 %11 %12 %15 %17 %19
@@ -34,15 +34,15 @@ OpName %12 "ATTRIB_2"
 OpName %15 "SV_Barycentrics"
 OpName %17 "FOO"
 OpName %19 "SV_Target"
-OpDecorate %10 PerVertexNV
+OpDecorate %10 PerVertexKHR
 OpDecorate %10 Location 0
-OpDecorate %11 PerVertexNV
+OpDecorate %11 PerVertexKHR
 OpDecorate %11 Location 0
 OpDecorate %11 Component 1
-OpDecorate %12 PerVertexNV
+OpDecorate %12 PerVertexKHR
 OpDecorate %12 Location 0
 OpDecorate %12 Component 2
-OpDecorate %15 BuiltIn BaryCoordNV
+OpDecorate %15 BuiltIn BaryCoordKHR
 OpDecorate %15 Sample
 OpDecorate %17 Centroid
 OpDecorate %17 Location 1
diff --git a/reference/shaders/dxil-builtin/barycentrics.frag b/reference/shaders/dxil-builtin/barycentrics.frag
index f7eabcd..e32cfd7 100644
--- a/reference/shaders/dxil-builtin/barycentrics.frag
+++ b/reference/shaders/dxil-builtin/barycentrics.frag
@@ -23,7 +23,7 @@ void main()
 ; Bound: 162
 ; Schema: 0
 OpCapability Shader
-OpCapability FragmentBarycentricNV
+OpCapability FragmentBarycentricKHR
 OpExtension "SPV_NV_fragment_shader_barycentric"
 OpMemoryModel Logical GLSL450
 OpEntryPoint Fragment %3 "main" %11 %16 %19 %20 %22 %24
@@ -35,12 +35,12 @@ OpName %19 "SV_Barycentrics"
 OpName %20 "SV_Barycentrics_1"
 OpName %22 "INDEX"
 OpName %24 "SV_Target"
-OpDecorate %11 PerVertexNV
+OpDecorate %11 PerVertexKHR
 OpDecorate %11 Location 0
-OpDecorate %16 PerVertexNV
+OpDecorate %16 PerVertexKHR
 OpDecorate %16 Location 1
-OpDecorate %19 BuiltIn BaryCoordNV
-OpDecorate %20 BuiltIn BaryCoordNoPerspNV
+OpDecorate %19 BuiltIn BaryCoordKHR
+OpDecorate %20 BuiltIn BaryCoordNoPerspKHR
 OpDecorate %20 Centroid
 OpDecorate %22 Flat
 OpDecorate %22 Location 3
diff --git a/reference/shaders/dxil-builtin/clip.demote-to-helper.frag b/reference/shaders/dxil-builtin/clip.demote-to-helper.frag
index a52b58b..b126c34 100644
--- a/reference/shaders/dxil-builtin/clip.demote-to-helper.frag
+++ b/reference/shaders/dxil-builtin/clip.demote-to-helper.frag
@@ -26,7 +26,7 @@ void main()
 ; Bound: 36
 ; Schema: 0
 OpCapability Shader
-OpCapability DemoteToHelperInvocationEXT
+OpCapability DemoteToHelperInvocation
 OpExtension "SPV_EXT_demote_to_helper_invocation"
 OpMemoryModel Logical GLSL450
 OpEntryPoint Fragment %3 "main" %8
@@ -72,7 +72,7 @@ OpFunctionEnd
 OpSelectionMerge %31 None
 OpBranchConditional %27 %30 %31
 %30 = OpLabel
-OpDemoteToHelperInvocationEXT
+OpDemoteToHelperInvocation
 OpBranch %31
 %31 = OpLabel
 OpReturn
diff --git a/reference/shaders/dxil-builtin/discard.demote-to-helper.frag b/reference/shaders/dxil-builtin/discard.demote-to-helper.frag
index 717a476..89b0195 100644
--- a/reference/shaders/dxil-builtin/discard.demote-to-helper.frag
+++ b/reference/shaders/dxil-builtin/discard.demote-to-helper.frag
@@ -27,7 +27,7 @@ void main()
 ; Bound: 29
 ; Schema: 0
 OpCapability Shader
-OpCapability DemoteToHelperInvocationEXT
+OpCapability DemoteToHelperInvocation
 OpExtension "SPV_EXT_demote_to_helper_invocation"
 OpMemoryModel Logical GLSL450
 OpEntryPoint Fragment %3 "main" %8
@@ -58,7 +58,7 @@ OpBranch %22
 OpSelectionMerge %27 None
 OpBranchConditional %15 %26 %23
 %26 = OpLabel
-OpDemoteToHelperInvocationEXT
+OpDemoteToHelperInvocation
 OpBranch %27
 %23 = OpLabel
 %17 = OpAccessChain %9 %8 %18
@@ -67,7 +67,7 @@ OpBranch %27
 OpSelectionMerge %25 None
 OpBranchConditional %20 %24 %25
 %24 = OpLabel
-OpDemoteToHelperInvocationEXT
+OpDemoteToHelperInvocation
 OpBranch %25
 %25 = OpLabel
 OpBranch %27
diff --git a/reference/shaders/dxil-builtin/get-dimensions-xyz-only.frag b/reference/shaders/dxil-builtin/get-dimensions-xyz-only.frag
index aa90694..5539174 100644
--- a/reference/shaders/dxil-builtin/get-dimensions-xyz-only.frag
+++ b/reference/shaders/dxil-builtin/get-dimensions-xyz-only.frag
@@ -46,7 +46,8 @@ void main()
     uint _144 = uint(imageSize(_61)) / 4u;
     uint _147 = uint(textureSize(_40)) * 4u;
     uint _150 = uint(imageSize(_62)) * 4u;
-    SV_Target = (((((((((((((((((((((((((((((((((((((uint(textureSize(_8, int(LEVEL))) + 32u) + _90.y) + _90.x) + _93.y) + _93.x) + _97.y) + _97.x) + _97.z) + _101.y) + _101.x) + _104.y) + _104.x) + _104.z) + _108.y) + _108.x) + _108.z) + _112.y) + _112.x) + _115.y) + _115.x) + _115.z) + (_119 * _119)) + (_121 * _121)) + _123.y) + _123.x) + _126.y) + _126.x) + _129.y) + _129.x) + _129.z) + _133.y) + _133.x) + _133.z) + (_137 * _137)) + (_140 * _140)) + (_144 * _144)) + (_147 * _147)) + (_150 * _150);
+    uint _180 = (((((((((((((((((((((((((((uint(textureSize(_8, int(LEVEL))) + 32u) + _90.y) + _90.x) + _93.y) + _93.x) + _97.y) + _97.x) + _97.z) + _101.y) + _101.x) + _104.y) + _104.x) + _104.z) + _108.y) + _108.x) + _108.z) + _112.y) + _112.x) + _115.y) + _115.x) + _115.z) + (_119 * _119)) + (_121 * _121)) + _123.y) + _123.x) + _126.y) + _126.x) + _129.y;
+    SV_Target = (((((((((_180 + _129.x) + _129.z) + _133.y) + _133.x) + _133.z) + (_137 * _137)) + (_140 * _140)) + (_144 * _144)) + (_147 * _147)) + (_150 * _150);
 }
 
 
diff --git a/reference/shaders/dxil-builtin/get-dimensions.frag b/reference/shaders/dxil-builtin/get-dimensions.frag
index c4c01fa..33b5e6d 100644
--- a/reference/shaders/dxil-builtin/get-dimensions.frag
+++ b/reference/shaders/dxil-builtin/get-dimensions.frag
@@ -54,7 +54,8 @@ void main()
     uint _179 = uint(imageSize(_61)) / 4u;
     uint _182 = uint(textureSize(_40)) * 4u;
     uint _185 = uint(imageSize(_62)) * 4u;
-    SV_Target = ((((((((((((((((((((((((((((((((((((((((((((((_91.w + 32u) + _91.x) + _99.y) + _99.x) + _99.w) + _106.y) + _106.x) + _106.w) + _114.y) + _114.x) + _114.z) + _114.w) + _121.y) + _121.x) + _121.w) + _128.y) + _128.x) + _128.z) + _128.w) + _135.y) + _135.x) + _135.z) + _135.w) + _142.y) + _142.x) + _142.w) + _149.y) + _149.x) + _149.z) + _149.w) + (_154 * _154)) + (_156 * _156)) + _158.y) + _158.x) + _161.y) + _161.x) + _164.y) + _164.x) + _164.z) + _168.y) + _168.x) + _168.z) + (_172 * _172)) + (_175 * _175)) + (_179 * _179)) + (_182 * _182)) + (_185 * _185);
+    uint _216 = ((((((((((((((((((((((((((((_91.w + 32u) + _91.x) + _99.y) + _99.x) + _99.w) + _106.y) + _106.x) + _106.w) + _114.y) + _114.x) + _114.z) + _114.w) + _121.y) + _121.x) + _121.w) + _128.y) + _128.x) + _128.z) + _128.w) + _135.y) + _135.x) + _135.z) + _135.w) + _142.y) + _142.x) + _142.w) + _149.y) + _149.x) + _149.z;
+    SV_Target = (((((((((((((((((_216 + _149.w) + (_154 * _154)) + (_156 * _156)) + _158.y) + _158.x) + _161.y) + _161.x) + _164.y) + _164.x) + _164.z) + _168.y) + _168.x) + _168.z) + (_172 * _172)) + (_175 * _175)) + (_179 * _179)) + (_182 * _182)) + (_185 * _185);
 }
 
 
diff --git a/reference/shaders/dxil-builtin/is-helper-lane.demote-to-helper.sm66.frag b/reference/shaders/dxil-builtin/is-helper-lane.demote-to-helper.sm66.frag
index 740be89..f4a2a0b 100644
--- a/reference/shaders/dxil-builtin/is-helper-lane.demote-to-helper.sm66.frag
+++ b/reference/shaders/dxil-builtin/is-helper-lane.demote-to-helper.sm66.frag
@@ -53,7 +53,7 @@ void main()
 ; Schema: 0
 OpCapability Shader
 OpCapability GroupNonUniformArithmetic
-OpCapability DemoteToHelperInvocationEXT
+OpCapability DemoteToHelperInvocation
 OpExtension "SPV_EXT_demote_to_helper_invocation"
 OpMemoryModel Logical GLSL450
 OpEntryPoint Fragment %3 "main" %8 %11
@@ -95,14 +95,14 @@ OpBranch %44
 OpSelectionMerge %49 None
 OpBranchConditional %21 %48 %45
 %48 = OpLabel
-OpDemoteToHelperInvocationEXT
+OpDemoteToHelperInvocation
 OpBranch %49
 %45 = OpLabel
 %23 = OpFOrdGreaterThan %20 %19 %24
 OpSelectionMerge %47 None
 OpBranchConditional %23 %46 %47
 %46 = OpLabel
-OpDemoteToHelperInvocationEXT
+OpDemoteToHelperInvocation
 OpBranch %47
 %47 = OpLabel
 OpBranch %49
diff --git a/reference/shaders/dxil-builtin/object-ray-direction.rany b/reference/shaders/dxil-builtin/object-ray-direction.rany
index faf9a26..b09cbca 100644
--- a/reference/shaders/dxil-builtin/object-ray-direction.rany
+++ b/reference/shaders/dxil-builtin/object-ray-direction.rany
@@ -19,13 +19,11 @@ vec3 _22;
 
 void main()
 {
-    vec3 _21 = _22;
+    vec3 _21;
     _21.x = gl_ObjectRayDirectionEXT.x;
-    vec3 _26 = _21;
-    _26.y = gl_ObjectRayDirectionEXT.y;
-    vec3 _30 = _26;
-    _30.z = gl_ObjectRayDirectionEXT.z;
-    payload._m0 = _30;
+    _21.y = gl_ObjectRayDirectionEXT.y;
+    _21.z = gl_ObjectRayDirectionEXT.z;
+    payload._m0 = _21;
 }
 
 
diff --git a/reference/shaders/dxil-builtin/object-ray-origin.rany b/reference/shaders/dxil-builtin/object-ray-origin.rany
index b3298ce..cca449b 100644
--- a/reference/shaders/dxil-builtin/object-ray-origin.rany
+++ b/reference/shaders/dxil-builtin/object-ray-origin.rany
@@ -19,13 +19,11 @@ vec3 _22;
 
 void main()
 {
-    vec3 _21 = _22;
+    vec3 _21;
     _21.x = gl_ObjectRayOriginEXT.x;
-    vec3 _26 = _21;
-    _26.y = gl_ObjectRayOriginEXT.y;
-    vec3 _30 = _26;
-    _30.z = gl_ObjectRayOriginEXT.z;
-    payload._m0 = _30;
+    _21.y = gl_ObjectRayOriginEXT.y;
+    _21.z = gl_ObjectRayOriginEXT.z;
+    payload._m0 = _21;
 }
 
 
diff --git a/reference/shaders/dxil-builtin/object-to-world-3x4.rany b/reference/shaders/dxil-builtin/object-to-world-3x4.rany
index 26813db..d257807 100644
--- a/reference/shaders/dxil-builtin/object-to-world-3x4.rany
+++ b/reference/shaders/dxil-builtin/object-to-world-3x4.rany
@@ -19,13 +19,11 @@ vec3 _59;
 
 void main()
 {
-    vec3 _58 = _59;
+    vec3 _58;
     _58.x = fma(gl_ObjectToWorldEXT[2u].x, payload._m0.z, fma(gl_ObjectToWorldEXT[1u].x, payload._m0.y, gl_ObjectToWorldEXT[0u].x * payload._m0.x)) + gl_ObjectToWorldEXT[3u].x;
-    vec3 _64 = _58;
-    _64.y = fma(gl_ObjectToWorldEXT[2u].y, payload._m0.z, fma(gl_ObjectToWorldEXT[1u].y, payload._m0.y, gl_ObjectToWorldEXT[0u].y * payload._m0.x)) + gl_ObjectToWorldEXT[3u].y;
-    vec3 _69 = _64;
-    _69.z = fma(gl_ObjectToWorldEXT[2u].z, payload._m0.z, fma(gl_ObjectToWorldEXT[1u].z, payload._m0.y, gl_ObjectToWorldEXT[0u].z * payload._m0.x)) + gl_ObjectToWorldEXT[3u].z;
-    payload._m0 = _69;
+    _58.y = fma(gl_ObjectToWorldEXT[2u].y, payload._m0.z, fma(gl_ObjectToWorldEXT[1u].y, payload._m0.y, gl_ObjectToWorldEXT[0u].y * payload._m0.x)) + gl_ObjectToWorldEXT[3u].y;
+    _58.z = fma(gl_ObjectToWorldEXT[2u].z, payload._m0.z, fma(gl_ObjectToWorldEXT[1u].z, payload._m0.y, gl_ObjectToWorldEXT[0u].z * payload._m0.x)) + gl_ObjectToWorldEXT[3u].z;
+    payload._m0 = _58;
 }
 
 
diff --git a/reference/shaders/dxil-builtin/object-to-world-4x3.rany b/reference/shaders/dxil-builtin/object-to-world-4x3.rany
index c42f1b4..c4acdf1 100644
--- a/reference/shaders/dxil-builtin/object-to-world-4x3.rany
+++ b/reference/shaders/dxil-builtin/object-to-world-4x3.rany
@@ -19,13 +19,11 @@ vec3 _59;
 
 void main()
 {
-    vec3 _58 = _59;
+    vec3 _58;
     _58.x = fma(payload._m0.z, gl_ObjectToWorldEXT[2u].x, fma(payload._m0.y, gl_ObjectToWorldEXT[1u].x, payload._m0.x * gl_ObjectToWorldEXT[0u].x)) + gl_ObjectToWorldEXT[3u].x;
-    vec3 _64 = _58;
-    _64.y = fma(payload._m0.z, gl_ObjectToWorldEXT[2u].y, fma(payload._m0.y, gl_ObjectToWorldEXT[1u].y, payload._m0.x * gl_ObjectToWorldEXT[0u].y)) + gl_ObjectToWorldEXT[3u].y;
-    vec3 _69 = _64;
-    _69.z = fma(payload._m0.z, gl_ObjectToWorldEXT[2u].z, fma(payload._m0.y, gl_ObjectToWorldEXT[1u].z, payload._m0.x * gl_ObjectToWorldEXT[0u].z)) + gl_ObjectToWorldEXT[3u].z;
-    payload._m0 = _69;
+    _58.y = fma(payload._m0.z, gl_ObjectToWorldEXT[2u].y, fma(payload._m0.y, gl_ObjectToWorldEXT[1u].y, payload._m0.x * gl_ObjectToWorldEXT[0u].y)) + gl_ObjectToWorldEXT[3u].y;
+    _58.z = fma(payload._m0.z, gl_ObjectToWorldEXT[2u].z, fma(payload._m0.y, gl_ObjectToWorldEXT[1u].z, payload._m0.x * gl_ObjectToWorldEXT[0u].z)) + gl_ObjectToWorldEXT[3u].z;
+    payload._m0 = _58;
 }
 
 
diff --git a/reference/shaders/dxil-builtin/ray-t-current.rany b/reference/shaders/dxil-builtin/ray-t-current.rany
index f4c336a..18f9f5e 100644
--- a/reference/shaders/dxil-builtin/ray-t-current.rany
+++ b/reference/shaders/dxil-builtin/ray-t-current.rany
@@ -19,13 +19,11 @@ vec3 _18;
 
 void main()
 {
-    vec3 _17 = _18;
+    vec3 _17;
     _17.x = gl_RayTmaxEXT;
-    vec3 _19 = _17;
-    _19.y = gl_RayTmaxEXT;
-    vec3 _20 = _19;
-    _20.z = gl_RayTmaxEXT;
-    payload._m0 = _20;
+    _17.y = gl_RayTmaxEXT;
+    _17.z = gl_RayTmaxEXT;
+    payload._m0 = _17;
 }
 
 
diff --git a/reference/shaders/dxil-builtin/ray-t-min.rany b/reference/shaders/dxil-builtin/ray-t-min.rany
index c94cdcb..500bba3 100644
--- a/reference/shaders/dxil-builtin/ray-t-min.rany
+++ b/reference/shaders/dxil-builtin/ray-t-min.rany
@@ -19,13 +19,11 @@ vec3 _18;
 
 void main()
 {
-    vec3 _17 = _18;
+    vec3 _17;
     _17.x = gl_RayTminEXT;
-    vec3 _19 = _17;
-    _19.y = gl_RayTminEXT;
-    vec3 _20 = _19;
-    _20.z = gl_RayTminEXT;
-    payload._m0 = _20;
+    _17.y = gl_RayTminEXT;
+    _17.z = gl_RayTminEXT;
+    payload._m0 = _17;
 }
 
 
diff --git a/reference/shaders/dxil-builtin/render-target-sample-position.frag b/reference/shaders/dxil-builtin/render-target-sample-position.frag
index 1dac314..b4ab6a4 100644
--- a/reference/shaders/dxil-builtin/render-target-sample-position.frag
+++ b/reference/shaders/dxil-builtin/render-target-sample-position.frag
@@ -7,7 +7,8 @@ layout(location = 0) out vec2 SV_Target;
 
 void main()
 {
-    uint _74 = ((uint(gl_SampleID) < _13) && (_13 <= 16u)) ? ((_13 - 1u) + uint(gl_SampleID)) : 0u;
+    uint _66 = _13 - 1u;
+    uint _74 = ((uint(gl_SampleID) < _13) && (_13 <= 16u)) ? (_66 + uint(gl_SampleID)) : 0u;
     SV_Target.x = _63[_74].x;
     SV_Target.y = _63[_74].y;
 }
diff --git a/reference/shaders/dxil-builtin/report-hit.rint b/reference/shaders/dxil-builtin/report-hit.rint
index 1d229ce..3688505 100644
--- a/reference/shaders/dxil-builtin/report-hit.rint
+++ b/reference/shaders/dxil-builtin/report-hit.rint
@@ -11,8 +11,9 @@ hitAttributeEXT _6 hit;
 
 void main()
 {
+    float _14 = gl_RayTmaxEXT;
     _6 _10;
-    _10._m0 = gl_RayTmaxEXT;
+    _10._m0 = _14;
     _6 _11;
     _11._m0 = gl_RayTminEXT;
     hit = _10;
diff --git a/reference/shaders/dxil-builtin/sm64-packed-arithmetic.ssbo.i8dot.noglsl.comp b/reference/shaders/dxil-builtin/sm64-packed-arithmetic.ssbo.i8dot.noglsl.comp
index 6fd75c1..4f1ff22 100644
--- a/reference/shaders/dxil-builtin/sm64-packed-arithmetic.ssbo.i8dot.noglsl.comp
+++ b/reference/shaders/dxil-builtin/sm64-packed-arithmetic.ssbo.i8dot.noglsl.comp
@@ -7,8 +7,8 @@ OpCapability Shader
 OpCapability Float16
 OpCapability Int16
 OpCapability StorageBuffer16BitAccess
-OpCapability DotProductInput4x8BitPackedKHR
-OpCapability DotProductKHR
+OpCapability DotProductInput4x8BitPacked
+OpCapability DotProduct
 OpExtension "SPV_KHR_integer_dot_product"
 OpMemoryModel Logical GLSL450
 OpEntryPoint GLCompute %3 "main" %34
@@ -122,7 +122,7 @@ OpBranch %121
 %43 = OpLoad %5 %42
 %44 = OpAccessChain %39 %9 %37 %38
 %45 = OpLoad %5 %44
-%46 = OpUDotKHR %5 %45 %43 PackedVectorFormat4x8BitKHR
+%46 = OpUDot %5 %45 %43 PackedVectorFormat4x8Bit
 %47 = OpIAdd %5 %41 %46
 %48 = OpAccessChain %39 %27 %37 %38
 OpStore %48 %47
@@ -132,7 +132,7 @@ OpStore %48 %47
 %52 = OpLoad %5 %51
 %53 = OpAccessChain %39 %9 %37 %38
 %54 = OpLoad %5 %53
-%55 = OpSDotKHR %5 %54 %52 PackedVectorFormat4x8BitKHR
+%55 = OpSDot %5 %54 %52 PackedVectorFormat4x8Bit
 %56 = OpIAdd %5 %50 %55
 %57 = OpAccessChain %39 %27 %37 %38
 OpStore %57 %56
diff --git a/reference/shaders/dxil-builtin/texture-load-feedback.frag b/reference/shaders/dxil-builtin/texture-load-feedback.frag
index 2eeaff8..71c2dc8 100644
--- a/reference/shaders/dxil-builtin/texture-load-feedback.frag
+++ b/reference/shaders/dxil-builtin/texture-load-feedback.frag
@@ -72,14 +72,14 @@ void main()
     float _166 = float(sparseTexelsResidentARB(int(_159._m4)));
     uint _309;
     vec4 _310;
-    _309 = sparseTexelFetchARB(_23, ivec2(uvec2(TEXCOORD.x, TEXCOORD.y)), TEXCOORD.z, _310);
+    _309 = sparseTexelFetchARB(_23, ivec2(uvec2(TEXCOORD.x, TEXCOORD.y)), int(TEXCOORD.z), _310);
     SparseTexel _169 = SparseTexel(_309, _310);
     vec4 _172 = _169._m1;
     _83 _177 = _83(_172.x, _172.y, _172.z, _172.w, _169._m0);
     float _184 = float(sparseTexelsResidentARB(int(_177._m4)));
     uint _311;
     vec4 _312;
-    _311 = sparseTexelFetchARB(_26, ivec3(uvec3(TEXCOORD.x, TEXCOORD.y, TEXCOORD.z)), TEXCOORD.w, _312);
+    _311 = sparseTexelFetchARB(_26, ivec3(uvec3(TEXCOORD.x, TEXCOORD.y, TEXCOORD.z)), int(TEXCOORD.w), _312);
     SparseTexel _187 = SparseTexel(_311, _312);
     vec4 _190 = _187._m1;
     _83 _195 = _83(_190.x, _190.y, _190.z, _190.w, _187._m0);
diff --git a/reference/shaders/dxil-builtin/texture-load-offset.frag b/reference/shaders/dxil-builtin/texture-load-offset.frag
index 4efadce..8c680c8 100644
--- a/reference/shaders/dxil-builtin/texture-load-offset.frag
+++ b/reference/shaders/dxil-builtin/texture-load-offset.frag
@@ -19,8 +19,8 @@ void main()
     vec4 _70 = texelFetchOffset(_14, ivec2(uvec2(TEXCOORD.x, TEXCOORD.y)), int(TEXCOORD.z), ivec2(3, 4));
     vec4 _80 = texelFetchOffset(_17, ivec3(uvec3(TEXCOORD.x, TEXCOORD.y, TEXCOORD.z)), int(TEXCOORD.w), ivec2(-4, -3));
     vec4 _88 = texelFetchOffset(_20, ivec3(uvec3(TEXCOORD.x, TEXCOORD.y, TEXCOORD.z)), int(TEXCOORD.w), ivec3(-4, 2, 3));
-    vec4 _96 = texelFetchOffset(_23, ivec2(uvec2(TEXCOORD.x, TEXCOORD.y)), ivec2(2, 3), TEXCOORD.z);
-    vec4 _104 = texelFetchOffset(_26, ivec3(uvec3(TEXCOORD.x, TEXCOORD.y, TEXCOORD.z)), ivec2(4, 5), TEXCOORD.w);
+    vec4 _96 = texelFetchOffset(_23, ivec2(uvec2(TEXCOORD.x, TEXCOORD.y)), ivec2(2, 3), int(TEXCOORD.z));
+    vec4 _104 = texelFetchOffset(_26, ivec3(uvec3(TEXCOORD.x, TEXCOORD.y, TEXCOORD.z)), ivec2(4, 5), int(TEXCOORD.w));
     SV_Target.x = (((((_61.x + _57.x) + _70.x) + _80.x) + _88.x) + _96.x) + _104.x;
     SV_Target.y = (((((_61.y + _57.y) + _70.y) + _80.y) + _88.y) + _96.y) + _104.y;
 }
diff --git a/reference/shaders/dxil-builtin/texture-load-signed.frag b/reference/shaders/dxil-builtin/texture-load-signed.frag
index 96e158c..d54dfea 100644
--- a/reference/shaders/dxil-builtin/texture-load-signed.frag
+++ b/reference/shaders/dxil-builtin/texture-load-signed.frag
@@ -24,8 +24,8 @@ void main()
     uvec4 _89 = uvec4(texelFetch(_14, ivec2(uvec2(TEXCOORD.x, TEXCOORD.y)), int(TEXCOORD.z)));
     uvec4 _97 = uvec4(texelFetch(_17, ivec3(uvec3(TEXCOORD.x, TEXCOORD.y, TEXCOORD.z)), int(TEXCOORD.w)));
     uvec4 _104 = uvec4(texelFetch(_20, ivec3(uvec3(TEXCOORD.x, TEXCOORD.y, TEXCOORD.z)), int(TEXCOORD.w)));
-    uvec4 _111 = uvec4(texelFetch(_23, ivec2(uvec2(TEXCOORD.x, TEXCOORD.y)), TEXCOORD.z));
-    uvec4 _118 = uvec4(texelFetch(_26, ivec3(uvec3(TEXCOORD.x, TEXCOORD.y, TEXCOORD.z)), TEXCOORD.w));
+    uvec4 _111 = uvec4(texelFetch(_23, ivec2(uvec2(TEXCOORD.x, TEXCOORD.y)), int(TEXCOORD.z)));
+    uvec4 _118 = uvec4(texelFetch(_26, ivec3(uvec3(TEXCOORD.x, TEXCOORD.y, TEXCOORD.z)), int(TEXCOORD.w)));
     uvec4 _124 = uvec4(imageLoad(_29, int(TEXCOORD.x)));
     uvec4 _131 = uvec4(imageLoad(_32, ivec2(uvec2(TEXCOORD.x, TEXCOORD.y))));
     uvec4 _138 = uvec4(imageLoad(_35, ivec2(uvec2(TEXCOORD.x, TEXCOORD.y))));
diff --git a/reference/shaders/dxil-builtin/texture-load.frag b/reference/shaders/dxil-builtin/texture-load.frag
index b80a89f..0338a9d 100644
--- a/reference/shaders/dxil-builtin/texture-load.frag
+++ b/reference/shaders/dxil-builtin/texture-load.frag
@@ -24,8 +24,8 @@ void main()
     vec4 _85 = texelFetch(_14, ivec2(uvec2(TEXCOORD.x, TEXCOORD.y)), int(TEXCOORD.z));
     vec4 _91 = texelFetch(_17, ivec3(uvec3(TEXCOORD.x, TEXCOORD.y, TEXCOORD.z)), int(TEXCOORD.w));
     vec4 _98 = texelFetch(_20, ivec3(uvec3(TEXCOORD.x, TEXCOORD.y, TEXCOORD.z)), int(TEXCOORD.w));
-    vec4 _104 = texelFetch(_23, ivec2(uvec2(TEXCOORD.x, TEXCOORD.y)), TEXCOORD.z);
-    vec4 _110 = texelFetch(_26, ivec3(uvec3(TEXCOORD.x, TEXCOORD.y, TEXCOORD.z)), TEXCOORD.w);
+    vec4 _104 = texelFetch(_23, ivec2(uvec2(TEXCOORD.x, TEXCOORD.y)), int(TEXCOORD.z));
+    vec4 _110 = texelFetch(_26, ivec3(uvec3(TEXCOORD.x, TEXCOORD.y, TEXCOORD.z)), int(TEXCOORD.w));
     vec4 _116 = imageLoad(_29, int(TEXCOORD.x));
     vec4 _121 = imageLoad(_32, ivec2(uvec2(TEXCOORD.x, TEXCOORD.y)));
     vec4 _127 = imageLoad(_35, ivec2(uvec2(TEXCOORD.x, TEXCOORD.y)));
diff --git a/reference/shaders/dxil-builtin/wave-active-ballot-discard.demote-to-helper.frag b/reference/shaders/dxil-builtin/wave-active-ballot-discard.demote-to-helper.frag
index aa21585..5c22543 100644
--- a/reference/shaders/dxil-builtin/wave-active-ballot-discard.demote-to-helper.frag
+++ b/reference/shaders/dxil-builtin/wave-active-ballot-discard.demote-to-helper.frag
@@ -28,7 +28,7 @@ void main()
 ; Schema: 0
 OpCapability Shader
 OpCapability GroupNonUniformBallot
-OpCapability DemoteToHelperInvocationEXT
+OpCapability DemoteToHelperInvocation
 OpExtension "SPV_EXT_demote_to_helper_invocation"
 OpMemoryModel Logical GLSL450
 OpEntryPoint Fragment %3 "main" %7 %10
@@ -64,7 +64,7 @@ OpBranch %31
 OpSelectionMerge %33 None
 OpBranchConditional %13 %32 %33
 %32 = OpLabel
-OpDemoteToHelperInvocationEXT
+OpDemoteToHelperInvocation
 OpBranch %33
 %33 = OpLabel
 %15 = OpULessThan %12 %11 %16
diff --git a/reference/shaders/dxil-builtin/world-ray-direction.rany b/reference/shaders/dxil-builtin/world-ray-direction.rany
index 43c623b..660c9e4 100644
--- a/reference/shaders/dxil-builtin/world-ray-direction.rany
+++ b/reference/shaders/dxil-builtin/world-ray-direction.rany
@@ -19,13 +19,11 @@ vec3 _22;
 
 void main()
 {
-    vec3 _21 = _22;
+    vec3 _21;
     _21.x = gl_WorldRayDirectionEXT.x;
-    vec3 _26 = _21;
-    _26.y = gl_WorldRayDirectionEXT.y;
-    vec3 _30 = _26;
-    _30.z = gl_WorldRayDirectionEXT.z;
-    payload._m0 = _30;
+    _21.y = gl_WorldRayDirectionEXT.y;
+    _21.z = gl_WorldRayDirectionEXT.z;
+    payload._m0 = _21;
 }
 
 
diff --git a/reference/shaders/dxil-builtin/world-ray-origin.rany b/reference/shaders/dxil-builtin/world-ray-origin.rany
index 67e1e0f..e3109ce 100644
--- a/reference/shaders/dxil-builtin/world-ray-origin.rany
+++ b/reference/shaders/dxil-builtin/world-ray-origin.rany
@@ -19,13 +19,11 @@ vec3 _22;
 
 void main()
 {
-    vec3 _21 = _22;
+    vec3 _21;
     _21.x = gl_WorldRayOriginEXT.x;
-    vec3 _26 = _21;
-    _26.y = gl_WorldRayOriginEXT.y;
-    vec3 _30 = _26;
-    _30.z = gl_WorldRayOriginEXT.z;
-    payload._m0 = _30;
+    _21.y = gl_WorldRayOriginEXT.y;
+    _21.z = gl_WorldRayOriginEXT.z;
+    payload._m0 = _21;
 }
 
 
diff --git a/reference/shaders/dxil-builtin/world-to-object-3x4.rany b/reference/shaders/dxil-builtin/world-to-object-3x4.rany
index a748e97..14177b8 100644
--- a/reference/shaders/dxil-builtin/world-to-object-3x4.rany
+++ b/reference/shaders/dxil-builtin/world-to-object-3x4.rany
@@ -19,13 +19,11 @@ vec3 _59;
 
 void main()
 {
-    vec3 _58 = _59;
+    vec3 _58;
     _58.x = fma(gl_WorldToObjectEXT[2u].x, payload._m0.z, fma(gl_WorldToObjectEXT[1u].x, payload._m0.y, gl_WorldToObjectEXT[0u].x * payload._m0.x)) + gl_WorldToObjectEXT[3u].x;
-    vec3 _64 = _58;
-    _64.y = fma(gl_WorldToObjectEXT[2u].y, payload._m0.z, fma(gl_WorldToObjectEXT[1u].y, payload._m0.y, gl_WorldToObjectEXT[0u].y * payload._m0.x)) + gl_WorldToObjectEXT[3u].y;
-    vec3 _69 = _64;
-    _69.z = fma(gl_WorldToObjectEXT[2u].z, payload._m0.z, fma(gl_WorldToObjectEXT[1u].z, payload._m0.y, gl_WorldToObjectEXT[0u].z * payload._m0.x)) + gl_WorldToObjectEXT[3u].z;
-    payload._m0 = _69;
+    _58.y = fma(gl_WorldToObjectEXT[2u].y, payload._m0.z, fma(gl_WorldToObjectEXT[1u].y, payload._m0.y, gl_WorldToObjectEXT[0u].y * payload._m0.x)) + gl_WorldToObjectEXT[3u].y;
+    _58.z = fma(gl_WorldToObjectEXT[2u].z, payload._m0.z, fma(gl_WorldToObjectEXT[1u].z, payload._m0.y, gl_WorldToObjectEXT[0u].z * payload._m0.x)) + gl_WorldToObjectEXT[3u].z;
+    payload._m0 = _58;
 }
 
 
diff --git a/reference/shaders/dxil-builtin/world-to-object-4x3.rany b/reference/shaders/dxil-builtin/world-to-object-4x3.rany
index 160f113..7fe5ec7 100644
--- a/reference/shaders/dxil-builtin/world-to-object-4x3.rany
+++ b/reference/shaders/dxil-builtin/world-to-object-4x3.rany
@@ -19,13 +19,11 @@ vec3 _59;
 
 void main()
 {
-    vec3 _58 = _59;
+    vec3 _58;
     _58.x = fma(payload._m0.z, gl_WorldToObjectEXT[2u].x, fma(payload._m0.y, gl_WorldToObjectEXT[1u].x, payload._m0.x * gl_WorldToObjectEXT[0u].x)) + gl_WorldToObjectEXT[3u].x;
-    vec3 _64 = _58;
-    _64.y = fma(payload._m0.z, gl_WorldToObjectEXT[2u].y, fma(payload._m0.y, gl_WorldToObjectEXT[1u].y, payload._m0.x * gl_WorldToObjectEXT[0u].y)) + gl_WorldToObjectEXT[3u].y;
-    vec3 _69 = _64;
-    _69.z = fma(payload._m0.z, gl_WorldToObjectEXT[2u].z, fma(payload._m0.y, gl_WorldToObjectEXT[1u].z, payload._m0.x * gl_WorldToObjectEXT[0u].z)) + gl_WorldToObjectEXT[3u].z;
-    payload._m0 = _69;
+    _58.y = fma(payload._m0.z, gl_WorldToObjectEXT[2u].y, fma(payload._m0.y, gl_WorldToObjectEXT[1u].y, payload._m0.x * gl_WorldToObjectEXT[0u].y)) + gl_WorldToObjectEXT[3u].y;
+    _58.z = fma(payload._m0.z, gl_WorldToObjectEXT[2u].z, fma(payload._m0.y, gl_WorldToObjectEXT[1u].z, payload._m0.x * gl_WorldToObjectEXT[0u].z)) + gl_WorldToObjectEXT[3u].z;
+    payload._m0 = _58;
 }
 
 
diff --git a/reference/shaders/fp16/saturate.sm60.frag b/reference/shaders/fp16/saturate.sm60.frag
index a4bfecc..5ed9216 100644
--- a/reference/shaders/fp16/saturate.sm60.frag
+++ b/reference/shaders/fp16/saturate.sm60.frag
@@ -5,14 +5,22 @@ layout(location = 0) out mediump vec4 SV_Target;
 
 void main()
 {
-    float _39 = isnan(0.0) ? V.x : (isnan(V.x) ? 0.0 : max(V.x, 0.0));
-    float _50 = isnan(0.0) ? V.y : (isnan(V.y) ? 0.0 : max(V.y, 0.0));
-    float _61 = isnan(0.0) ? V.z : (isnan(V.z) ? 0.0 : max(V.z, 0.0));
-    float _72 = isnan(0.0) ? V.w : (isnan(V.w) ? 0.0 : max(V.w, 0.0));
-    SV_Target.x = isnan(1.0) ? _39 : (isnan(_39) ? 1.0 : min(_39, 1.0));
-    SV_Target.y = isnan(1.0) ? _50 : (isnan(_50) ? 1.0 : min(_50, 1.0));
-    SV_Target.z = isnan(1.0) ? _61 : (isnan(_61) ? 1.0 : min(_61, 1.0));
-    SV_Target.w = isnan(1.0) ? _72 : (isnan(_72) ? 1.0 : min(_72, 1.0));
+    mediump float _15 = V.x;
+    float hp_copy_15 = _15;
+    mediump float _18 = V.y;
+    float hp_copy_18 = _18;
+    mediump float _21 = V.z;
+    float hp_copy_21 = _21;
+    mediump float _24 = V.w;
+    float hp_copy_24 = _24;
+    float _40 = isnan(0.0) ? hp_copy_15 : (isnan(hp_copy_15) ? 0.0 : max(hp_copy_15, 0.0));
+    float _52 = isnan(0.0) ? hp_copy_18 : (isnan(hp_copy_18) ? 0.0 : max(hp_copy_18, 0.0));
+    float _64 = isnan(0.0) ? hp_copy_21 : (isnan(hp_copy_21) ? 0.0 : max(hp_copy_21, 0.0));
+    float _76 = isnan(0.0) ? hp_copy_24 : (isnan(hp_copy_24) ? 0.0 : max(hp_copy_24, 0.0));
+    SV_Target.x = isnan(1.0) ? _40 : (isnan(_40) ? 1.0 : min(_40, 1.0));
+    SV_Target.y = isnan(1.0) ? _52 : (isnan(_52) ? 1.0 : min(_52, 1.0));
+    SV_Target.z = isnan(1.0) ? _64 : (isnan(_64) ? 1.0 : min(_64, 1.0));
+    SV_Target.w = isnan(1.0) ? _76 : (isnan(_76) ? 1.0 : min(_76, 1.0));
 }
 
 
diff --git a/reference/shaders/llvm-builtin/fcmp_ne.frag b/reference/shaders/llvm-builtin/fcmp_ne.frag
index f6fe33d..f5adc87 100644
--- a/reference/shaders/llvm-builtin/fcmp_ne.frag
+++ b/reference/shaders/llvm-builtin/fcmp_ne.frag
@@ -5,7 +5,7 @@ layout(location = 0) out float SV_Target;
 
 void main()
 {
-    SV_Target = (!(A.x == A.y)) ? A.z : A.w;
+    SV_Target = (A.x != A.y) ? A.z : A.w;
 }
 
 
diff --git a/reference/shaders/resources/cbv-legacy-fp16-fp64.root-descriptor.frag b/reference/shaders/resources/cbv-legacy-fp16-fp64.root-descriptor.frag
index 996b3f0..f2ccf13 100644
--- a/reference/shaders/resources/cbv-legacy-fp16-fp64.root-descriptor.frag
+++ b/reference/shaders/resources/cbv-legacy-fp16-fp64.root-descriptor.frag
@@ -9,12 +9,7 @@
 #extension GL_EXT_shader_16bit_storage : require
 #extension GL_ARB_gpu_shader_int64 : require
 #extension GL_EXT_buffer_reference : require
-
-struct AddCarry
-{
-    uint _m0;
-    uint _m1;
-};
+#extension GL_EXT_buffer_reference_uvec2 : require
 
 struct CBVComposite16x8
 {
@@ -28,16 +23,16 @@ struct CBVComposite16x8
     float16_t _m7;
 };
 
-layout(buffer_reference) buffer PhysicalPointerFloat4NonWrite;
-layout(buffer_reference) buffer PhysicalPointerUint642NonWrite;
-layout(buffer_reference, std430) readonly buffer PhysicalPointerFloat4NonWrite
+layout(buffer_reference) buffer PhysicalPointerFloat4NonWriteCBVArray;
+layout(buffer_reference) buffer PhysicalPointerUint642NonWriteCBVArray;
+layout(buffer_reference, buffer_reference_align = 16, std430) readonly buffer PhysicalPointerFloat4NonWriteCBVArray
 {
-    vec4 value;
+    vec4 value[4096];
 };
 
-layout(buffer_reference, std430) readonly buffer PhysicalPointerUint642NonWrite
+layout(buffer_reference, buffer_reference_align = 16, std430) readonly buffer PhysicalPointerUint642NonWriteCBVArray
 {
-    u64vec2 value;
+    u64vec2 value[4096];
 };
 
 layout(push_constant, std430) uniform RootConstants
@@ -52,27 +47,19 @@ layout(location = 0) out vec4 SV_Target;
 
 void main()
 {
-    AddCarry _23;
-    _23._m0 = uaddCarry(registers._m0.x, 0u * 16u, _23._m1);
-    PhysicalPointerFloat4NonWrite _30 = PhysicalPointerFloat4NonWrite(uvec2(_23._m0, registers._m0.y + _23._m1));
-    AddCarry _43;
-    _43._m0 = uaddCarry(registers._m0.x, 1u * 16u, _43._m1);
-    PhysicalPointerFloat4NonWrite _48 = PhysicalPointerFloat4NonWrite(uvec2(_43._m0, registers._m0.y + _43._m1));
-    f16vec2 _56 = unpackFloat2x16(floatBitsToUint(_48.value.x));
-    f16vec2 _59 = unpackFloat2x16(floatBitsToUint(_48.value.y));
-    f16vec2 _62 = unpackFloat2x16(floatBitsToUint(_48.value.z));
-    f16vec2 _65 = unpackFloat2x16(floatBitsToUint(_48.value.w));
-    CBVComposite16x8 _69 = CBVComposite16x8(_56.x, _56.y, _59.x, _59.y, _62.x, _62.y, _65.x, _65.y);
-    AddCarry _99;
-    _99._m0 = uaddCarry(registers._m0.x, 2u * 16u, _99._m1);
-    PhysicalPointerUint642NonWrite _107 = PhysicalPointerUint642NonWrite(uvec2(_99._m0, registers._m0.y + _99._m1));
-    AddCarry _117;
-    _117._m0 = uaddCarry(registers._m0.x, 3u * 16u, _117._m1);
-    PhysicalPointerUint642NonWrite _122 = PhysicalPointerUint642NonWrite(uvec2(_117._m0, registers._m0.y + _117._m1));
-    SV_Target.x = ((float(_69._m0) + _30.value.x) + float(_69._m4)) + float(int64_t(_107.value.x));
-    SV_Target.y = ((float(_69._m1) + _30.value.y) + float(_69._m5)) + float(int64_t(_107.value.y));
-    SV_Target.z = ((float(_69._m2) + _30.value.z) + float(_69._m6)) + float(int64_t(_122.value.x));
-    SV_Target.w = ((float(_69._m3) + _30.value.w) + float(_69._m7)) + float(int64_t(_122.value.y));
+    PhysicalPointerFloat4NonWriteCBVArray _22 = PhysicalPointerFloat4NonWriteCBVArray(registers._m0);
+    PhysicalPointerFloat4NonWriteCBVArray _32 = PhysicalPointerFloat4NonWriteCBVArray(registers._m0);
+    f16vec2 _40 = unpackFloat2x16(floatBitsToUint(_32.value[1u].x));
+    f16vec2 _43 = unpackFloat2x16(floatBitsToUint(_32.value[1u].y));
+    f16vec2 _46 = unpackFloat2x16(floatBitsToUint(_32.value[1u].z));
+    f16vec2 _49 = unpackFloat2x16(floatBitsToUint(_32.value[1u].w));
+    CBVComposite16x8 _53 = CBVComposite16x8(_40.x, _40.y, _43.x, _43.y, _46.x, _46.y, _49.x, _49.y);
+    PhysicalPointerUint642NonWriteCBVArray _84 = PhysicalPointerUint642NonWriteCBVArray(registers._m0);
+    PhysicalPointerUint642NonWriteCBVArray _91 = PhysicalPointerUint642NonWriteCBVArray(registers._m0);
+    SV_Target.x = ((float(_53._m0) + _22.value[0u].x) + float(_53._m4)) + float(int64_t(_84.value[2u].x));
+    SV_Target.y = ((float(_53._m1) + _22.value[0u].y) + float(_53._m5)) + float(int64_t(_84.value[2u].y));
+    SV_Target.z = ((float(_53._m2) + _22.value[0u].z) + float(_53._m6)) + float(int64_t(_91.value[3u].x));
+    SV_Target.w = ((float(_53._m3) + _22.value[0u].w) + float(_53._m7)) + float(int64_t(_91.value[3u].y));
 }
 
 
@@ -81,7 +68,7 @@ void main()
 ; SPIR-V
 ; Version: 1.3
 ; Generator: Unknown(30017); 21022
-; Bound: 142
+; Bound: 111
 ; Schema: 0
 OpCapability Shader
 OpCapability Float16
@@ -95,24 +82,25 @@ OpName %3 "main"
 OpName %7 "RootConstants"
 OpName %9 "registers"
 OpName %13 "SV_Target"
-OpName %22 "AddCarry"
-OpName %28 "PhysicalPointerFloat4NonWrite"
-OpMemberName %28 0 "value"
-OpName %68 "CBVComposite16x8"
-OpName %105 "PhysicalPointerUint642NonWrite"
-OpMemberName %105 0 "value"
+OpName %20 "PhysicalPointerFloat4NonWriteCBVArray"
+OpMemberName %20 0 "value"
+OpName %52 "CBVComposite16x8"
+OpName %82 "PhysicalPointerUint642NonWriteCBVArray"
+OpMemberName %82 0 "value"
 OpDecorate %7 Block
 OpMemberDecorate %7 0 Offset 0
 OpMemberDecorate %7 1 Offset 8
 OpMemberDecorate %7 2 Offset 16
 OpMemberDecorate %7 3 Offset 24
 OpDecorate %13 Location 0
-OpMemberDecorate %28 0 Offset 0
-OpDecorate %28 Block
-OpMemberDecorate %28 0 NonWritable
-OpMemberDecorate %105 0 Offset 0
-OpDecorate %105 Block
-OpMemberDecorate %105 0 NonWritable
+OpDecorate %19 ArrayStride 16
+OpMemberDecorate %20 0 Offset 0
+OpDecorate %20 Block
+OpMemberDecorate %20 0 NonWritable
+OpDecorate %81 ArrayStride 16
+OpMemberDecorate %82 0 Offset 0
+OpDecorate %82 Block
+OpMemberDecorate %82 0 NonWritable
 %1 = OpTypeVoid
 %2 = OpTypeFunction %1
 %5 = OpTypeInt 32 0
@@ -126,138 +114,107 @@ OpMemberDecorate %105 0 NonWritable
 %13 = OpVariable %12 Output
 %14 = OpTypePointer PushConstant %6
 %16 = OpConstant %5 0
-%19 = OpConstant %5 16
-%22 = OpTypeStruct %5 %5
-%28 = OpTypeStruct %11
-%29 = OpTypePointer PhysicalStorageBuffer %28
-%31 = OpTypePointer PhysicalStorageBuffer %11
-%38 = OpConstant %5 1
-%40 = OpTypeFloat 16
-%51 = OpTypeVector %40 2
-%68 = OpTypeStruct %40 %40 %40 %40 %40 %40 %40 %40
-%94 = OpConstant %5 2
-%96 = OpTypeInt 64 0
-%104 = OpTypeVector %96 2
-%105 = OpTypeStruct %104
-%106 = OpTypePointer PhysicalStorageBuffer %105
-%108 = OpTypePointer PhysicalStorageBuffer %104
-%113 = OpConstant %5 3
-%135 = OpTypePointer Output %10
+%18 = OpConstant %5 4096
+%19 = OpTypeArray %11 %18
+%20 = OpTypeStruct %19
+%21 = OpTypePointer PhysicalStorageBuffer %20
+%23 = OpTypePointer PhysicalStorageBuffer %11
+%30 = OpTypeFloat 16
+%31 = OpConstant %5 1
+%35 = OpTypeVector %30 2
+%52 = OpTypeStruct %30 %30 %30 %30 %30 %30 %30 %30
+%78 = OpTypeInt 64 0
+%79 = OpConstant %5 2
+%80 = OpTypeVector %78 2
+%81 = OpTypeArray %80 %18
+%82 = OpTypeStruct %81
+%83 = OpTypePointer PhysicalStorageBuffer %82
+%85 = OpTypePointer PhysicalStorageBuffer %80
+%90 = OpConstant %5 3
+%104 = OpTypePointer Output %10
 %3 = OpFunction %1 None %2
 %4 = OpLabel
-OpBranch %140
-%140 = OpLabel
+OpBranch %109
+%109 = OpLabel
 %15 = OpAccessChain %14 %9 %16
 %17 = OpLoad %6 %15
-%18 = OpIMul %5 %16 %19
-%20 = OpCompositeExtract %5 %17 0
-%21 = OpCompositeExtract %5 %17 1
-%23 = OpIAddCarry %22 %20 %18
-%24 = OpCompositeExtract %5 %23 0
-%25 = OpCompositeExtract %5 %23 1
-%26 = OpIAdd %5 %21 %25
-%27 = OpCompositeConstruct %6 %24 %26
-%30 = OpBitcast %29 %27
-%32 = OpAccessChain %31 %30 %16
-%33 = OpLoad %11 %32 Aligned 16
-%34 = OpCompositeExtract %10 %33 0
-%35 = OpCompositeExtract %10 %33 1
-%36 = OpCompositeExtract %10 %33 2
-%37 = OpCompositeExtract %10 %33 3
-%39 = OpIMul %5 %38 %19
-%41 = OpCompositeExtract %5 %17 0
-%42 = OpCompositeExtract %5 %17 1
-%43 = OpIAddCarry %22 %41 %39
-%44 = OpCompositeExtract %5 %43 0
-%45 = OpCompositeExtract %5 %43 1
-%46 = OpIAdd %5 %42 %45
-%47 = OpCompositeConstruct %6 %44 %46
-%48 = OpBitcast %29 %47
-%49 = OpAccessChain %31 %48 %16
-%50 = OpLoad %11 %49 Aligned 16
-%52 = OpCompositeExtract %10 %50 0
-%53 = OpCompositeExtract %10 %50 1
-%54 = OpCompositeExtract %10 %50 2
-%55 = OpCompositeExtract %10 %50 3
-%56 = OpBitcast %51 %52
-%57 = OpCompositeExtract %40 %56 0
-%58 = OpCompositeExtract %40 %56 1
-%59 = OpBitcast %51 %53
-%60 = OpCompositeExtract %40 %59 0
-%61 = OpCompositeExtract %40 %59 1
-%62 = OpBitcast %51 %54
-%63 = OpCompositeExtract %40 %62 0
-%64 = OpCompositeExtract %40 %62 1
-%65 = OpBitcast %51 %55
-%66 = OpCompositeExtract %40 %65 0
-%67 = OpCompositeExtract %40 %65 1
-%69 = OpCompositeConstruct %68 %57 %58 %60 %61 %63 %64 %66 %67
-%70 = OpCompositeExtract %40 %69 0
-%71 = OpCompositeExtract %40 %69 1
-%72 = OpCompositeExtract %40 %69 2
-%73 = OpCompositeExtract %40 %69 3
-%74 = OpFConvert %10 %70
-%75 = OpFConvert %10 %71
-%76 = OpFConvert %10 %72
-%77 = OpFConvert %10 %73
-%78 = OpFAdd %10 %74 %34
-%79 = OpFAdd %10 %75 %35
-%80 = OpFAdd %10 %76 %36
-%81 = OpFAdd %10 %77 %37
-%82 = OpCompositeExtract %40 %69 4
-%83 = OpCompositeExtract %40 %69 5
-%84 = OpCompositeExtract %40 %69 6
-%85 = OpCompositeExtract %40 %69 7
-%86 = OpFConvert %10 %82
-%87 = OpFConvert %10 %83
-%88 = OpFConvert %10 %84
-%89 = OpFConvert %10 %85
-%90 = OpFAdd %10 %78 %86
-%91 = OpFAdd %10 %79 %87
-%92 = OpFAdd %10 %80 %88
-%93 = OpFAdd %10 %81 %89
-%95 = OpIMul %5 %94 %19
-%97 = OpCompositeExtract %5 %17 0
-%98 = OpCompositeExtract %5 %17 1
-%99 = OpIAddCarry %22 %97 %95
-%100 = OpCompositeExtract %5 %99 0
-%101 = OpCompositeExtract %5 %99 1
-%102 = OpIAdd %5 %98 %101
-%103 = OpCompositeConstruct %6 %100 %102
-%107 = OpBitcast %106 %103
-%109 = OpAccessChain %108 %107 %16
-%110 = OpLoad %104 %109 Aligned 16
-%111 = OpCompositeExtract %96 %110 0
-%112 = OpCompositeExtract %96 %110 1
-%114 = OpIMul %5 %113 %19
-%115 = OpCompositeExtract %5 %17 0
-%116 = OpCompositeExtract %5 %17 1
-%117 = OpIAddCarry %22 %115 %114
-%118 = OpCompositeExtract %5 %117 0
-%119 = OpCompositeExtract %5 %117 1
-%120 = OpIAdd %5 %116 %119
-%121 = OpCompositeConstruct %6 %118 %120
-%122 = OpBitcast %106 %121
-%123 = OpAccessChain %108 %122 %16
-%124 = OpLoad %104 %123 Aligned 16
-%125 = OpCompositeExtract %96 %124 0
-%126 = OpCompositeExtract %96 %124 1
-%127 = OpConvertSToF %10 %111
-%128 = OpConvertSToF %10 %112
-%129 = OpConvertSToF %10 %125
-%130 = OpConvertSToF %10 %126
-%131 = OpFAdd %10 %90 %127
-%132 = OpFAdd %10 %91 %128
-%133 = OpFAdd %10 %92 %129
-%134 = OpFAdd %10 %93 %130
-%136 = OpAccessChain %135 %13 %16
-OpStore %136 %131
-%137 = OpAccessChain %135 %13 %38
-OpStore %137 %132
-%138 = OpAccessChain %135 %13 %94
-OpStore %138 %133
-%139 = OpAccessChain %135 %13 %113
-OpStore %139 %134
+%22 = OpBitcast %21 %17
+%24 = OpInBoundsAccessChain %23 %22 %16 %16
+%25 = OpLoad %11 %24 Aligned 16
+%26 = OpCompositeExtract %10 %25 0
+%27 = OpCompositeExtract %10 %25 1
+%28 = OpCompositeExtract %10 %25 2
+%29 = OpCompositeExtract %10 %25 3
+%32 = OpBitcast %21 %17
+%33 = OpInBoundsAccessChain %23 %32 %16 %31
+%34 = OpLoad %11 %33 Aligned 16
+%36 = OpCompositeExtract %10 %34 0
+%37 = OpCompositeExtract %10 %34 1
+%38 = OpCompositeExtract %10 %34 2
+%39 = OpCompositeExtract %10 %34 3
+%40 = OpBitcast %35 %36
+%41 = OpCompositeExtract %30 %40 0
+%42 = OpCompositeExtract %30 %40 1
+%43 = OpBitcast %35 %37
+%44 = OpCompositeExtract %30 %43 0
+%45 = OpCompositeExtract %30 %43 1
+%46 = OpBitcast %35 %38
+%47 = OpCompositeExtract %30 %46 0
+%48 = OpCompositeExtract %30 %46 1
+%49 = OpBitcast %35 %39
+%50 = OpCompositeExtract %30 %49 0
+%51 = OpCompositeExtract %30 %49 1
+%53 = OpCompositeConstruct %52 %41 %42 %44 %45 %47 %48 %50 %51
+%54 = OpCompositeExtract %30 %53 0
+%55 = OpCompositeExtract %30 %53 1
+%56 = OpCompositeExtract %30 %53 2
+%57 = OpCompositeExtract %30 %53 3
+%58 = OpFConvert %10 %54
+%59 = OpFConvert %10 %55
+%60 = OpFConvert %10 %56
+%61 = OpFConvert %10 %57
+%62 = OpFAdd %10 %58 %26
+%63 = OpFAdd %10 %59 %27
+%64 = OpFAdd %10 %60 %28
+%65 = OpFAdd %10 %61 %29
+%66 = OpCompositeExtract %30 %53 4
+%67 = OpCompositeExtract %30 %53 5
+%68 = OpCompositeExtract %30 %53 6
+%69 = OpCompositeExtract %30 %53 7
+%70 = OpFConvert %10 %66
+%71 = OpFConvert %10 %67
+%72 = OpFConvert %10 %68
+%73 = OpFConvert %10 %69
+%74 = OpFAdd %10 %62 %70
+%75 = OpFAdd %10 %63 %71
+%76 = OpFAdd %10 %64 %72
+%77 = OpFAdd %10 %65 %73
+%84 = OpBitcast %83 %17
+%86 = OpInBoundsAccessChain %85 %84 %16 %79
+%87 = OpLoad %80 %86 Aligned 16
+%88 = OpCompositeExtract %78 %87 0
+%89 = OpCompositeExtract %78 %87 1
+%91 = OpBitcast %83 %17
+%92 = OpInBoundsAccessChain %85 %91 %16 %90
+%93 = OpLoad %80 %92 Aligned 16
+%94 = OpCompositeExtract %78 %93 0
+%95 = OpCompositeExtract %78 %93 1
+%96 = OpConvertSToF %10 %88
+%97 = OpConvertSToF %10 %89
+%98 = OpConvertSToF %10 %94
+%99 = OpConvertSToF %10 %95
+%100 = OpFAdd %10 %74 %96
+%101 = OpFAdd %10 %75 %97
+%102 = OpFAdd %10 %76 %98
+%103 = OpFAdd %10 %77 %99
+%105 = OpAccessChain %104 %13 %16
+OpStore %105 %100
+%106 = OpAccessChain %104 %13 %31
+OpStore %106 %101
+%107 = OpAccessChain %104 %13 %79
+OpStore %107 %102
+%108 = OpAccessChain %104 %13 %90
+OpStore %108 %103
 OpReturn
 OpFunctionEnd
 #endif
diff --git a/reference/shaders/resources/cbv-legacy-fp16-fp64.root-descriptor.sm60.frag b/reference/shaders/resources/cbv-legacy-fp16-fp64.root-descriptor.sm60.frag
index ec79c5f..8b2179b 100644
--- a/reference/shaders/resources/cbv-legacy-fp16-fp64.root-descriptor.sm60.frag
+++ b/reference/shaders/resources/cbv-legacy-fp16-fp64.root-descriptor.sm60.frag
@@ -1,23 +1,18 @@
 #version 460
 #extension GL_ARB_gpu_shader_int64 : require
 #extension GL_EXT_buffer_reference : require
+#extension GL_EXT_buffer_reference_uvec2 : require
 
-struct AddCarry
+layout(buffer_reference) buffer PhysicalPointerFloat4NonWriteCBVArray;
+layout(buffer_reference) buffer PhysicalPointerUint642NonWriteCBVArray;
+layout(buffer_reference, buffer_reference_align = 16, std430) readonly buffer PhysicalPointerFloat4NonWriteCBVArray
 {
-    uint _m0;
-    uint _m1;
+    vec4 value[4096];
 };
 
-layout(buffer_reference) buffer PhysicalPointerFloat4NonWrite;
-layout(buffer_reference) buffer PhysicalPointerUint642NonWrite;
-layout(buffer_reference, std430) readonly buffer PhysicalPointerFloat4NonWrite
+layout(buffer_reference, buffer_reference_align = 16, std430) readonly buffer PhysicalPointerUint642NonWriteCBVArray
 {
-    vec4 value;
-};
-
-layout(buffer_reference, std430) readonly buffer PhysicalPointerUint642NonWrite
-{
-    u64vec2 value;
+    u64vec2 value[4096];
 };
 
 layout(push_constant, std430) uniform RootConstants
@@ -32,25 +27,15 @@ layout(location = 0) out vec4 SV_Target;
 
 void main()
 {
-    AddCarry _23;
-    _23._m0 = uaddCarry(registers._m0.x, 0u * 16u, _23._m1);
-    PhysicalPointerFloat4NonWrite _30 = PhysicalPointerFloat4NonWrite(uvec2(_23._m0, registers._m0.y + _23._m1));
-    AddCarry _42;
-    _42._m0 = uaddCarry(registers._m0.x, 1u * 16u, _42._m1);
-    PhysicalPointerFloat4NonWrite _47 = PhysicalPointerFloat4NonWrite(uvec2(_42._m0, registers._m0.y + _42._m1));
-    AddCarry _62;
-    _62._m0 = uaddCarry(registers._m0.x, 2u * 16u, _62._m1);
-    PhysicalPointerFloat4NonWrite _67 = PhysicalPointerFloat4NonWrite(uvec2(_62._m0, registers._m0.y + _62._m1));
-    AddCarry _83;
-    _83._m0 = uaddCarry(registers._m0.x, 3u * 16u, _83._m1);
-    PhysicalPointerUint642NonWrite _91 = PhysicalPointerUint642NonWrite(uvec2(_83._m0, registers._m0.y + _83._m1));
-    AddCarry _101;
-    _101._m0 = uaddCarry(registers._m0.x, 4u * 16u, _101._m1);
-    PhysicalPointerUint642NonWrite _106 = PhysicalPointerUint642NonWrite(uvec2(_101._m0, registers._m0.y + _101._m1));
-    SV_Target.x = ((_47.value.x + _30.value.x) + _67.value.x) + float(int64_t(_91.value.x));
-    SV_Target.y = ((_47.value.y + _30.value.y) + _67.value.y) + float(int64_t(_91.value.y));
-    SV_Target.z = ((_47.value.z + _30.value.z) + _67.value.z) + float(int64_t(_106.value.x));
-    SV_Target.w = ((_47.value.w + _30.value.w) + _67.value.w) + float(int64_t(_106.value.y));
+    PhysicalPointerFloat4NonWriteCBVArray _22 = PhysicalPointerFloat4NonWriteCBVArray(registers._m0);
+    PhysicalPointerFloat4NonWriteCBVArray _31 = PhysicalPointerFloat4NonWriteCBVArray(registers._m0);
+    PhysicalPointerFloat4NonWriteCBVArray _43 = PhysicalPointerFloat4NonWriteCBVArray(registers._m0);
+    PhysicalPointerUint642NonWriteCBVArray _60 = PhysicalPointerUint642NonWriteCBVArray(registers._m0);
+    PhysicalPointerUint642NonWriteCBVArray _67 = PhysicalPointerUint642NonWriteCBVArray(registers._m0);
+    SV_Target.x = ((_31.value[1u].x + _22.value[0u].x) + _43.value[2u].x) + float(int64_t(_60.value[3u].x));
+    SV_Target.y = ((_31.value[1u].y + _22.value[0u].y) + _43.value[2u].y) + float(int64_t(_60.value[3u].y));
+    SV_Target.z = ((_31.value[1u].z + _22.value[0u].z) + _43.value[2u].z) + float(int64_t(_67.value[4u].x));
+    SV_Target.w = ((_31.value[1u].w + _22.value[0u].w) + _43.value[2u].w) + float(int64_t(_67.value[4u].y));
 }
 
 
@@ -59,7 +44,7 @@ void main()
 ; SPIR-V
 ; Version: 1.3
 ; Generator: Unknown(30017); 21022
-; Bound: 126
+; Bound: 87
 ; Schema: 0
 OpCapability Shader
 OpCapability Int64
@@ -72,23 +57,24 @@ OpName %3 "main"
 OpName %7 "RootConstants"
 OpName %9 "registers"
 OpName %13 "SV_Target"
-OpName %22 "AddCarry"
-OpName %28 "PhysicalPointerFloat4NonWrite"
-OpMemberName %28 0 "value"
-OpName %89 "PhysicalPointerUint642NonWrite"
-OpMemberName %89 0 "value"
+OpName %20 "PhysicalPointerFloat4NonWriteCBVArray"
+OpMemberName %20 0 "value"
+OpName %58 "PhysicalPointerUint642NonWriteCBVArray"
+OpMemberName %58 0 "value"
 OpDecorate %7 Block
 OpMemberDecorate %7 0 Offset 0
 OpMemberDecorate %7 1 Offset 8
 OpMemberDecorate %7 2 Offset 16
 OpMemberDecorate %7 3 Offset 24
 OpDecorate %13 Location 0
-OpMemberDecorate %28 0 Offset 0
-OpDecorate %28 Block
-OpMemberDecorate %28 0 NonWritable
-OpMemberDecorate %89 0 Offset 0
-OpDecorate %89 Block
-OpMemberDecorate %89 0 NonWritable
+OpDecorate %19 ArrayStride 16
+OpMemberDecorate %20 0 Offset 0
+OpDecorate %20 Block
+OpMemberDecorate %20 0 NonWritable
+OpDecorate %57 ArrayStride 16
+OpMemberDecorate %58 0 Offset 0
+OpDecorate %58 Block
+OpMemberDecorate %58 0 NonWritable
 %1 = OpTypeVoid
 %2 = OpTypeFunction %1
 %5 = OpTypeInt 32 0
@@ -102,122 +88,83 @@ OpMemberDecorate %89 0 NonWritable
 %13 = OpVariable %12 Output
 %14 = OpTypePointer PushConstant %6
 %16 = OpConstant %5 0
-%19 = OpConstant %5 16
-%22 = OpTypeStruct %5 %5
-%28 = OpTypeStruct %11
-%29 = OpTypePointer PhysicalStorageBuffer %28
-%31 = OpTypePointer PhysicalStorageBuffer %11
-%38 = OpConstant %5 1
-%58 = OpConstant %5 2
-%78 = OpConstant %5 3
-%80 = OpTypeInt 64 0
-%88 = OpTypeVector %80 2
-%89 = OpTypeStruct %88
-%90 = OpTypePointer PhysicalStorageBuffer %89
-%92 = OpTypePointer PhysicalStorageBuffer %88
-%97 = OpConstant %5 4
-%119 = OpTypePointer Output %10
+%18 = OpConstant %5 4096
+%19 = OpTypeArray %11 %18
+%20 = OpTypeStruct %19
+%21 = OpTypePointer PhysicalStorageBuffer %20
+%23 = OpTypePointer PhysicalStorageBuffer %11
+%30 = OpConstant %5 1
+%42 = OpConstant %5 2
+%54 = OpTypeInt 64 0
+%55 = OpConstant %5 3
+%56 = OpTypeVector %54 2
+%57 = OpTypeArray %56 %18
+%58 = OpTypeStruct %57
+%59 = OpTypePointer PhysicalStorageBuffer %58
+%61 = OpTypePointer PhysicalStorageBuffer %56
+%66 = OpConstant %5 4
+%80 = OpTypePointer Output %10
 %3 = OpFunction %1 None %2
 %4 = OpLabel
-OpBranch %124
-%124 = OpLabel
+OpBranch %85
+%85 = OpLabel
 %15 = OpAccessChain %14 %9 %16
 %17 = OpLoad %6 %15
-%18 = OpIMul %5 %16 %19
-%20 = OpCompositeExtract %5 %17 0
-%21 = OpCompositeExtract %5 %17 1
-%23 = OpIAddCarry %22 %20 %18
-%24 = OpCompositeExtract %5 %23 0
-%25 = OpCompositeExtract %5 %23 1
-%26 = OpIAdd %5 %21 %25
-%27 = OpCompositeConstruct %6 %24 %26
-%30 = OpBitcast %29 %27
-%32 = OpAccessChain %31 %30 %16
+%22 = OpBitcast %21 %17
+%24 = OpInBoundsAccessChain %23 %22 %16 %16
+%25 = OpLoad %11 %24 Aligned 16
+%26 = OpCompositeExtract %10 %25 0
+%27 = OpCompositeExtract %10 %25 1
+%28 = OpCompositeExtract %10 %25 2
+%29 = OpCompositeExtract %10 %25 3
+%31 = OpBitcast %21 %17
+%32 = OpInBoundsAccessChain %23 %31 %16 %30
 %33 = OpLoad %11 %32 Aligned 16
 %34 = OpCompositeExtract %10 %33 0
 %35 = OpCompositeExtract %10 %33 1
 %36 = OpCompositeExtract %10 %33 2
 %37 = OpCompositeExtract %10 %33 3
-%39 = OpIMul %5 %38 %19
-%40 = OpCompositeExtract %5 %17 0
-%41 = OpCompositeExtract %5 %17 1
-%42 = OpIAddCarry %22 %40 %39
-%43 = OpCompositeExtract %5 %42 0
-%44 = OpCompositeExtract %5 %42 1
-%45 = OpIAdd %5 %41 %44
-%46 = OpCompositeConstruct %6 %43 %45
-%47 = OpBitcast %29 %46
-%48 = OpAccessChain %31 %47 %16
-%49 = OpLoad %11 %48 Aligned 16
-%50 = OpCompositeExtract %10 %49 0
-%51 = OpCompositeExtract %10 %49 1
-%52 = OpCompositeExtract %10 %49 2
-%53 = OpCompositeExtract %10 %49 3
-%54 = OpFAdd %10 %50 %34
-%55 = OpFAdd %10 %51 %35
-%56 = OpFAdd %10 %52 %36
-%57 = OpFAdd %10 %53 %37
-%59 = OpIMul %5 %58 %19
-%60 = OpCompositeExtract %5 %17 0
-%61 = OpCompositeExtract %5 %17 1
-%62 = OpIAddCarry %22 %60 %59
-%63 = OpCompositeExtract %5 %62 0
-%64 = OpCompositeExtract %5 %62 1
-%65 = OpIAdd %5 %61 %64
-%66 = OpCompositeConstruct %6 %63 %65
-%67 = OpBitcast %29 %66
-%68 = OpAccessChain %31 %67 %16
-%69 = OpLoad %11 %68 Aligned 16
-%70 = OpCompositeExtract %10 %69 0
-%71 = OpCompositeExtract %10 %69 1
-%72 = OpCompositeExtract %10 %69 2
-%73 = OpCompositeExtract %10 %69 3
-%74 = OpFAdd %10 %54 %70
-%75 = OpFAdd %10 %55 %71
-%76 = OpFAdd %10 %56 %72
-%77 = OpFAdd %10 %57 %73
-%79 = OpIMul %5 %78 %19
-%81 = OpCompositeExtract %5 %17 0
-%82 = OpCompositeExtract %5 %17 1
-%83 = OpIAddCarry %22 %81 %79
-%84 = OpCompositeExtract %5 %83 0
-%85 = OpCompositeExtract %5 %83 1
-%86 = OpIAdd %5 %82 %85
-%87 = OpCompositeConstruct %6 %84 %86
-%91 = OpBitcast %90 %87
-%93 = OpAccessChain %92 %91 %16
-%94 = OpLoad %88 %93 Aligned 16
-%95 = OpCompositeExtract %80 %94 0
-%96 = OpCompositeExtract %80 %94 1
-%98 = OpIMul %5 %97 %19
-%99 = OpCompositeExtract %5 %17 0
-%100 = OpCompositeExtract %5 %17 1
-%101 = OpIAddCarry %22 %99 %98
-%102 = OpCompositeExtract %5 %101 0
-%103 = OpCompositeExtract %5 %101 1
-%104 = OpIAdd %5 %100 %103
-%105 = OpCompositeConstruct %6 %102 %104
-%106 = OpBitcast %90 %105
-%107 = OpAccessChain %92 %106 %16
-%108 = OpLoad %88 %107 Aligned 16
-%109 = OpCompositeExtract %80 %108 0
-%110 = OpCompositeExtract %80 %108 1
-%111 = OpConvertSToF %10 %95
-%112 = OpConvertSToF %10 %96
-%113 = OpConvertSToF %10 %109
-%114 = OpConvertSToF %10 %110
-%115 = OpFAdd %10 %74 %111
-%116 = OpFAdd %10 %75 %112
-%117 = OpFAdd %10 %76 %113
-%118 = OpFAdd %10 %77 %114
-%120 = OpAccessChain %119 %13 %16
-OpStore %120 %115
-%121 = OpAccessChain %119 %13 %38
-OpStore %121 %116
-%122 = OpAccessChain %119 %13 %58
-OpStore %122 %117
-%123 = OpAccessChain %119 %13 %78
-OpStore %123 %118
+%38 = OpFAdd %10 %34 %26
+%39 = OpFAdd %10 %35 %27
+%40 = OpFAdd %10 %36 %28
+%41 = OpFAdd %10 %37 %29
+%43 = OpBitcast %21 %17
+%44 = OpInBoundsAccessChain %23 %43 %16 %42
+%45 = OpLoad %11 %44 Aligned 16
+%46 = OpCompositeExtract %10 %45 0
+%47 = OpCompositeExtract %10 %45 1
+%48 = OpCompositeExtract %10 %45 2
+%49 = OpCompositeExtract %10 %45 3
+%50 = OpFAdd %10 %38 %46
+%51 = OpFAdd %10 %39 %47
+%52 = OpFAdd %10 %40 %48
+%53 = OpFAdd %10 %41 %49
+%60 = OpBitcast %59 %17
+%62 = OpInBoundsAccessChain %61 %60 %16 %55
+%63 = OpLoad %56 %62 Aligned 16
+%64 = OpCompositeExtract %54 %63 0
+%65 = OpCompositeExtract %54 %63 1
+%67 = OpBitcast %59 %17
+%68 = OpInBoundsAccessChain %61 %67 %16 %66
+%69 = OpLoad %56 %68 Aligned 16
+%70 = OpCompositeExtract %54 %69 0
+%71 = OpCompositeExtract %54 %69 1
+%72 = OpConvertSToF %10 %64
+%73 = OpConvertSToF %10 %65
+%74 = OpConvertSToF %10 %70
+%75 = OpConvertSToF %10 %71
+%76 = OpFAdd %10 %50 %72
+%77 = OpFAdd %10 %51 %73
+%78 = OpFAdd %10 %52 %74
+%79 = OpFAdd %10 %53 %75
+%81 = OpAccessChain %80 %13 %16
+OpStore %81 %76
+%82 = OpAccessChain %80 %13 %30
+OpStore %82 %77
+%83 = OpAccessChain %80 %13 %42
+OpStore %83 %78
+%84 = OpAccessChain %80 %13 %55
+OpStore %84 %79
 OpReturn
 OpFunctionEnd
 #endif
diff --git a/reference/shaders/resources/cbv-legacy-fp16-fp64.root-descriptor.sm60.native-fp16.frag b/reference/shaders/resources/cbv-legacy-fp16-fp64.root-descriptor.sm60.native-fp16.frag
index 4f60fc9..6ff9f96 100644
--- a/reference/shaders/resources/cbv-legacy-fp16-fp64.root-descriptor.sm60.native-fp16.frag
+++ b/reference/shaders/resources/cbv-legacy-fp16-fp64.root-descriptor.sm60.native-fp16.frag
@@ -9,23 +9,18 @@
 #extension GL_EXT_shader_16bit_storage : require
 #extension GL_ARB_gpu_shader_int64 : require
 #extension GL_EXT_buffer_reference : require
+#extension GL_EXT_buffer_reference_uvec2 : require
 
-struct AddCarry
+layout(buffer_reference) buffer PhysicalPointerFloat4NonWriteCBVArray;
+layout(buffer_reference) buffer PhysicalPointerUint642NonWriteCBVArray;
+layout(buffer_reference, buffer_reference_align = 16, std430) readonly buffer PhysicalPointerFloat4NonWriteCBVArray
 {
-    uint _m0;
-    uint _m1;
+    vec4 value[4096];
 };
 
-layout(buffer_reference) buffer PhysicalPointerFloat4NonWrite;
-layout(buffer_reference) buffer PhysicalPointerUint642NonWrite;
-layout(buffer_reference, std430) readonly buffer PhysicalPointerFloat4NonWrite
+layout(buffer_reference, buffer_reference_align = 16, std430) readonly buffer PhysicalPointerUint642NonWriteCBVArray
 {
-    vec4 value;
-};
-
-layout(buffer_reference, std430) readonly buffer PhysicalPointerUint642NonWrite
-{
-    u64vec2 value;
+    u64vec2 value[4096];
 };
 
 layout(push_constant, std430) uniform RootConstants
@@ -40,25 +35,15 @@ layout(location = 0) out vec4 SV_Target;
 
 void main()
 {
-    AddCarry _23;
-    _23._m0 = uaddCarry(registers._m0.x, 0u * 16u, _23._m1);
-    PhysicalPointerFloat4NonWrite _30 = PhysicalPointerFloat4NonWrite(uvec2(_23._m0, registers._m0.y + _23._m1));
-    AddCarry _42;
-    _42._m0 = uaddCarry(registers._m0.x, 1u * 16u, _42._m1);
-    f16vec4 _52 = f16vec4(PhysicalPointerFloat4NonWrite(uvec2(_42._m0, registers._m0.y + _42._m1)).value);
-    AddCarry _69;
-    _69._m0 = uaddCarry(registers._m0.x, 2u * 16u, _69._m1);
-    f16vec4 _77 = f16vec4(PhysicalPointerFloat4NonWrite(uvec2(_69._m0, registers._m0.y + _69._m1)).value);
-    AddCarry _95;
-    _95._m0 = uaddCarry(registers._m0.x, 3u * 16u, _95._m1);
-    PhysicalPointerUint642NonWrite _103 = PhysicalPointerUint642NonWrite(uvec2(_95._m0, registers._m0.y + _95._m1));
-    AddCarry _113;
-    _113._m0 = uaddCarry(registers._m0.x, 4u * 16u, _113._m1);
-    PhysicalPointerUint642NonWrite _118 = PhysicalPointerUint642NonWrite(uvec2(_113._m0, registers._m0.y + _113._m1));
-    SV_Target.x = ((float(_52.x) + _30.value.x) + float(_77.x)) + float(int64_t(_103.value.x));
-    SV_Target.y = ((float(_52.y) + _30.value.y) + float(_77.y)) + float(int64_t(_103.value.y));
-    SV_Target.z = ((float(_52.z) + _30.value.z) + float(_77.z)) + float(int64_t(_118.value.x));
-    SV_Target.w = ((float(_52.w) + _30.value.w) + float(_77.w)) + float(int64_t(_118.value.y));
+    PhysicalPointerFloat4NonWriteCBVArray _22 = PhysicalPointerFloat4NonWriteCBVArray(registers._m0);
+    f16vec4 _36 = f16vec4(PhysicalPointerFloat4NonWriteCBVArray(registers._m0).value[1u]);
+    f16vec4 _53 = f16vec4(PhysicalPointerFloat4NonWriteCBVArray(registers._m0).value[2u]);
+    PhysicalPointerUint642NonWriteCBVArray _72 = PhysicalPointerUint642NonWriteCBVArray(registers._m0);
+    PhysicalPointerUint642NonWriteCBVArray _79 = PhysicalPointerUint642NonWriteCBVArray(registers._m0);
+    SV_Target.x = ((float(_36.x) + _22.value[0u].x) + float(_53.x)) + float(int64_t(_72.value[3u].x));
+    SV_Target.y = ((float(_36.y) + _22.value[0u].y) + float(_53.y)) + float(int64_t(_72.value[3u].y));
+    SV_Target.z = ((float(_36.z) + _22.value[0u].z) + float(_53.z)) + float(int64_t(_79.value[4u].x));
+    SV_Target.w = ((float(_36.w) + _22.value[0u].w) + float(_53.w)) + float(int64_t(_79.value[4u].y));
 }
 
 
@@ -67,7 +52,7 @@ void main()
 ; SPIR-V
 ; Version: 1.3
 ; Generator: Unknown(30017); 21022
-; Bound: 138
+; Bound: 99
 ; Schema: 0
 OpCapability Shader
 OpCapability Float16
@@ -81,23 +66,24 @@ OpName %3 "main"
 OpName %7 "RootConstants"
 OpName %9 "registers"
 OpName %13 "SV_Target"
-OpName %22 "AddCarry"
-OpName %28 "PhysicalPointerFloat4NonWrite"
-OpMemberName %28 0 "value"
-OpName %101 "PhysicalPointerUint642NonWrite"
-OpMemberName %101 0 "value"
+OpName %20 "PhysicalPointerFloat4NonWriteCBVArray"
+OpMemberName %20 0 "value"
+OpName %70 "PhysicalPointerUint642NonWriteCBVArray"
+OpMemberName %70 0 "value"
 OpDecorate %7 Block
 OpMemberDecorate %7 0 Offset 0
 OpMemberDecorate %7 1 Offset 8
 OpMemberDecorate %7 2 Offset 16
 OpMemberDecorate %7 3 Offset 24
 OpDecorate %13 Location 0
-OpMemberDecorate %28 0 Offset 0
-OpDecorate %28 Block
-OpMemberDecorate %28 0 NonWritable
-OpMemberDecorate %101 0 Offset 0
-OpDecorate %101 Block
-OpMemberDecorate %101 0 NonWritable
+OpDecorate %19 ArrayStride 16
+OpMemberDecorate %20 0 Offset 0
+OpDecorate %20 Block
+OpMemberDecorate %20 0 NonWritable
+OpDecorate %69 ArrayStride 16
+OpMemberDecorate %70 0 Offset 0
+OpDecorate %70 Block
+OpMemberDecorate %70 0 NonWritable
 %1 = OpTypeVoid
 %2 = OpTypeFunction %1
 %5 = OpTypeInt 32 0
@@ -111,134 +97,95 @@ OpMemberDecorate %101 0 NonWritable
 %13 = OpVariable %12 Output
 %14 = OpTypePointer PushConstant %6
 %16 = OpConstant %5 0
-%19 = OpConstant %5 16
-%22 = OpTypeStruct %5 %5
-%28 = OpTypeStruct %11
-%29 = OpTypePointer PhysicalStorageBuffer %28
-%31 = OpTypePointer PhysicalStorageBuffer %11
-%38 = OpConstant %5 1
-%50 = OpTypeFloat 16
-%51 = OpTypeVector %50 4
-%65 = OpConstant %5 2
-%90 = OpConstant %5 3
-%92 = OpTypeInt 64 0
-%100 = OpTypeVector %92 2
-%101 = OpTypeStruct %100
-%102 = OpTypePointer PhysicalStorageBuffer %101
-%104 = OpTypePointer PhysicalStorageBuffer %100
-%109 = OpConstant %5 4
-%131 = OpTypePointer Output %10
+%18 = OpConstant %5 4096
+%19 = OpTypeArray %11 %18
+%20 = OpTypeStruct %19
+%21 = OpTypePointer PhysicalStorageBuffer %20
+%23 = OpTypePointer PhysicalStorageBuffer %11
+%30 = OpConstant %5 1
+%34 = OpTypeFloat 16
+%35 = OpTypeVector %34 4
+%49 = OpConstant %5 2
+%66 = OpTypeInt 64 0
+%67 = OpConstant %5 3
+%68 = OpTypeVector %66 2
+%69 = OpTypeArray %68 %18
+%70 = OpTypeStruct %69
+%71 = OpTypePointer PhysicalStorageBuffer %70
+%73 = OpTypePointer PhysicalStorageBuffer %68
+%78 = OpConstant %5 4
+%92 = OpTypePointer Output %10
 %3 = OpFunction %1 None %2
 %4 = OpLabel
-OpBranch %136
-%136 = OpLabel
+OpBranch %97
+%97 = OpLabel
 %15 = OpAccessChain %14 %9 %16
 %17 = OpLoad %6 %15
-%18 = OpIMul %5 %16 %19
-%20 = OpCompositeExtract %5 %17 0
-%21 = OpCompositeExtract %5 %17 1
-%23 = OpIAddCarry %22 %20 %18
-%24 = OpCompositeExtract %5 %23 0
-%25 = OpCompositeExtract %5 %23 1
-%26 = OpIAdd %5 %21 %25
-%27 = OpCompositeConstruct %6 %24 %26
-%30 = OpBitcast %29 %27
-%32 = OpAccessChain %31 %30 %16
+%22 = OpBitcast %21 %17
+%24 = OpInBoundsAccessChain %23 %22 %16 %16
+%25 = OpLoad %11 %24 Aligned 16
+%26 = OpCompositeExtract %10 %25 0
+%27 = OpCompositeExtract %10 %25 1
+%28 = OpCompositeExtract %10 %25 2
+%29 = OpCompositeExtract %10 %25 3
+%31 = OpBitcast %21 %17
+%32 = OpInBoundsAccessChain %23 %31 %16 %30
 %33 = OpLoad %11 %32 Aligned 16
-%34 = OpCompositeExtract %10 %33 0
-%35 = OpCompositeExtract %10 %33 1
-%36 = OpCompositeExtract %10 %33 2
-%37 = OpCompositeExtract %10 %33 3
-%39 = OpIMul %5 %38 %19
-%40 = OpCompositeExtract %5 %17 0
-%41 = OpCompositeExtract %5 %17 1
-%42 = OpIAddCarry %22 %40 %39
-%43 = OpCompositeExtract %5 %42 0
-%44 = OpCompositeExtract %5 %42 1
-%45 = OpIAdd %5 %41 %44
-%46 = OpCompositeConstruct %6 %43 %45
-%47 = OpBitcast %29 %46
-%48 = OpAccessChain %31 %47 %16
-%49 = OpLoad %11 %48 Aligned 16
-%52 = OpFConvert %51 %49
-%53 = OpCompositeExtract %50 %52 0
-%54 = OpCompositeExtract %50 %52 1
-%55 = OpCompositeExtract %50 %52 2
-%56 = OpCompositeExtract %50 %52 3
-%57 = OpFConvert %10 %53
+%36 = OpFConvert %35 %33
+%37 = OpCompositeExtract %34 %36 0
+%38 = OpCompositeExtract %34 %36 1
+%39 = OpCompositeExtract %34 %36 2
+%40 = OpCompositeExtract %34 %36 3
+%41 = OpFConvert %10 %37
+%42 = OpFConvert %10 %38
+%43 = OpFConvert %10 %39
+%44 = OpFConvert %10 %40
+%45 = OpFAdd %10 %41 %26
+%46 = OpFAdd %10 %42 %27
+%47 = OpFAdd %10 %43 %28
+%48 = OpFAdd %10 %44 %29
+%50 = OpBitcast %21 %17
+%51 = OpInBoundsAccessChain %23 %50 %16 %49
+%52 = OpLoad %11 %51 Aligned 16
+%53 = OpFConvert %35 %52
+%54 = OpCompositeExtract %34 %53 0
+%55 = OpCompositeExtract %34 %53 1
+%56 = OpCompositeExtract %34 %53 2
+%57 = OpCompositeExtract %34 %53 3
 %58 = OpFConvert %10 %54
 %59 = OpFConvert %10 %55
 %60 = OpFConvert %10 %56
-%61 = OpFAdd %10 %57 %34
-%62 = OpFAdd %10 %58 %35
-%63 = OpFAdd %10 %59 %36
-%64 = OpFAdd %10 %60 %37
-%66 = OpIMul %5 %65 %19
-%67 = OpCompositeExtract %5 %17 0
-%68 = OpCompositeExtract %5 %17 1
-%69 = OpIAddCarry %22 %67 %66
-%70 = OpCompositeExtract %5 %69 0
-%71 = OpCompositeExtract %5 %69 1
-%72 = OpIAdd %5 %68 %71
-%73 = OpCompositeConstruct %6 %70 %72
-%74 = OpBitcast %29 %73
-%75 = OpAccessChain %31 %74 %16
-%76 = OpLoad %11 %75 Aligned 16
-%77 = OpFConvert %51 %76
-%78 = OpCompositeExtract %50 %77 0
-%79 = OpCompositeExtract %50 %77 1
-%80 = OpCompositeExtract %50 %77 2
-%81 = OpCompositeExtract %50 %77 3
-%82 = OpFConvert %10 %78
-%83 = OpFConvert %10 %79
-%84 = OpFConvert %10 %80
-%85 = OpFConvert %10 %81
-%86 = OpFAdd %10 %61 %82
-%87 = OpFAdd %10 %62 %83
-%88 = OpFAdd %10 %63 %84
-%89 = OpFAdd %10 %64 %85
-%91 = OpIMul %5 %90 %19
-%93 = OpCompositeExtract %5 %17 0
-%94 = OpCompositeExtract %5 %17 1
-%95 = OpIAddCarry %22 %93 %91
-%96 = OpCompositeExtract %5 %95 0
-%97 = OpCompositeExtract %5 %95 1
-%98 = OpIAdd %5 %94 %97
-%99 = OpCompositeConstruct %6 %96 %98
-%103 = OpBitcast %102 %99
-%105 = OpAccessChain %104 %103 %16
-%106 = OpLoad %100 %105 Aligned 16
-%107 = OpCompositeExtract %92 %106 0
-%108 = OpCompositeExtract %92 %106 1
-%110 = OpIMul %5 %109 %19
-%111 = OpCompositeExtract %5 %17 0
-%112 = OpCompositeExtract %5 %17 1
-%113 = OpIAddCarry %22 %111 %110
-%114 = OpCompositeExtract %5 %113 0
-%115 = OpCompositeExtract %5 %113 1
-%116 = OpIAdd %5 %112 %115
-%117 = OpCompositeConstruct %6 %114 %116
-%118 = OpBitcast %102 %117
-%119 = OpAccessChain %104 %118 %16
-%120 = OpLoad %100 %119 Aligned 16
-%121 = OpCompositeExtract %92 %120 0
-%122 = OpCompositeExtract %92 %120 1
-%123 = OpConvertSToF %10 %107
-%124 = OpConvertSToF %10 %108
-%125 = OpConvertSToF %10 %121
-%126 = OpConvertSToF %10 %122
-%127 = OpFAdd %10 %86 %123
-%128 = OpFAdd %10 %87 %124
-%129 = OpFAdd %10 %88 %125
-%130 = OpFAdd %10 %89 %126
-%132 = OpAccessChain %131 %13 %16
-OpStore %132 %127
-%133 = OpAccessChain %131 %13 %38
-OpStore %133 %128
-%134 = OpAccessChain %131 %13 %65
-OpStore %134 %129
-%135 = OpAccessChain %131 %13 %90
-OpStore %135 %130
+%61 = OpFConvert %10 %57
+%62 = OpFAdd %10 %45 %58
+%63 = OpFAdd %10 %46 %59
+%64 = OpFAdd %10 %47 %60
+%65 = OpFAdd %10 %48 %61
+%72 = OpBitcast %71 %17
+%74 = OpInBoundsAccessChain %73 %72 %16 %67
+%75 = OpLoad %68 %74 Aligned 16
+%76 = OpCompositeExtract %66 %75 0
+%77 = OpCompositeExtract %66 %75 1
+%79 = OpBitcast %71 %17
+%80 = OpInBoundsAccessChain %73 %79 %16 %78
+%81 = OpLoad %68 %80 Aligned 16
+%82 = OpCompositeExtract %66 %81 0
+%83 = OpCompositeExtract %66 %81 1
+%84 = OpConvertSToF %10 %76
+%85 = OpConvertSToF %10 %77
+%86 = OpConvertSToF %10 %82
+%87 = OpConvertSToF %10 %83
+%88 = OpFAdd %10 %62 %84
+%89 = OpFAdd %10 %63 %85
+%90 = OpFAdd %10 %64 %86
+%91 = OpFAdd %10 %65 %87
+%93 = OpAccessChain %92 %13 %16
+OpStore %93 %88
+%94 = OpAccessChain %92 %13 %30
+OpStore %94 %89
+%95 = OpAccessChain %92 %13 %49
+OpStore %95 %90
+%96 = OpAccessChain %92 %13 %67
+OpStore %96 %91
 OpReturn
 OpFunctionEnd
 #endif
diff --git a/reference/shaders/resources/cbv.no-legacy-cbuf-layout.local-root-signature.rmiss b/reference/shaders/resources/cbv.no-legacy-cbuf-layout.local-root-signature.rmiss
index 5c816f7..0d79ef9 100644
--- a/reference/shaders/resources/cbv.no-legacy-cbuf-layout.local-root-signature.rmiss
+++ b/reference/shaders/resources/cbv.no-legacy-cbuf-layout.local-root-signature.rmiss
@@ -30,24 +30,18 @@ uvec4 _54;
 
 void main()
 {
-    vec4 _37 = _38;
+    vec4 _37;
     _37.x = uintBitsToFloat(SBT._m0[0u]);
-    vec4 _39 = _37;
-    _39.y = float(SBT._m0[1u]);
-    vec4 _40 = _39;
-    _40.z = float(int(SBT._m0[2u]));
-    vec4 _41 = _40;
-    _41.w = 1.0;
-    uvec4 _53 = _54;
+    _37.y = float(SBT._m0[1u]);
+    _37.z = float(int(SBT._m0[2u]));
+    _37.w = 1.0;
+    uvec4 _53;
     _53.x = uint(int(uintBitsToFloat(SBT._m1[0u])));
-    uvec4 _55 = _53;
-    _55.y = uint(int(uintBitsToFloat(SBT._m1[1u])));
-    uvec4 _56 = _55;
-    _56.z = SBT._m1[2u];
-    uvec4 _57 = _56;
-    _57.w = SBT._m0[2u];
-    payload._m0 = _41;
-    payload._m1 = _57;
+    _53.y = uint(int(uintBitsToFloat(SBT._m1[1u])));
+    _53.z = SBT._m1[2u];
+    _53.w = SBT._m0[2u];
+    payload._m0 = _37;
+    payload._m1 = _53;
 }
 
 
diff --git a/reference/shaders/resources/cbv.root-descriptor.no-legacy-cbuf-layout.frag b/reference/shaders/resources/cbv.root-descriptor.no-legacy-cbuf-layout.frag
index 8eb931e..fb6a0ab 100644
--- a/reference/shaders/resources/cbv.root-descriptor.no-legacy-cbuf-layout.frag
+++ b/reference/shaders/resources/cbv.root-descriptor.no-legacy-cbuf-layout.frag
@@ -9,29 +9,24 @@
 #endif
 #extension GL_EXT_shader_16bit_storage : require
 #extension GL_EXT_buffer_reference : require
+#extension GL_EXT_buffer_reference_uvec2 : require
 
-struct AddCarry
+layout(buffer_reference) buffer PhysicalPointerFloatNonWriteCBVArray;
+layout(buffer_reference) buffer PhysicalPointerUint64NonWriteCBVArray;
+layout(buffer_reference) buffer PhysicalPointerHalfNonWriteCBVArray;
+layout(buffer_reference, buffer_reference_align = 4, std430) readonly buffer PhysicalPointerFloatNonWriteCBVArray
 {
-    uint _m0;
-    uint _m1;
+    float value[16384];
 };
 
-layout(buffer_reference) buffer PhysicalPointerFloatNonWrite;
-layout(buffer_reference) buffer PhysicalPointerUint64NonWrite;
-layout(buffer_reference) buffer PhysicalPointerHalfNonWrite;
-layout(buffer_reference, std430) readonly buffer PhysicalPointerFloatNonWrite
+layout(buffer_reference, buffer_reference_align = 8, std430) readonly buffer PhysicalPointerUint64NonWriteCBVArray
 {
-    float value;
+    uint64_t value[8192];
 };
 
-layout(buffer_reference, std430) readonly buffer PhysicalPointerUint64NonWrite
+layout(buffer_reference, buffer_reference_align = 2, std430) readonly buffer PhysicalPointerHalfNonWriteCBVArray
 {
-    uint64_t value;
-};
-
-layout(buffer_reference, std430) readonly buffer PhysicalPointerHalfNonWrite
-{
-    float16_t value;
+    float16_t value[32768];
 };
 
 layout(push_constant, std430) uniform RootConstants
@@ -46,34 +41,10 @@ layout(location = 0) out vec4 SV_Target;
 
 void main()
 {
-    AddCarry _21;
-    _21._m0 = uaddCarry(registers._m0.x, 0u, _21._m1);
-    AddCarry _35;
-    _35._m0 = uaddCarry(registers._m0.x, 4u, _35._m1);
-    AddCarry _46;
-    _46._m0 = uaddCarry(registers._m0.x, 8u, _46._m1);
-    AddCarry _57;
-    _57._m0 = uaddCarry(registers._m0.x, 12u, _57._m1);
-    AddCarry _69;
-    _69._m0 = uaddCarry(registers._m0.x, 32u, _69._m1);
-    AddCarry _83;
-    _83._m0 = uaddCarry(registers._m0.x, 40u, _83._m1);
-    AddCarry _94;
-    _94._m0 = uaddCarry(registers._m0.x, 48u, _94._m1);
-    AddCarry _105;
-    _105._m0 = uaddCarry(registers._m0.x, 56u, _105._m1);
-    AddCarry _125;
-    _125._m0 = uaddCarry(registers._m0.x, 16u, _125._m1);
-    AddCarry _139;
-    _139._m0 = uaddCarry(registers._m0.x, 20u, _139._m1);
-    AddCarry _150;
-    _150._m0 = uaddCarry(registers._m0.x, 24u, _150._m1);
-    AddCarry _161;
-    _161._m0 = uaddCarry(registers._m0.x, 28u, _161._m1);
-    SV_Target.x = (float(int64_t(PhysicalPointerUint64NonWrite(uvec2(_69._m0, registers._m0.y + _69._m1)).value)) + PhysicalPointerFloatNonWrite(uvec2(_21._m0, registers._m0.y + _21._m1)).value) + float(PhysicalPointerHalfNonWrite(uvec2(_125._m0, registers._m0.y + _125._m1)).value);
-    SV_Target.y = (float(int64_t(PhysicalPointerUint64NonWrite(uvec2(_83._m0, registers._m0.y + _83._m1)).value)) + PhysicalPointerFloatNonWrite(uvec2(_35._m0, registers._m0.y + _35._m1)).value) + float(PhysicalPointerHalfNonWrite(uvec2(_139._m0, registers._m0.y + _139._m1)).value);
-    SV_Target.z = (float(int64_t(PhysicalPointerUint64NonWrite(uvec2(_94._m0, registers._m0.y + _94._m1)).value)) + PhysicalPointerFloatNonWrite(uvec2(_46._m0, registers._m0.y + _46._m1)).value) + float(PhysicalPointerHalfNonWrite(uvec2(_150._m0, registers._m0.y + _150._m1)).value);
-    SV_Target.w = (float(int64_t(PhysicalPointerUint64NonWrite(uvec2(_105._m0, registers._m0.y + _105._m1)).value)) + PhysicalPointerFloatNonWrite(uvec2(_57._m0, registers._m0.y + _57._m1)).value) + float(PhysicalPointerHalfNonWrite(uvec2(_161._m0, registers._m0.y + _161._m1)).value);
+    SV_Target.x = (float(int64_t(PhysicalPointerUint64NonWriteCBVArray(registers._m0).value[4u])) + PhysicalPointerFloatNonWriteCBVArray(registers._m0).value[0u]) + float(PhysicalPointerHalfNonWriteCBVArray(registers._m0).value[8u]);
+    SV_Target.y = (float(int64_t(PhysicalPointerUint64NonWriteCBVArray(registers._m0).value[5u])) + PhysicalPointerFloatNonWriteCBVArray(registers._m0).value[1u]) + float(PhysicalPointerHalfNonWriteCBVArray(registers._m0).value[10u]);
+    SV_Target.z = (float(int64_t(PhysicalPointerUint64NonWriteCBVArray(registers._m0).value[6u])) + PhysicalPointerFloatNonWriteCBVArray(registers._m0).value[2u]) + float(PhysicalPointerHalfNonWriteCBVArray(registers._m0).value[12u]);
+    SV_Target.w = (float(int64_t(PhysicalPointerUint64NonWriteCBVArray(registers._m0).value[7u])) + PhysicalPointerFloatNonWriteCBVArray(registers._m0).value[3u]) + float(PhysicalPointerHalfNonWriteCBVArray(registers._m0).value[14u]);
 }
 
 
@@ -82,7 +53,7 @@ void main()
 ; SPIR-V
 ; Version: 1.3
 ; Generator: Unknown(30017); 21022
-; Bound: 187
+; Bound: 105
 ; Schema: 0
 OpCapability Shader
 OpCapability Float16
@@ -96,28 +67,30 @@ OpName %3 "main"
 OpName %7 "RootConstants"
 OpName %9 "registers"
 OpName %13 "SV_Target"
-OpName %20 "AddCarry"
-OpName %26 "PhysicalPointerFloatNonWrite"
-OpMemberName %26 0 "value"
-OpName %74 "PhysicalPointerUint64NonWrite"
-OpMemberName %74 0 "value"
-OpName %130 "PhysicalPointerHalfNonWrite"
-OpMemberName %130 0 "value"
+OpName %20 "PhysicalPointerFloatNonWriteCBVArray"
+OpMemberName %20 0 "value"
+OpName %42 "PhysicalPointerUint64NonWriteCBVArray"
+OpMemberName %42 0 "value"
+OpName %72 "PhysicalPointerHalfNonWriteCBVArray"
+OpMemberName %72 0 "value"
 OpDecorate %7 Block
 OpMemberDecorate %7 0 Offset 0
 OpMemberDecorate %7 1 Offset 8
 OpMemberDecorate %7 2 Offset 16
 OpMemberDecorate %7 3 Offset 24
 OpDecorate %13 Location 0
-OpMemberDecorate %26 0 Offset 0
-OpDecorate %26 Block
-OpMemberDecorate %26 0 NonWritable
-OpMemberDecorate %74 0 Offset 0
-OpDecorate %74 Block
-OpMemberDecorate %74 0 NonWritable
-OpMemberDecorate %130 0 Offset 0
-OpDecorate %130 Block
-OpMemberDecorate %130 0 NonWritable
+OpDecorate %19 ArrayStride 4
+OpMemberDecorate %20 0 Offset 0
+OpDecorate %20 Block
+OpMemberDecorate %20 0 NonWritable
+OpDecorate %41 ArrayStride 8
+OpMemberDecorate %42 0 Offset 0
+OpDecorate %42 Block
+OpMemberDecorate %42 0 NonWritable
+OpDecorate %71 ArrayStride 2
+OpMemberDecorate %72 0 Offset 0
+OpDecorate %72 Block
+OpMemberDecorate %72 0 NonWritable
 %1 = OpTypeVoid
 %2 = OpTypeFunction %1
 %5 = OpTypeInt 32 0
@@ -131,183 +104,101 @@ OpMemberDecorate %130 0 NonWritable
 %13 = OpVariable %12 Output
 %14 = OpTypePointer PushConstant %6
 %16 = OpConstant %5 0
-%20 = OpTypeStruct %5 %5
-%26 = OpTypeStruct %10
-%27 = OpTypePointer PhysicalStorageBuffer %26
-%29 = OpTypePointer PhysicalStorageBuffer %10
-%32 = OpConstant %5 4
-%43 = OpConstant %5 8
-%54 = OpConstant %5 12
-%65 = OpConstant %5 32
-%66 = OpTypeInt 64 0
-%74 = OpTypeStruct %66
-%75 = OpTypePointer PhysicalStorageBuffer %74
-%77 = OpTypePointer PhysicalStorageBuffer %66
-%80 = OpConstant %5 40
-%91 = OpConstant %5 48
-%102 = OpConstant %5 56
-%121 = OpConstant %5 16
-%122 = OpTypeFloat 16
-%130 = OpTypeStruct %122
-%131 = OpTypePointer PhysicalStorageBuffer %130
-%133 = OpTypePointer PhysicalStorageBuffer %122
-%136 = OpConstant %5 20
-%147 = OpConstant %5 24
-%158 = OpConstant %5 28
-%177 = OpTypePointer Output %10
-%180 = OpConstant %5 1
-%182 = OpConstant %5 2
-%184 = OpConstant %5 3
+%18 = OpConstant %5 16384
+%19 = OpTypeArray %10 %18
+%20 = OpTypeStruct %19
+%21 = OpTypePointer PhysicalStorageBuffer %20
+%23 = OpTypePointer PhysicalStorageBuffer %10
+%26 = OpConstant %5 1
+%30 = OpConstant %5 2
+%34 = OpConstant %5 3
+%38 = OpTypeInt 64 0
+%39 = OpConstant %5 4
+%40 = OpConstant %5 8192
+%41 = OpTypeArray %38 %40
+%42 = OpTypeStruct %41
+%43 = OpTypePointer PhysicalStorageBuffer %42
+%45 = OpTypePointer PhysicalStorageBuffer %38
+%48 = OpConstant %5 5
+%52 = OpConstant %5 6
+%56 = OpConstant %5 7
+%68 = OpTypeFloat 16
+%69 = OpConstant %5 8
+%70 = OpConstant %5 32768
+%71 = OpTypeArray %68 %70
+%72 = OpTypeStruct %71
+%73 = OpTypePointer PhysicalStorageBuffer %72
+%75 = OpTypePointer PhysicalStorageBuffer %68
+%78 = OpConstant %5 10
+%82 = OpConstant %5 12
+%86 = OpConstant %5 14
+%98 = OpTypePointer Output %10
 %3 = OpFunction %1 None %2
 %4 = OpLabel
-OpBranch %185
-%185 = OpLabel
+OpBranch %103
+%103 = OpLabel
 %15 = OpAccessChain %14 %9 %16
 %17 = OpLoad %6 %15
-%18 = OpCompositeExtract %5 %17 0
-%19 = OpCompositeExtract %5 %17 1
-%21 = OpIAddCarry %20 %18 %16
-%22 = OpCompositeExtract %5 %21 0
-%23 = OpCompositeExtract %5 %21 1
-%24 = OpIAdd %5 %19 %23
-%25 = OpCompositeConstruct %6 %22 %24
-%28 = OpBitcast %27 %25
-%30 = OpAccessChain %29 %28 %16
-%31 = OpLoad %10 %30 Aligned 4
-%33 = OpCompositeExtract %5 %17 0
-%34 = OpCompositeExtract %5 %17 1
-%35 = OpIAddCarry %20 %33 %32
-%36 = OpCompositeExtract %5 %35 0
-%37 = OpCompositeExtract %5 %35 1
-%38 = OpIAdd %5 %34 %37
-%39 = OpCompositeConstruct %6 %36 %38
-%40 = OpBitcast %27 %39
-%41 = OpAccessChain %29 %40 %16
-%42 = OpLoad %10 %41 Aligned 4
-%44 = OpCompositeExtract %5 %17 0
-%45 = OpCompositeExtract %5 %17 1
-%46 = OpIAddCarry %20 %44 %43
-%47 = OpCompositeExtract %5 %46 0
-%48 = OpCompositeExtract %5 %46 1
-%49 = OpIAdd %5 %45 %48
-%50 = OpCompositeConstruct %6 %47 %49
-%51 = OpBitcast %27 %50
-%52 = OpAccessChain %29 %51 %16
-%53 = OpLoad %10 %52 Aligned 4
-%55 = OpCompositeExtract %5 %17 0
-%56 = OpCompositeExtract %5 %17 1
-%57 = OpIAddCarry %20 %55 %54
-%58 = OpCompositeExtract %5 %57 0
-%59 = OpCompositeExtract %5 %57 1
-%60 = OpIAdd %5 %56 %59
-%61 = OpCompositeConstruct %6 %58 %60
-%62 = OpBitcast %27 %61
-%63 = OpAccessChain %29 %62 %16
-%64 = OpLoad %10 %63 Aligned 4
-%67 = OpCompositeExtract %5 %17 0
-%68 = OpCompositeExtract %5 %17 1
-%69 = OpIAddCarry %20 %67 %65
-%70 = OpCompositeExtract %5 %69 0
-%71 = OpCompositeExtract %5 %69 1
-%72 = OpIAdd %5 %68 %71
-%73 = OpCompositeConstruct %6 %70 %72
-%76 = OpBitcast %75 %73
-%78 = OpAccessChain %77 %76 %16
-%79 = OpLoad %66 %78 Aligned 8
-%81 = OpCompositeExtract %5 %17 0
-%82 = OpCompositeExtract %5 %17 1
-%83 = OpIAddCarry %20 %81 %80
-%84 = OpCompositeExtract %5 %83 0
-%85 = OpCompositeExtract %5 %83 1
-%86 = OpIAdd %5 %82 %85
-%87 = OpCompositeConstruct %6 %84 %86
-%88 = OpBitcast %75 %87
-%89 = OpAccessChain %77 %88 %16
-%90 = OpLoad %66 %89 Aligned 8
-%92 = OpCompositeExtract %5 %17 0
-%93 = OpCompositeExtract %5 %17 1
-%94 = OpIAddCarry %20 %92 %91
-%95 = OpCompositeExtract %5 %94 0
-%96 = OpCompositeExtract %5 %94 1
-%97 = OpIAdd %5 %93 %96
-%98 = OpCompositeConstruct %6 %95 %97
-%99 = OpBitcast %75 %98
-%100 = OpAccessChain %77 %99 %16
-%101 = OpLoad %66 %100 Aligned 8
-%103 = OpCompositeExtract %5 %17 0
-%104 = OpCompositeExtract %5 %17 1
-%105 = OpIAddCarry %20 %103 %102
-%106 = OpCompositeExtract %5 %105 0
-%107 = OpCompositeExtract %5 %105 1
-%108 = OpIAdd %5 %104 %107
-%109 = OpCompositeConstruct %6 %106 %108
-%110 = OpBitcast %75 %109
-%111 = OpAccessChain %77 %110 %16
-%112 = OpLoad %66 %111 Aligned 8
-%113 = OpConvertSToF %10 %79
-%114 = OpConvertSToF %10 %90
-%115 = OpConvertSToF %10 %101
-%116 = OpConvertSToF %10 %112
-%117 = OpFAdd %10 %113 %31
-%118 = OpFAdd %10 %114 %42
-%119 = OpFAdd %10 %115 %53
-%120 = OpFAdd %10 %116 %64
-%123 = OpCompositeExtract %5 %17 0
-%124 = OpCompositeExtract %5 %17 1
-%125 = OpIAddCarry %20 %123 %121
-%126 = OpCompositeExtract %5 %125 0
-%127 = OpCompositeExtract %5 %125 1
-%128 = OpIAdd %5 %124 %127
-%129 = OpCompositeConstruct %6 %126 %128
-%132 = OpBitcast %131 %129
-%134 = OpAccessChain %133 %132 %16
-%135 = OpLoad %122 %134 Aligned 2
-%137 = OpCompositeExtract %5 %17 0
-%138 = OpCompositeExtract %5 %17 1
-%139 = OpIAddCarry %20 %137 %136
-%140 = OpCompositeExtract %5 %139 0
-%141 = OpCompositeExtract %5 %139 1
-%142 = OpIAdd %5 %138 %141
-%143 = OpCompositeConstruct %6 %140 %142
-%144 = OpBitcast %131 %143
-%145 = OpAccessChain %133 %144 %16
-%146 = OpLoad %122 %145 Aligned 2
-%148 = OpCompositeExtract %5 %17 0
-%149 = OpCompositeExtract %5 %17 1
-%150 = OpIAddCarry %20 %148 %147
-%151 = OpCompositeExtract %5 %150 0
-%152 = OpCompositeExtract %5 %150 1
-%153 = OpIAdd %5 %149 %152
-%154 = OpCompositeConstruct %6 %151 %153
-%155 = OpBitcast %131 %154
-%156 = OpAccessChain %133 %155 %16
-%157 = OpLoad %122 %156 Aligned 2
-%159 = OpCompositeExtract %5 %17 0
-%160 = OpCompositeExtract %5 %17 1
-%161 = OpIAddCarry %20 %159 %158
-%162 = OpCompositeExtract %5 %161 0
-%163 = OpCompositeExtract %5 %161 1
-%164 = OpIAdd %5 %160 %163
-%165 = OpCompositeConstruct %6 %162 %164
-%166 = OpBitcast %131 %165
-%167 = OpAccessChain %133 %166 %16
-%168 = OpLoad %122 %167 Aligned 2
-%169 = OpFConvert %10 %135
-%170 = OpFConvert %10 %146
-%171 = OpFConvert %10 %157
-%172 = OpFConvert %10 %168
-%173 = OpFAdd %10 %117 %169
-%174 = OpFAdd %10 %118 %170
-%175 = OpFAdd %10 %119 %171
-%176 = OpFAdd %10 %120 %172
-%178 = OpAccessChain %177 %13 %16
-OpStore %178 %173
-%179 = OpAccessChain %177 %13 %180
-OpStore %179 %174
-%181 = OpAccessChain %177 %13 %182
-OpStore %181 %175
-%183 = OpAccessChain %177 %13 %184
-OpStore %183 %176
+%22 = OpBitcast %21 %17
+%24 = OpInBoundsAccessChain %23 %22 %16 %16
+%25 = OpLoad %10 %24 Aligned 4
+%27 = OpBitcast %21 %17
+%28 = OpInBoundsAccessChain %23 %27 %16 %26
+%29 = OpLoad %10 %28 Aligned 4
+%31 = OpBitcast %21 %17
+%32 = OpInBoundsAccessChain %23 %31 %16 %30
+%33 = OpLoad %10 %32 Aligned 4
+%35 = OpBitcast %21 %17
+%36 = OpInBoundsAccessChain %23 %35 %16 %34
+%37 = OpLoad %10 %36 Aligned 4
+%44 = OpBitcast %43 %17
+%46 = OpInBoundsAccessChain %45 %44 %16 %39
+%47 = OpLoad %38 %46 Aligned 8
+%49 = OpBitcast %43 %17
+%50 = OpInBoundsAccessChain %45 %49 %16 %48
+%51 = OpLoad %38 %50 Aligned 8
+%53 = OpBitcast %43 %17
+%54 = OpInBoundsAccessChain %45 %53 %16 %52
+%55 = OpLoad %38 %54 Aligned 8
+%57 = OpBitcast %43 %17
+%58 = OpInBoundsAccessChain %45 %57 %16 %56
+%59 = OpLoad %38 %58 Aligned 8
+%60 = OpConvertSToF %10 %47
+%61 = OpConvertSToF %10 %51
+%62 = OpConvertSToF %10 %55
+%63 = OpConvertSToF %10 %59
+%64 = OpFAdd %10 %60 %25
+%65 = OpFAdd %10 %61 %29
+%66 = OpFAdd %10 %62 %33
+%67 = OpFAdd %10 %63 %37
+%74 = OpBitcast %73 %17
+%76 = OpInBoundsAccessChain %75 %74 %16 %69
+%77 = OpLoad %68 %76 Aligned 2
+%79 = OpBitcast %73 %17
+%80 = OpInBoundsAccessChain %75 %79 %16 %78
+%81 = OpLoad %68 %80 Aligned 2
+%83 = OpBitcast %73 %17
+%84 = OpInBoundsAccessChain %75 %83 %16 %82
+%85 = OpLoad %68 %84 Aligned 2
+%87 = OpBitcast %73 %17
+%88 = OpInBoundsAccessChain %75 %87 %16 %86
+%89 = OpLoad %68 %88 Aligned 2
+%90 = OpFConvert %10 %77
+%91 = OpFConvert %10 %81
+%92 = OpFConvert %10 %85
+%93 = OpFConvert %10 %89
+%94 = OpFAdd %10 %64 %90
+%95 = OpFAdd %10 %65 %91
+%96 = OpFAdd %10 %66 %92
+%97 = OpFAdd %10 %67 %93
+%99 = OpAccessChain %98 %13 %16
+OpStore %99 %94
+%100 = OpAccessChain %98 %13 %26
+OpStore %100 %95
+%101 = OpAccessChain %98 %13 %30
+OpStore %101 %96
+%102 = OpAccessChain %98 %13 %34
+OpStore %102 %97
 OpReturn
 OpFunctionEnd
 #endif
diff --git a/reference/shaders/resources/root-bda.root-descriptor.comp b/reference/shaders/resources/root-bda.root-descriptor.comp
index 547a2cf..c527517 100644
--- a/reference/shaders/resources/root-bda.root-descriptor.comp
+++ b/reference/shaders/resources/root-bda.root-descriptor.comp
@@ -8,54 +8,56 @@
 #endif
 #extension GL_EXT_shader_16bit_storage : require
 #extension GL_EXT_buffer_reference : require
+#extension GL_EXT_scalar_block_layout : require
+#extension GL_EXT_buffer_reference_uvec2 : require
 layout(local_size_x = 1, local_size_y = 1, local_size_z = 1) in;
 
-struct AddCarry
+layout(buffer_reference) buffer PhysicalPointerFloat4NonWriteArray;
+layout(buffer_reference) buffer PhysicalPointerFloat4NonWriteCBVArray;
+layout(buffer_reference) buffer PhysicalPointerFloat4CoherentArray;
+layout(buffer_reference) buffer PhysicalPointerHalfArray;
+layout(buffer_reference) buffer PhysicalPointerHalf2Array;
+layout(buffer_reference) buffer PhysicalPointerHalf3Array;
+layout(buffer_reference) buffer PhysicalPointerHalf4Array;
+layout(buffer_reference) buffer PhysicalPointerUint2Array;
+layout(buffer_reference, buffer_reference_align = 4, std430) readonly buffer PhysicalPointerFloat4NonWriteArray
 {
-    uint _m0;
-    uint _m1;
+    vec4 value[];
 };
 
-layout(buffer_reference) buffer PhysicalPointerFloat4NonWrite;
-layout(buffer_reference) buffer PhysicalPointerFloat4Coherent;
-layout(buffer_reference) buffer PhysicalPointerHalf;
-layout(buffer_reference) buffer PhysicalPointerHalf2;
-layout(buffer_reference) buffer PhysicalPointerHalf3;
-layout(buffer_reference) buffer PhysicalPointerHalf4;
-layout(buffer_reference) buffer PhysicalPointerUint2;
-layout(buffer_reference, std430) readonly buffer PhysicalPointerFloat4NonWrite
+layout(buffer_reference, buffer_reference_align = 16, std430) readonly buffer PhysicalPointerFloat4NonWriteCBVArray
 {
-    vec4 value;
+    vec4 value[4096];
 };
 
-layout(buffer_reference, std430) coherent buffer PhysicalPointerFloat4Coherent
+layout(buffer_reference, buffer_reference_align = 4, std430) coherent buffer PhysicalPointerFloat4CoherentArray
 {
-    vec4 value;
+    vec4 value[];
 };
 
-layout(buffer_reference, std430) buffer PhysicalPointerHalf
+layout(buffer_reference, buffer_reference_align = 2, std430) buffer PhysicalPointerHalfArray
 {
-    float16_t value;
+    float16_t value[];
 };
 
-layout(buffer_reference, std430) buffer PhysicalPointerHalf2
+layout(buffer_reference, buffer_reference_align = 2, std430) buffer PhysicalPointerHalf2Array
 {
-    f16vec2 value;
+    f16vec2 value[];
 };
 
-layout(buffer_reference, std430) buffer PhysicalPointerHalf3
+layout(buffer_reference, buffer_reference_align = 2, scalar) buffer PhysicalPointerHalf3Array
 {
-    f16vec3 value;
+    f16vec3 value[];
 };
 
-layout(buffer_reference, std430) buffer PhysicalPointerHalf4
+layout(buffer_reference, buffer_reference_align = 2, std430) buffer PhysicalPointerHalf4Array
 {
-    f16vec4 value;
+    f16vec4 value[];
 };
 
-layout(buffer_reference, std430) buffer PhysicalPointerUint2
+layout(buffer_reference, buffer_reference_align = 4, std430) buffer PhysicalPointerUint2Array
 {
-    uvec2 value;
+    uvec2 value[];
 };
 
 layout(push_constant, std430) uniform RootConstants
@@ -66,36 +68,19 @@ layout(push_constant, std430) uniform RootConstants
     uvec2 _m3;
 } registers;
 
-float16_t _100;
-uint _164;
+float16_t _72;
+uint _112;
 
 void main()
 {
-    AddCarry _39;
-    _39._m0 = uaddCarry(registers._m1.x, (gl_GlobalInvocationID.x * 16u) + 0u, _39._m1);
-    PhysicalPointerFloat4NonWrite _44 = PhysicalPointerFloat4NonWrite(uvec2(_39._m0, registers._m1.y + _39._m1));
-    AddCarry _55;
-    _55._m0 = uaddCarry(registers._m0.x, gl_GlobalInvocationID.x * 16u, _55._m1);
-    PhysicalPointerFloat4NonWrite _60 = PhysicalPointerFloat4NonWrite(uvec2(_55._m0, registers._m0.y + _55._m1));
-    AddCarry _77;
-    _77._m0 = uaddCarry(registers._m2.x, (gl_GlobalInvocationID.x * 16u) + 0u, _77._m1);
-    PhysicalPointerFloat4Coherent(uvec2(_77._m0, registers._m2.y + _77._m1)).value = vec4(_60.value.x + _44.value.x, _60.value.y + _44.value.y, _60.value.z + _44.value.z, _60.value.w + _44.value.w);
-    AddCarry _91;
-    _91._m0 = uaddCarry(registers._m3.x, gl_GlobalInvocationID.x << 1u, _91._m1);
-    PhysicalPointerHalf(uvec2(_91._m0, registers._m3.y + _91._m1)).value = float16_t(1.0);
-    AddCarry _107;
-    _107._m0 = uaddCarry(registers._m3.x, gl_GlobalInvocationID.x << 2u, _107._m1);
-    PhysicalPointerHalf2(uvec2(_107._m0, registers._m3.y + _107._m1)).value = f16vec2(float16_t(1.0), float16_t(2.0));
-    AddCarry _124;
-    _124._m0 = uaddCarry(registers._m3.x, gl_GlobalInvocationID.x * 6u, _124._m1);
-    PhysicalPointerHalf3(uvec2(_124._m0, registers._m3.y + _124._m1)).value = f16vec3(float16_t(1.0), float16_t(2.0), float16_t(3.0));
-    uint _134 = gl_GlobalInvocationID.x << 3u;
-    AddCarry _140;
-    _140._m0 = uaddCarry(registers._m3.x, _134, _140._m1);
-    PhysicalPointerHalf4(uvec2(_140._m0, registers._m3.y + _140._m1)).value = f16vec4(float16_t(1.0), float16_t(2.0), float16_t(3.0), float16_t(4.0));
-    AddCarry _154;
-    _154._m0 = uaddCarry(registers._m3.x, _134, _154._m1);
-    PhysicalPointerUint2(uvec2(_154._m0, registers._m3.y + _154._m1)).value = uvec2(4u, 5u);
+    PhysicalPointerFloat4NonWriteArray _34 = PhysicalPointerFloat4NonWriteArray(registers._m1);
+    PhysicalPointerFloat4NonWriteCBVArray _46 = PhysicalPointerFloat4NonWriteCBVArray(registers._m0);
+    PhysicalPointerFloat4CoherentArray(registers._m2).value[gl_GlobalInvocationID.x] = vec4(_46.value[gl_GlobalInvocationID.x].x + _34.value[gl_GlobalInvocationID.x].x, _46.value[gl_GlobalInvocationID.x].y + _34.value[gl_GlobalInvocationID.x].y, _46.value[gl_GlobalInvocationID.x].z + _34.value[gl_GlobalInvocationID.x].z, _46.value[gl_GlobalInvocationID.x].w + _34.value[gl_GlobalInvocationID.x].w);
+    PhysicalPointerHalfArray(registers._m3).value[gl_GlobalInvocationID.x] = float16_t(1.0);
+    PhysicalPointerHalf2Array(registers._m3).value[gl_GlobalInvocationID.x] = f16vec2(float16_t(1.0), float16_t(2.0));
+    PhysicalPointerHalf3Array(registers._m3).value[gl_GlobalInvocationID.x] = f16vec3(float16_t(1.0), float16_t(2.0), float16_t(3.0));
+    PhysicalPointerHalf4Array(registers._m3).value[gl_GlobalInvocationID.x] = f16vec4(float16_t(1.0), float16_t(2.0), float16_t(3.0), float16_t(4.0));
+    PhysicalPointerUint2Array(registers._m3).value[gl_GlobalInvocationID.x] = uvec2(4u, 5u);
 }
 
 
@@ -104,7 +89,7 @@ void main()
 ; SPIR-V
 ; Version: 1.3
 ; Generator: Unknown(30017); 21022
-; Bound: 168
+; Bound: 116
 ; Schema: 0
 OpCapability Shader
 OpCapability Float16
@@ -116,43 +101,55 @@ OpExecutionMode %3 LocalSize 1 1 1
 OpName %3 "main"
 OpName %7 "RootConstants"
 OpName %9 "registers"
-OpName %31 "PhysicalPointerFloat4NonWrite"
-OpMemberName %31 0 "value"
-OpName %38 "AddCarry"
-OpName %71 "PhysicalPointerFloat4Coherent"
-OpMemberName %71 0 "value"
-OpName %87 "PhysicalPointerHalf"
+OpName %32 "PhysicalPointerFloat4NonWriteArray"
+OpMemberName %32 0 "value"
+OpName %44 "PhysicalPointerFloat4NonWriteCBVArray"
+OpMemberName %44 0 "value"
+OpName %58 "PhysicalPointerFloat4CoherentArray"
+OpMemberName %58 0 "value"
+OpName %66 "PhysicalPointerHalfArray"
+OpMemberName %66 0 "value"
+OpName %76 "PhysicalPointerHalf2Array"
+OpMemberName %76 0 "value"
+OpName %87 "PhysicalPointerHalf3Array"
 OpMemberName %87 0 "value"
-OpName %103 "PhysicalPointerHalf2"
-OpMemberName %103 0 "value"
-OpName %120 "PhysicalPointerHalf3"
-OpMemberName %120 0 "value"
-OpName %136 "PhysicalPointerHalf4"
-OpMemberName %136 0 "value"
-OpName %150 "PhysicalPointerUint2"
-OpMemberName %150 0 "value"
+OpName %97 "PhysicalPointerHalf4Array"
+OpMemberName %97 0 "value"
+OpName %105 "PhysicalPointerUint2Array"
+OpMemberName %105 0 "value"
 OpDecorate %7 Block
 OpMemberDecorate %7 0 Offset 0
 OpMemberDecorate %7 1 Offset 8
 OpMemberDecorate %7 2 Offset 16
 OpMemberDecorate %7 3 Offset 24
 OpDecorate %25 BuiltIn GlobalInvocationId
-OpMemberDecorate %31 0 Offset 0
-OpDecorate %31 Block
-OpMemberDecorate %31 0 NonWritable
-OpMemberDecorate %71 0 Offset 0
-OpDecorate %71 Block
-OpMemberDecorate %71 0 Coherent
+OpDecorate %31 ArrayStride 16
+OpMemberDecorate %32 0 Offset 0
+OpDecorate %32 Block
+OpMemberDecorate %32 0 NonWritable
+OpDecorate %43 ArrayStride 16
+OpMemberDecorate %44 0 Offset 0
+OpDecorate %44 Block
+OpMemberDecorate %44 0 NonWritable
+OpDecorate %57 ArrayStride 16
+OpMemberDecorate %58 0 Offset 0
+OpDecorate %58 Block
+OpMemberDecorate %58 0 Coherent
+OpDecorate %65 ArrayStride 2
+OpMemberDecorate %66 0 Offset 0
+OpDecorate %66 Block
+OpDecorate %75 ArrayStride 4
+OpMemberDecorate %76 0 Offset 0
+OpDecorate %76 Block
+OpDecorate %86 ArrayStride 6
 OpMemberDecorate %87 0 Offset 0
 OpDecorate %87 Block
-OpMemberDecorate %103 0 Offset 0
-OpDecorate %103 Block
-OpMemberDecorate %120 0 Offset 0
-OpDecorate %120 Block
-OpMemberDecorate %136 0 Offset 0
-OpDecorate %136 Block
-OpMemberDecorate %150 0 Offset 0
-OpDecorate %150 Block
+OpDecorate %96 ArrayStride 8
+OpMemberDecorate %97 0 Offset 0
+OpDecorate %97 Block
+OpDecorate %104 ArrayStride 8
+OpMemberDecorate %105 0 Offset 0
+OpDecorate %105 Block
 %1 = OpTypeVoid
 %2 = OpTypeFunction %1
 %5 = OpTypeInt 32 0
@@ -171,45 +168,54 @@ OpDecorate %150 Block
 %26 = OpTypePointer Input %5
 %29 = OpTypeFloat 32
 %30 = OpTypeVector %29 4
-%31 = OpTypeStruct %30
-%32 = OpTypePointer PhysicalStorageBuffer %31
-%34 = OpConstant %5 16
-%38 = OpTypeStruct %5 %5
-%45 = OpTypePointer PhysicalStorageBuffer %30
-%71 = OpTypeStruct %30
-%72 = OpTypePointer PhysicalStorageBuffer %71
-%86 = OpTypeFloat 16
+%31 = OpTypeRuntimeArray %30
+%32 = OpTypeStruct %31
+%33 = OpTypePointer PhysicalStorageBuffer %32
+%35 = OpTypePointer PhysicalStorageBuffer %30
+%42 = OpConstant %5 4096
+%43 = OpTypeArray %30 %42
+%44 = OpTypeStruct %43
+%45 = OpTypePointer PhysicalStorageBuffer %44
+%57 = OpTypeRuntimeArray %30
+%58 = OpTypeStruct %57
+%59 = OpTypePointer PhysicalStorageBuffer %58
+%64 = OpTypeFloat 16
+%65 = OpTypeRuntimeArray %64
+%66 = OpTypeStruct %65
+%67 = OpTypePointer PhysicalStorageBuffer %66
+%69 = OpTypePointer PhysicalStorageBuffer %64
+%71 = OpConstant %64 0x1p+0
+%74 = OpTypeVector %64 2
+%75 = OpTypeRuntimeArray %74
+%76 = OpTypeStruct %75
+%77 = OpTypePointer PhysicalStorageBuffer %76
+%79 = OpTypePointer PhysicalStorageBuffer %74
+%81 = OpConstant %64 0x1p+1
+%84 = OpConstant %5 6
+%85 = OpTypeVector %64 3
+%86 = OpTypeRuntimeArray %85
 %87 = OpTypeStruct %86
 %88 = OpTypePointer PhysicalStorageBuffer %87
-%97 = OpTypePointer PhysicalStorageBuffer %86
-%99 = OpConstant %86 0x1p+0
-%102 = OpTypeVector %86 2
-%103 = OpTypeStruct %102
-%104 = OpTypePointer PhysicalStorageBuffer %103
-%113 = OpTypePointer PhysicalStorageBuffer %102
-%115 = OpConstant %86 0x1p+1
-%118 = OpConstant %5 6
-%119 = OpTypeVector %86 3
-%120 = OpTypeStruct %119
-%121 = OpTypePointer PhysicalStorageBuffer %120
-%130 = OpTypePointer PhysicalStorageBuffer %119
-%132 = OpConstant %86 0x1.8p+1
-%135 = OpTypeVector %86 4
-%136 = OpTypeStruct %135
-%137 = OpTypePointer PhysicalStorageBuffer %136
-%146 = OpTypePointer PhysicalStorageBuffer %135
-%148 = OpConstant %86 0x1p+2
-%150 = OpTypeStruct %6
-%151 = OpTypePointer PhysicalStorageBuffer %150
-%160 = OpTypePointer PhysicalStorageBuffer %6
-%162 = OpConstant %5 4
-%163 = OpConstant %5 5
+%90 = OpTypePointer PhysicalStorageBuffer %85
+%92 = OpConstant %64 0x1.8p+1
+%95 = OpTypeVector %64 4
+%96 = OpTypeRuntimeArray %95
+%97 = OpTypeStruct %96
+%98 = OpTypePointer PhysicalStorageBuffer %97
+%100 = OpTypePointer PhysicalStorageBuffer %95
+%102 = OpConstant %64 0x1p+2
+%104 = OpTypeRuntimeArray %6
+%105 = OpTypeStruct %104
+%106 = OpTypePointer PhysicalStorageBuffer %105
+%108 = OpTypePointer PhysicalStorageBuffer %6
+%110 = OpConstant %5 4
+%111 = OpConstant %5 5
 %3 = OpFunction %1 None %2
 %4 = OpLabel
-%100 = OpUndef %86
-%164 = OpUndef %5
-OpBranch %166
-%166 = OpLabel
+%72 = OpUndef %64
+%112 = OpUndef %5
+OpBranch %114
+%114 = OpLabel
 %11 = OpAccessChain %10 %9 %12
 %13 = OpLoad %6 %11
 %14 = OpAccessChain %10 %9 %15
@@ -220,112 +226,51 @@ OpBranch %166
 %22 = OpLoad %6 %20
 %27 = OpAccessChain %26 %25 %21
 %28 = OpLoad %5 %27
-%33 = OpIMul %5 %28 %34
-%35 = OpIAdd %5 %33 %21
-%36 = OpCompositeExtract %5 %19 0
-%37 = OpCompositeExtract %5 %19 1
-%39 = OpIAddCarry %38 %36 %35
-%40 = OpCompositeExtract %5 %39 0
-%41 = OpCompositeExtract %5 %39 1
-%42 = OpIAdd %5 %37 %41
-%43 = OpCompositeConstruct %6 %40 %42
-%44 = OpBitcast %32 %43
-%46 = OpAccessChain %45 %44 %21
-%47 = OpLoad %30 %46 Aligned 4
-%48 = OpCompositeExtract %29 %47 0
-%49 = OpCompositeExtract %29 %47 1
-%50 = OpCompositeExtract %29 %47 2
-%51 = OpCompositeExtract %29 %47 3
-%52 = OpIMul %5 %28 %34
-%53 = OpCompositeExtract %5 %22 0
-%54 = OpCompositeExtract %5 %22 1
-%55 = OpIAddCarry %38 %53 %52
-%56 = OpCompositeExtract %5 %55 0
-%57 = OpCompositeExtract %5 %55 1
-%58 = OpIAdd %5 %54 %57
-%59 = OpCompositeConstruct %6 %56 %58
-%60 = OpBitcast %32 %59
-%61 = OpAccessChain %45 %60 %21
-%62 = OpLoad %30 %61 Aligned 16
-%63 = OpCompositeExtract %29 %62 0
-%64 = OpCompositeExtract %29 %62 1
-%65 = OpCompositeExtract %29 %62 2
-%66 = OpCompositeExtract %29 %62 3
-%67 = OpFAdd %29 %63 %48
-%68 = OpFAdd %29 %64 %49
-%69 = OpFAdd %29 %65 %50
-%70 = OpFAdd %29 %66 %51
-%73 = OpIMul %5 %28 %34
-%74 = OpIAdd %5 %73 %21
-%75 = OpCompositeExtract %5 %16 0
-%76 = OpCompositeExtract %5 %16 1
-%77 = OpIAddCarry %38 %75 %74
-%78 = OpCompositeExtract %5 %77 0
-%79 = OpCompositeExtract %5 %77 1
-%80 = OpIAdd %5 %76 %79
-%81 = OpCompositeConstruct %6 %78 %80
-%82 = OpBitcast %72 %81
-%83 = OpAccessChain %45 %82 %21
-%84 = OpCompositeConstruct %30 %67 %68 %69 %70
-OpStore %83 %84 Aligned 4
-%85 = OpShiftLeftLogical %5 %28 %18
-%89 = OpCompositeExtract %5 %13 0
-%90 = OpCompositeExtract %5 %13 1
-%91 = OpIAddCarry %38 %89 %85
-%92 = OpCompositeExtract %5 %91 0
-%93 = OpCompositeExtract %5 %91 1
-%94 = OpIAdd %5 %90 %93
-%95 = OpCompositeConstruct %6 %92 %94
-%96 = OpBitcast %88 %95
-%98 = OpAccessChain %97 %96 %21
-OpStore %98 %99 Aligned 2
-%101 = OpShiftLeftLogical %5 %28 %15
-%105 = OpCompositeExtract %5 %13 0
-%106 = OpCompositeExtract %5 %13 1
-%107 = OpIAddCarry %38 %105 %101
-%108 = OpCompositeExtract %5 %107 0
-%109 = OpCompositeExtract %5 %107 1
-%110 = OpIAdd %5 %106 %109
-%111 = OpCompositeConstruct %6 %108 %110
-%112 = OpBitcast %104 %111
-%114 = OpAccessChain %113 %112 %21
-%116 = OpCompositeConstruct %102 %99 %115
-OpStore %114 %116 Aligned 2
-%117 = OpIMul %5 %28 %118
-%122 = OpCompositeExtract %5 %13 0
-%123 = OpCompositeExtract %5 %13 1
-%124 = OpIAddCarry %38 %122 %117
-%125 = OpCompositeExtract %5 %124 0
-%126 = OpCompositeExtract %5 %124 1
-%127 = OpIAdd %5 %123 %126
-%128 = OpCompositeConstruct %6 %125 %127
-%129 = OpBitcast %121 %128
-%131 = OpAccessChain %130 %129 %21
-%133 = OpCompositeConstruct %119 %99 %115 %132
-OpStore %131 %133 Aligned 2
-%134 = OpShiftLeftLogical %5 %28 %12
-%138 = OpCompositeExtract %5 %13 0
-%139 = OpCompositeExtract %5 %13 1
-%140 = OpIAddCarry %38 %138 %134
-%141 = OpCompositeExtract %5 %140 0
-%142 = OpCompositeExtract %5 %140 1
-%143 = OpIAdd %5 %139 %142
-%144 = OpCompositeConstruct %6 %141 %143
-%145 = OpBitcast %137 %144
-%147 = OpAccessChain %146 %145 %21
-%149 = OpCompositeConstruct %135 %99 %115 %132 %148
-OpStore %147 %149 Aligned 2
-%152 = OpCompositeExtract %5 %13 0
-%153 = OpCompositeExtract %5 %13 1
-%154 = OpIAddCarry %38 %152 %134
-%155 = OpCompositeExtract %5 %154 0
-%156 = OpCompositeExtract %5 %154 1
-%157 = OpIAdd %5 %153 %156
-%158 = OpCompositeConstruct %6 %155 %157
-%159 = OpBitcast %151 %158
-%161 = OpAccessChain %160 %159 %21
-%165 = OpCompositeConstruct %6 %162 %163
-OpStore %161 %165 Aligned 4
+%34 = OpBitcast %33 %19
+%36 = OpAccessChain %35 %34 %21 %28
+%37 = OpLoad %30 %36 Aligned 4
+%38 = OpCompositeExtract %29 %37 0
+%39 = OpCompositeExtract %29 %37 1
+%40 = OpCompositeExtract %29 %37 2
+%41 = OpCompositeExtract %29 %37 3
+%46 = OpBitcast %45 %22
+%47 = OpInBoundsAccessChain %35 %46 %21 %28
+%48 = OpLoad %30 %47 Aligned 16
+%49 = OpCompositeExtract %29 %48 0
+%50 = OpCompositeExtract %29 %48 1
+%51 = OpCompositeExtract %29 %48 2
+%52 = OpCompositeExtract %29 %48 3
+%53 = OpFAdd %29 %49 %38
+%54 = OpFAdd %29 %50 %39
+%55 = OpFAdd %29 %51 %40
+%56 = OpFAdd %29 %52 %41
+%60 = OpBitcast %59 %16
+%61 = OpAccessChain %35 %60 %21 %28
+%62 = OpCompositeConstruct %30 %53 %54 %55 %56
+OpStore %61 %62 Aligned 4
+%63 = OpShiftLeftLogical %5 %28 %18
+%68 = OpBitcast %67 %13
+%70 = OpAccessChain %69 %68 %21 %28
+OpStore %70 %71 Aligned 2
+%73 = OpShiftLeftLogical %5 %28 %15
+%78 = OpBitcast %77 %13
+%80 = OpAccessChain %79 %78 %21 %28
+%82 = OpCompositeConstruct %74 %71 %81
+OpStore %80 %82 Aligned 2
+%83 = OpIMul %5 %28 %84
+%89 = OpBitcast %88 %13
+%91 = OpAccessChain %90 %89 %21 %28
+%93 = OpCompositeConstruct %85 %71 %81 %92
+OpStore %91 %93 Aligned 2
+%94 = OpShiftLeftLogical %5 %28 %12
+%99 = OpBitcast %98 %13
+%101 = OpAccessChain %100 %99 %21 %28
+%103 = OpCompositeConstruct %95 %71 %81 %92 %102
+OpStore %101 %103 Aligned 2
+%107 = OpBitcast %106 %13
+%109 = OpAccessChain %108 %107 %21 %28
+%113 = OpCompositeConstruct %6 %110 %111
+OpStore %109 %113 Aligned 4
 OpReturn
 OpFunctionEnd
 #endif
diff --git a/reference/shaders/resources/root-bda.root-descriptor.sm60.comp b/reference/shaders/resources/root-bda.root-descriptor.sm60.comp
index 1ebdb75..a2fbd63 100644
--- a/reference/shaders/resources/root-bda.root-descriptor.sm60.comp
+++ b/reference/shaders/resources/root-bda.root-descriptor.sm60.comp
@@ -1,23 +1,24 @@
 #version 460
 #extension GL_EXT_buffer_reference : require
+#extension GL_EXT_buffer_reference_uvec2 : require
 layout(local_size_x = 1, local_size_y = 1, local_size_z = 1) in;
 
-struct AddCarry
+layout(buffer_reference) buffer PhysicalPointerFloat4NonWriteArray;
+layout(buffer_reference) buffer PhysicalPointerFloat4NonWriteCBVArray;
+layout(buffer_reference) buffer PhysicalPointerFloat4CoherentArray;
+layout(buffer_reference, buffer_reference_align = 4, std430) readonly buffer PhysicalPointerFloat4NonWriteArray
 {
-    uint _m0;
-    uint _m1;
+    vec4 value[];
 };
 
-layout(buffer_reference) buffer PhysicalPointerFloat4NonWrite;
-layout(buffer_reference) buffer PhysicalPointerFloat4Coherent;
-layout(buffer_reference, std430) readonly buffer PhysicalPointerFloat4NonWrite
+layout(buffer_reference, buffer_reference_align = 16, std430) readonly buffer PhysicalPointerFloat4NonWriteCBVArray
 {
-    vec4 value;
+    vec4 value[4096];
 };
 
-layout(buffer_reference, std430) coherent buffer PhysicalPointerFloat4Coherent
+layout(buffer_reference, buffer_reference_align = 4, std430) coherent buffer PhysicalPointerFloat4CoherentArray
 {
-    vec4 value;
+    vec4 value[];
 };
 
 layout(push_constant, std430) uniform RootConstants
@@ -30,15 +31,9 @@ layout(push_constant, std430) uniform RootConstants
 
 void main()
 {
-    AddCarry _36;
-    _36._m0 = uaddCarry(registers._m1.x, (gl_GlobalInvocationID.x * 16u) + 0u, _36._m1);
-    PhysicalPointerFloat4NonWrite _41 = PhysicalPointerFloat4NonWrite(uvec2(_36._m0, registers._m1.y + _36._m1));
-    AddCarry _52;
-    _52._m0 = uaddCarry(registers._m0.x, gl_GlobalInvocationID.x * 16u, _52._m1);
-    PhysicalPointerFloat4NonWrite _57 = PhysicalPointerFloat4NonWrite(uvec2(_52._m0, registers._m0.y + _52._m1));
-    AddCarry _74;
-    _74._m0 = uaddCarry(registers._m2.x, (gl_GlobalInvocationID.x * 16u) + 0u, _74._m1);
-    PhysicalPointerFloat4Coherent(uvec2(_74._m0, registers._m2.y + _74._m1)).value = vec4(_57.value.x + _41.value.x, _57.value.y + _41.value.y, _57.value.z + _41.value.z, _57.value.w + _41.value.w);
+    PhysicalPointerFloat4NonWriteArray _31 = PhysicalPointerFloat4NonWriteArray(registers._m1);
+    PhysicalPointerFloat4NonWriteCBVArray _43 = PhysicalPointerFloat4NonWriteCBVArray(registers._m0);
+    PhysicalPointerFloat4CoherentArray(registers._m2).value[gl_GlobalInvocationID.x] = vec4(_43.value[gl_GlobalInvocationID.x].x + _31.value[gl_GlobalInvocationID.x].x, _43.value[gl_GlobalInvocationID.x].y + _31.value[gl_GlobalInvocationID.x].y, _43.value[gl_GlobalInvocationID.x].z + _31.value[gl_GlobalInvocationID.x].z, _43.value[gl_GlobalInvocationID.x].w + _31.value[gl_GlobalInvocationID.x].w);
 }
 
 
@@ -47,7 +42,7 @@ void main()
 ; SPIR-V
 ; Version: 1.3
 ; Generator: Unknown(30017); 21022
-; Bound: 84
+; Bound: 62
 ; Schema: 0
 OpCapability Shader
 OpCapability PhysicalStorageBufferAddresses
@@ -58,23 +53,30 @@ OpExecutionMode %3 LocalSize 1 1 1
 OpName %3 "main"
 OpName %7 "RootConstants"
 OpName %9 "registers"
-OpName %28 "PhysicalPointerFloat4NonWrite"
-OpMemberName %28 0 "value"
-OpName %35 "AddCarry"
-OpName %68 "PhysicalPointerFloat4Coherent"
-OpMemberName %68 0 "value"
+OpName %29 "PhysicalPointerFloat4NonWriteArray"
+OpMemberName %29 0 "value"
+OpName %41 "PhysicalPointerFloat4NonWriteCBVArray"
+OpMemberName %41 0 "value"
+OpName %55 "PhysicalPointerFloat4CoherentArray"
+OpMemberName %55 0 "value"
 OpDecorate %7 Block
 OpMemberDecorate %7 0 Offset 0
 OpMemberDecorate %7 1 Offset 8
 OpMemberDecorate %7 2 Offset 16
 OpMemberDecorate %7 3 Offset 24
 OpDecorate %22 BuiltIn GlobalInvocationId
-OpMemberDecorate %28 0 Offset 0
-OpDecorate %28 Block
-OpMemberDecorate %28 0 NonWritable
-OpMemberDecorate %68 0 Offset 0
-OpDecorate %68 Block
-OpMemberDecorate %68 0 Coherent
+OpDecorate %28 ArrayStride 16
+OpMemberDecorate %29 0 Offset 0
+OpDecorate %29 Block
+OpMemberDecorate %29 0 NonWritable
+OpDecorate %40 ArrayStride 16
+OpMemberDecorate %41 0 Offset 0
+OpDecorate %41 Block
+OpMemberDecorate %41 0 NonWritable
+OpDecorate %54 ArrayStride 16
+OpMemberDecorate %55 0 Offset 0
+OpDecorate %55 Block
+OpMemberDecorate %55 0 Coherent
 %1 = OpTypeVoid
 %2 = OpTypeFunction %1
 %5 = OpTypeInt 32 0
@@ -92,17 +94,21 @@ OpMemberDecorate %68 0 Coherent
 %23 = OpTypePointer Input %5
 %26 = OpTypeFloat 32
 %27 = OpTypeVector %26 4
-%28 = OpTypeStruct %27
-%29 = OpTypePointer PhysicalStorageBuffer %28
-%31 = OpConstant %5 16
-%35 = OpTypeStruct %5 %5
-%42 = OpTypePointer PhysicalStorageBuffer %27
-%68 = OpTypeStruct %27
-%69 = OpTypePointer PhysicalStorageBuffer %68
+%28 = OpTypeRuntimeArray %27
+%29 = OpTypeStruct %28
+%30 = OpTypePointer PhysicalStorageBuffer %29
+%32 = OpTypePointer PhysicalStorageBuffer %27
+%39 = OpConstant %5 4096
+%40 = OpTypeArray %27 %39
+%41 = OpTypeStruct %40
+%42 = OpTypePointer PhysicalStorageBuffer %41
+%54 = OpTypeRuntimeArray %27
+%55 = OpTypeStruct %54
+%56 = OpTypePointer PhysicalStorageBuffer %55
 %3 = OpFunction %1 None %2
 %4 = OpLabel
-OpBranch %82
-%82 = OpLabel
+OpBranch %60
+%60 = OpLabel
 %11 = OpAccessChain %10 %9 %12
 %13 = OpLoad %6 %11
 %14 = OpAccessChain %10 %9 %15
@@ -111,54 +117,28 @@ OpBranch %82
 %19 = OpLoad %6 %17
 %24 = OpAccessChain %23 %22 %18
 %25 = OpLoad %5 %24
-%30 = OpIMul %5 %25 %31
-%32 = OpIAdd %5 %30 %18
-%33 = OpCompositeExtract %5 %16 0
-%34 = OpCompositeExtract %5 %16 1
-%36 = OpIAddCarry %35 %33 %32
-%37 = OpCompositeExtract %5 %36 0
-%38 = OpCompositeExtract %5 %36 1
-%39 = OpIAdd %5 %34 %38
-%40 = OpCompositeConstruct %6 %37 %39
-%41 = OpBitcast %29 %40
-%43 = OpAccessChain %42 %41 %18
-%44 = OpLoad %27 %43 Aligned 4
-%45 = OpCompositeExtract %26 %44 0
-%46 = OpCompositeExtract %26 %44 1
-%47 = OpCompositeExtract %26 %44 2
-%48 = OpCompositeExtract %26 %44 3
-%49 = OpIMul %5 %25 %31
-%50 = OpCompositeExtract %5 %19 0
-%51 = OpCompositeExtract %5 %19 1
-%52 = OpIAddCarry %35 %50 %49
-%53 = OpCompositeExtract %5 %52 0
-%54 = OpCompositeExtract %5 %52 1
-%55 = OpIAdd %5 %51 %54
-%56 = OpCompositeConstruct %6 %53 %55
-%57 = OpBitcast %29 %56
-%58 = OpAccessChain %42 %57 %18
-%59 = OpLoad %27 %58 Aligned 16
-%60 = OpCompositeExtract %26 %59 0
-%61 = OpCompositeExtract %26 %59 1
-%62 = OpCompositeExtract %26 %59 2
-%63 = OpCompositeExtract %26 %59 3
-%64 = OpFAdd %26 %60 %45
-%65 = OpFAdd %26 %61 %46
-%66 = OpFAdd %26 %62 %47
-%67 = OpFAdd %26 %63 %48
-%70 = OpIMul %5 %25 %31
-%71 = OpIAdd %5 %70 %18
-%72 = OpCompositeExtract %5 %13 0
-%73 = OpCompositeExtract %5 %13 1
-%74 = OpIAddCarry %35 %72 %71
-%75 = OpCompositeExtract %5 %74 0
-%76 = OpCompositeExtract %5 %74 1
-%77 = OpIAdd %5 %73 %76
-%78 = OpCompositeConstruct %6 %75 %77
-%79 = OpBitcast %69 %78
-%80 = OpAccessChain %42 %79 %18
-%81 = OpCompositeConstruct %27 %64 %65 %66 %67
-OpStore %80 %81 Aligned 4
+%31 = OpBitcast %30 %16
+%33 = OpAccessChain %32 %31 %18 %25
+%34 = OpLoad %27 %33 Aligned 4
+%35 = OpCompositeExtract %26 %34 0
+%36 = OpCompositeExtract %26 %34 1
+%37 = OpCompositeExtract %26 %34 2
+%38 = OpCompositeExtract %26 %34 3
+%43 = OpBitcast %42 %19
+%44 = OpInBoundsAccessChain %32 %43 %18 %25
+%45 = OpLoad %27 %44 Aligned 16
+%46 = OpCompositeExtract %26 %45 0
+%47 = OpCompositeExtract %26 %45 1
+%48 = OpCompositeExtract %26 %45 2
+%49 = OpCompositeExtract %26 %45 3
+%50 = OpFAdd %26 %46 %35
+%51 = OpFAdd %26 %47 %36
+%52 = OpFAdd %26 %48 %37
+%53 = OpFAdd %26 %49 %38
+%57 = OpBitcast %56 %13
+%58 = OpAccessChain %32 %57 %18 %25
+%59 = OpCompositeConstruct %27 %50 %51 %52 %53
+OpStore %58 %59 Aligned 4
 OpReturn
 OpFunctionEnd
 #endif
diff --git a/reference/shaders/resources/root-constant-with-bda.root-descriptor.root-constant.comp b/reference/shaders/resources/root-constant-with-bda.root-descriptor.root-constant.comp
index 0b0c145..68722cb 100644
--- a/reference/shaders/resources/root-constant-with-bda.root-descriptor.root-constant.comp
+++ b/reference/shaders/resources/root-constant-with-bda.root-descriptor.root-constant.comp
@@ -1,23 +1,18 @@
 #version 460
 #extension GL_EXT_buffer_reference : require
+#extension GL_EXT_buffer_reference_uvec2 : require
 layout(local_size_x = 1, local_size_y = 1, local_size_z = 1) in;
 
-struct AddCarry
+layout(buffer_reference) buffer PhysicalPointerFloat4NonWriteArray;
+layout(buffer_reference) buffer PhysicalPointerFloat4CoherentArray;
+layout(buffer_reference, buffer_reference_align = 4, std430) readonly buffer PhysicalPointerFloat4NonWriteArray
 {
-    uint _m0;
-    uint _m1;
+    vec4 value[];
 };
 
-layout(buffer_reference) buffer PhysicalPointerFloat4NonWrite;
-layout(buffer_reference) buffer PhysicalPointerFloat4Coherent;
-layout(buffer_reference, std430) readonly buffer PhysicalPointerFloat4NonWrite
+layout(buffer_reference, buffer_reference_align = 4, std430) coherent buffer PhysicalPointerFloat4CoherentArray
 {
-    vec4 value;
-};
-
-layout(buffer_reference, std430) coherent buffer PhysicalPointerFloat4Coherent
-{
-    vec4 value;
+    vec4 value[];
 };
 
 layout(push_constant, std430) uniform RootConstants
@@ -46,18 +41,14 @@ layout(push_constant, std430) uniform RootConstants
 
 void main()
 {
-    AddCarry _34;
-    _34._m0 = uaddCarry(registers._m1.x, (gl_GlobalInvocationID.x * 16u) + 0u, _34._m1);
-    PhysicalPointerFloat4NonWrite _39 = PhysicalPointerFloat4NonWrite(uvec2(_34._m0, registers._m1.y + _34._m1));
-    vec4 _62 = uintBitsToFloat(uvec4(registers._m4, registers._m5, registers._m6, registers._m7));
-    float _63 = _62.x;
-    vec4 _81 = uintBitsToFloat(uvec4(registers._m8, registers._m9, registers._m10, registers._m11));
-    float _82 = _81.y;
-    vec4 _100 = uintBitsToFloat(uvec4(registers._m12, registers._m13, registers._m14, registers._m15));
-    float _101 = _100.z;
-    AddCarry _112;
-    _112._m0 = uaddCarry(registers._m2.x, (gl_GlobalInvocationID.x * 16u) + 0u, _112._m1);
-    PhysicalPointerFloat4Coherent(uvec2(_112._m0, registers._m2.y + _112._m1)).value = vec4(((_63 + _39.value.x) + _82) + _101, ((_63 + _39.value.y) + _82) + _101, ((_63 + _39.value.z) + _82) + _101, ((_63 + _39.value.w) + _82) + _101);
+    PhysicalPointerFloat4NonWriteArray _29 = PhysicalPointerFloat4NonWriteArray(registers._m1);
+    vec4 _52 = uintBitsToFloat(uvec4(registers._m4, registers._m5, registers._m6, registers._m7));
+    float _53 = _52.x;
+    vec4 _71 = uintBitsToFloat(uvec4(registers._m8, registers._m9, registers._m10, registers._m11));
+    float _72 = _71.y;
+    vec4 _90 = uintBitsToFloat(uvec4(registers._m12, registers._m13, registers._m14, registers._m15));
+    float _91 = _90.z;
+    PhysicalPointerFloat4CoherentArray(registers._m2).value[gl_GlobalInvocationID.x] = vec4(((_53 + _29.value[gl_GlobalInvocationID.x].x) + _72) + _91, ((_53 + _29.value[gl_GlobalInvocationID.x].y) + _72) + _91, ((_53 + _29.value[gl_GlobalInvocationID.x].z) + _72) + _91, ((_53 + _29.value[gl_GlobalInvocationID.x].w) + _72) + _91);
 }
 
 
@@ -66,7 +57,7 @@ void main()
 ; SPIR-V
 ; Version: 1.3
 ; Generator: Unknown(30017); 21022
-; Bound: 122
+; Bound: 104
 ; Schema: 0
 OpCapability Shader
 OpCapability PhysicalStorageBufferAddresses
@@ -77,11 +68,10 @@ OpExecutionMode %3 LocalSize 1 1 1
 OpName %3 "main"
 OpName %7 "RootConstants"
 OpName %9 "registers"
-OpName %26 "PhysicalPointerFloat4NonWrite"
-OpMemberName %26 0 "value"
-OpName %33 "AddCarry"
-OpName %106 "PhysicalPointerFloat4Coherent"
-OpMemberName %106 0 "value"
+OpName %27 "PhysicalPointerFloat4NonWriteArray"
+OpMemberName %27 0 "value"
+OpName %97 "PhysicalPointerFloat4CoherentArray"
+OpMemberName %97 0 "value"
 OpDecorate %7 Block
 OpMemberDecorate %7 0 Offset 0
 OpMemberDecorate %7 1 Offset 8
@@ -104,12 +94,14 @@ OpMemberDecorate %7 17 Offset 84
 OpMemberDecorate %7 18 Offset 88
 OpMemberDecorate %7 19 Offset 92
 OpDecorate %19 BuiltIn GlobalInvocationId
-OpMemberDecorate %26 0 Offset 0
-OpDecorate %26 Block
-OpMemberDecorate %26 0 NonWritable
-OpMemberDecorate %106 0 Offset 0
-OpDecorate %106 Block
-OpMemberDecorate %106 0 Coherent
+OpDecorate %26 ArrayStride 16
+OpMemberDecorate %27 0 Offset 0
+OpDecorate %27 Block
+OpMemberDecorate %27 0 NonWritable
+OpDecorate %96 ArrayStride 16
+OpMemberDecorate %97 0 Offset 0
+OpDecorate %97 Block
+OpMemberDecorate %97 0 Coherent
 %1 = OpTypeVoid
 %2 = OpTypeFunction %1
 %5 = OpTypeInt 32 0
@@ -127,111 +119,93 @@ OpMemberDecorate %106 0 Coherent
 %22 = OpConstant %5 0
 %24 = OpTypeFloat 32
 %25 = OpTypeVector %24 4
-%26 = OpTypeStruct %25
-%27 = OpTypePointer PhysicalStorageBuffer %26
-%29 = OpConstant %5 16
-%33 = OpTypeStruct %5 %5
-%40 = OpTypePointer PhysicalStorageBuffer %25
-%47 = OpTypePointer PushConstant %5
-%49 = OpConstant %5 4
-%52 = OpConstant %5 5
-%55 = OpConstant %5 6
-%58 = OpConstant %5 7
-%60 = OpTypeVector %5 4
-%69 = OpConstant %5 8
-%72 = OpConstant %5 9
-%75 = OpConstant %5 10
-%78 = OpConstant %5 11
-%88 = OpConstant %5 12
-%91 = OpConstant %5 13
-%94 = OpConstant %5 14
-%97 = OpConstant %5 15
-%106 = OpTypeStruct %25
-%107 = OpTypePointer PhysicalStorageBuffer %106
+%26 = OpTypeRuntimeArray %25
+%27 = OpTypeStruct %26
+%28 = OpTypePointer PhysicalStorageBuffer %27
+%30 = OpTypePointer PhysicalStorageBuffer %25
+%37 = OpTypePointer PushConstant %5
+%39 = OpConstant %5 4
+%42 = OpConstant %5 5
+%45 = OpConstant %5 6
+%48 = OpConstant %5 7
+%50 = OpTypeVector %5 4
+%59 = OpConstant %5 8
+%62 = OpConstant %5 9
+%65 = OpConstant %5 10
+%68 = OpConstant %5 11
+%78 = OpConstant %5 12
+%81 = OpConstant %5 13
+%84 = OpConstant %5 14
+%87 = OpConstant %5 15
+%96 = OpTypeRuntimeArray %25
+%97 = OpTypeStruct %96
+%98 = OpTypePointer PhysicalStorageBuffer %97
 %3 = OpFunction %1 None %2
 %4 = OpLabel
-OpBranch %120
-%120 = OpLabel
+OpBranch %102
+%102 = OpLabel
 %11 = OpAccessChain %10 %9 %12
 %13 = OpLoad %6 %11
 %14 = OpAccessChain %10 %9 %15
 %16 = OpLoad %6 %14
 %21 = OpAccessChain %20 %19 %22
 %23 = OpLoad %5 %21
-%28 = OpIMul %5 %23 %29
-%30 = OpIAdd %5 %28 %22
-%31 = OpCompositeExtract %5 %16 0
-%32 = OpCompositeExtract %5 %16 1
-%34 = OpIAddCarry %33 %31 %30
-%35 = OpCompositeExtract %5 %34 0
-%36 = OpCompositeExtract %5 %34 1
-%37 = OpIAdd %5 %32 %36
-%38 = OpCompositeConstruct %6 %35 %37
-%39 = OpBitcast %27 %38
-%41 = OpAccessChain %40 %39 %22
-%42 = OpLoad %25 %41 Aligned 4
-%43 = OpCompositeExtract %24 %42 0
-%44 = OpCompositeExtract %24 %42 1
-%45 = OpCompositeExtract %24 %42 2
-%46 = OpCompositeExtract %24 %42 3
-%48 = OpAccessChain %47 %9 %49
-%50 = OpLoad %5 %48
-%51 = OpAccessChain %47 %9 %52
-%53 = OpLoad %5 %51
-%54 = OpAccessChain %47 %9 %55
-%56 = OpLoad %5 %54
-%57 = OpAccessChain %47 %9 %58
-%59 = OpLoad %5 %57
-%61 = OpCompositeConstruct %60 %50 %53 %56 %59
-%62 = OpBitcast %25 %61
-%63 = OpCompositeExtract %24 %62 0
-%64 = OpFAdd %24 %63 %43
-%65 = OpFAdd %24 %63 %44
-%66 = OpFAdd %24 %63 %45
-%67 = OpFAdd %24 %63 %46
-%68 = OpAccessChain %47 %9 %69
-%70 = OpLoad %5 %68
-%71 = OpAccessChain %47 %9 %72
-%73 = OpLoad %5 %71
-%74 = OpAccessChain %47 %9 %75
-%76 = OpLoad %5 %74
-%77 = OpAccessChain %47 %9 %78
+%29 = OpBitcast %28 %16
+%31 = OpAccessChain %30 %29 %22 %23
+%32 = OpLoad %25 %31 Aligned 4
+%33 = OpCompositeExtract %24 %32 0
+%34 = OpCompositeExtract %24 %32 1
+%35 = OpCompositeExtract %24 %32 2
+%36 = OpCompositeExtract %24 %32 3
+%38 = OpAccessChain %37 %9 %39
+%40 = OpLoad %5 %38
+%41 = OpAccessChain %37 %9 %42
+%43 = OpLoad %5 %41
+%44 = OpAccessChain %37 %9 %45
+%46 = OpLoad %5 %44
+%47 = OpAccessChain %37 %9 %48
+%49 = OpLoad %5 %47
+%51 = OpCompositeConstruct %50 %40 %43 %46 %49
+%52 = OpBitcast %25 %51
+%53 = OpCompositeExtract %24 %52 0
+%54 = OpFAdd %24 %53 %33
+%55 = OpFAdd %24 %53 %34
+%56 = OpFAdd %24 %53 %35
+%57 = OpFAdd %24 %53 %36
+%58 = OpAccessChain %37 %9 %59
+%60 = OpLoad %5 %58
+%61 = OpAccessChain %37 %9 %62
+%63 = OpLoad %5 %61
+%64 = OpAccessChain %37 %9 %65
+%66 = OpLoad %5 %64
+%67 = OpAccessChain %37 %9 %68
+%69 = OpLoad %5 %67
+%70 = OpCompositeConstruct %50 %60 %63 %66 %69
+%71 = OpBitcast %25 %70
+%72 = OpCompositeExtract %24 %71 1
+%73 = OpFAdd %24 %54 %72
+%74 = OpFAdd %24 %55 %72
+%75 = OpFAdd %24 %56 %72
+%76 = OpFAdd %24 %57 %72
+%77 = OpAccessChain %37 %9 %78
 %79 = OpLoad %5 %77
-%80 = OpCompositeConstruct %60 %70 %73 %76 %79
-%81 = OpBitcast %25 %80
-%82 = OpCompositeExtract %24 %81 1
-%83 = OpFAdd %24 %64 %82
-%84 = OpFAdd %24 %65 %82
-%85 = OpFAdd %24 %66 %82
-%86 = OpFAdd %24 %67 %82
-%87 = OpAccessChain %47 %9 %88
-%89 = OpLoad %5 %87
-%90 = OpAccessChain %47 %9 %91
-%92 = OpLoad %5 %90
-%93 = OpAccessChain %47 %9 %94
-%95 = OpLoad %5 %93
-%96 = OpAccessChain %47 %9 %97
-%98 = OpLoad %5 %96
-%99 = OpCompositeConstruct %60 %89 %92 %95 %98
-%100 = OpBitcast %25 %99
-%101 = OpCompositeExtract %24 %100 2
-%102 = OpFAdd %24 %83 %101
-%103 = OpFAdd %24 %84 %101
-%104 = OpFAdd %24 %85 %101
-%105 = OpFAdd %24 %86 %101
-%108 = OpIMul %5 %23 %29
-%109 = OpIAdd %5 %108 %22
-%110 = OpCompositeExtract %5 %13 0
-%111 = OpCompositeExtract %5 %13 1
-%112 = OpIAddCarry %33 %110 %109
-%113 = OpCompositeExtract %5 %112 0
-%114 = OpCompositeExtract %5 %112 1
-%115 = OpIAdd %5 %111 %114
-%116 = OpCompositeConstruct %6 %113 %115
-%117 = OpBitcast %107 %116
-%118 = OpAccessChain %40 %117 %22
-%119 = OpCompositeConstruct %25 %102 %103 %104 %105
-OpStore %118 %119 Aligned 4
+%80 = OpAccessChain %37 %9 %81
+%82 = OpLoad %5 %80
+%83 = OpAccessChain %37 %9 %84
+%85 = OpLoad %5 %83
+%86 = OpAccessChain %37 %9 %87
+%88 = OpLoad %5 %86
+%89 = OpCompositeConstruct %50 %79 %82 %85 %88
+%90 = OpBitcast %25 %89
+%91 = OpCompositeExtract %24 %90 2
+%92 = OpFAdd %24 %73 %91
+%93 = OpFAdd %24 %74 %91
+%94 = OpFAdd %24 %75 %91
+%95 = OpFAdd %24 %76 %91
+%99 = OpBitcast %98 %13
+%100 = OpAccessChain %30 %99 %22 %23
+%101 = OpCompositeConstruct %25 %92 %93 %94 %95
+OpStore %100 %101 Aligned 4
 OpReturn
 OpFunctionEnd
 #endif
diff --git a/reference/shaders/resources/rt-resources.bindless.local-root-signature.rmiss b/reference/shaders/resources/rt-resources.bindless.local-root-signature.rmiss
index e4815b2..04cc0ab 100644
--- a/reference/shaders/resources/rt-resources.bindless.local-root-signature.rmiss
+++ b/reference/shaders/resources/rt-resources.bindless.local-root-signature.rmiss
@@ -2,7 +2,9 @@
 #extension GL_EXT_ray_tracing : require
 #extension GL_EXT_buffer_reference : require
 #extension GL_EXT_nonuniform_qualifier : require
+#extension GL_EXT_scalar_block_layout : require
 #extension GL_EXT_samplerless_texture_functions : require
+#extension GL_EXT_buffer_reference_uvec2 : require
 
 struct _37
 {
@@ -10,52 +12,52 @@ struct _37
     uint _m1;
 };
 
-struct AddCarry
+layout(buffer_reference) buffer PhysicalPointerFloat4NonWriteCBVArray;
+layout(buffer_reference) buffer PhysicalPointerFloat4NonWriteArray;
+layout(buffer_reference) buffer PhysicalPointerUintNonWriteArray;
+layout(buffer_reference) buffer PhysicalPointerUint2NonWriteArray;
+layout(buffer_reference) buffer PhysicalPointerUint3NonWriteArray;
+layout(buffer_reference) buffer PhysicalPointerUint4NonWriteArray;
+layout(buffer_reference) buffer PhysicalPointerFloatArray;
+layout(buffer_reference) buffer PhysicalPointerUintArray;
+layout(buffer_reference, buffer_reference_align = 16, std430) readonly buffer PhysicalPointerFloat4NonWriteCBVArray
 {
-    uint _m0;
-    uint _m1;
+    vec4 value[4096];
 };
 
-layout(buffer_reference) buffer PhysicalPointerFloat4NonWrite;
-layout(buffer_reference) buffer PhysicalPointerUintNonWrite;
-layout(buffer_reference) buffer PhysicalPointerUint2NonWrite;
-layout(buffer_reference) buffer PhysicalPointerUint3NonWrite;
-layout(buffer_reference) buffer PhysicalPointerUint4NonWrite;
-layout(buffer_reference) buffer PhysicalPointerFloat;
-layout(buffer_reference) buffer PhysicalPointerUint;
-layout(buffer_reference, std430) readonly buffer PhysicalPointerFloat4NonWrite
+layout(buffer_reference, buffer_reference_align = 4, std430) readonly buffer PhysicalPointerFloat4NonWriteArray
 {
-    vec4 value;
+    vec4 value[];
 };
 
-layout(buffer_reference, std430) readonly buffer PhysicalPointerUintNonWrite
+layout(buffer_reference, buffer_reference_align = 4, std430) readonly buffer PhysicalPointerUintNonWriteArray
 {
-    uint value;
+    uint value[];
 };
 
-layout(buffer_reference, std430) readonly buffer PhysicalPointerUint2NonWrite
+layout(buffer_reference, buffer_reference_align = 4, std430) readonly buffer PhysicalPointerUint2NonWriteArray
 {
-    uvec2 value;
+    uvec2 value[];
 };
 
-layout(buffer_reference, std430) readonly buffer PhysicalPointerUint3NonWrite
+layout(buffer_reference, buffer_reference_align = 4, scalar) readonly buffer PhysicalPointerUint3NonWriteArray
 {
-    uvec3 value;
+    uvec3 value[];
 };
 
-layout(buffer_reference, std430) readonly buffer PhysicalPointerUint4NonWrite
+layout(buffer_reference, buffer_reference_align = 4, std430) readonly buffer PhysicalPointerUint4NonWriteArray
 {
-    uvec4 value;
+    uvec4 value[];
 };
 
-layout(buffer_reference, std430) buffer PhysicalPointerFloat
+layout(buffer_reference, buffer_reference_align = 4, std430) buffer PhysicalPointerFloatArray
 {
-    float value;
+    float value[];
 };
 
-layout(buffer_reference, std430) buffer PhysicalPointerUint
+layout(buffer_reference, buffer_reference_align = 4, std430) buffer PhysicalPointerUintArray
 {
-    uint value;
+    uint value[];
 };
 
 layout(shaderRecordEXT, std430) buffer SBTBlock
@@ -95,8 +97,8 @@ layout(set = 3, binding = 0, r32f) uniform readonly image2D _25[];
 layout(set = 2, binding = 0) uniform sampler _36[];
 layout(location = 0) rayPayloadInEXT _37 payload;
 
-vec4 _425;
-float _440;
+vec4 _372;
+float _378;
 
 void main()
 {
@@ -110,53 +112,29 @@ void main()
     uint _146 = ((SBT._m9.x >> 6u) + 13u) + _58;
     vec4 _169 = uintBitsToFloat(uvec4(SBT._m0[0u], SBT._m0[1u], SBT._m0[2u], SBT._m0[3u]));
     vec4 _182 = uintBitsToFloat(uvec4(SBT._m0[4u], 0u, 0u, 0u));
-    AddCarry _196;
-    _196._m0 = uaddCarry(SBT._m6.x, 1u * 16u, _196._m1);
-    PhysicalPointerFloat4NonWrite _203 = PhysicalPointerFloat4NonWrite(uvec2(_196._m0, SBT._m6.y + _196._m1));
-    vec4 _232 = textureLod(nonuniformEXT(sampler2D(_21[registers._m0 + _59], _36[(SBT._m10.x >> 5u) + 13u])), vec2(0.5), 0.0);
-    vec4 _258 = textureLod(nonuniformEXT(sampler2D(_21[registers._m0 + _58], _36[((SBT._m10.x >> 5u) + 14u) + (_58 ^ 1u)])), vec2(0.5), 0.0);
-    AddCarry _274;
-    _274._m0 = uaddCarry(SBT._m2.x, (_58 * 16u) + 0u, _274._m1);
-    PhysicalPointerFloat4NonWrite _279 = PhysicalPointerFloat4NonWrite(uvec2(_274._m0, SBT._m2.y + _274._m1));
-    uint _290 = _58 << 2u;
-    AddCarry _297;
-    _297._m0 = uaddCarry(SBT._m4.x, _290, _297._m1);
-    float _306 = uintBitsToFloat(PhysicalPointerUintNonWrite(uvec2(_297._m0, SBT._m4.y + _297._m1)).value);
-    AddCarry _316;
-    _316._m0 = uaddCarry(SBT._m4.x, _58 << 3u, _316._m1);
-    PhysicalPointerUint2NonWrite _321 = PhysicalPointerUint2NonWrite(uvec2(_316._m0, SBT._m4.y + _316._m1));
-    float _327 = uintBitsToFloat(_321.value.x);
-    float _328 = uintBitsToFloat(_321.value.y);
-    AddCarry _339;
-    _339._m0 = uaddCarry(SBT._m4.x, _58 * 12u, _339._m1);
-    PhysicalPointerUint3NonWrite _344 = PhysicalPointerUint3NonWrite(uvec2(_339._m0, SBT._m4.y + _339._m1));
-    float _353 = uintBitsToFloat(_344.value.z);
-    AddCarry _363;
-    _363._m0 = uaddCarry(SBT._m4.x, _58 << 4u, _363._m1);
-    PhysicalPointerUint4NonWrite _368 = PhysicalPointerUint4NonWrite(uvec2(_363._m0, SBT._m4.y + _363._m1));
-    AddCarry _392;
-    _392._m0 = uaddCarry(SBT._m3.x, (_58 * 4u) + 0u, _392._m1);
-    PhysicalPointerFloat _397 = PhysicalPointerFloat(uvec2(_392._m0, SBT._m3.y + _392._m1));
-    AddCarry _411;
-    _411._m0 = uaddCarry(SBT._m5.x, _290, _411._m1);
-    float _419 = uintBitsToFloat(PhysicalPointerUint(uvec2(_411._m0, SBT._m5.y + _411._m1)).value);
-    float _420 = ((((((((((((((((_67.x + _80.x) + _99.x) + _119.x) + _32[nonuniformEXT(_53)]._m0[0u].x) + _32[nonuniformEXT(_146)]._m0[0u].x) + _169.x) + _182.x) + _203.value.x) + _232.x) + _258.x) + _279.value.x) + _306) + _327) + uintBitsToFloat(_344.value.x)) + uintBitsToFloat(_368.value.x)) + _397.value) + _419;
-    float _421 = ((((((((((((((((_67.y + _80.y) + _99.y) + _119.y) + _32[nonuniformEXT(_53)]._m0[0u].y) + _32[nonuniformEXT(_146)]._m0[0u].y) + _169.y) + _182.y) + _203.value.y) + _232.y) + _258.y) + _279.value.y) + _306) + _328) + uintBitsToFloat(_344.value.y)) + uintBitsToFloat(_368.value.y)) + _397.value) + _419;
-    vec4 _424 = _425;
-    _424.x = _420;
-    vec4 _426 = _424;
-    _426.y = _421;
-    vec4 _427 = _426;
-    _427.z = ((((((((((((((((_67.z + _80.z) + _99.z) + _119.z) + _32[nonuniformEXT(_53)]._m0[0u].z) + _32[nonuniformEXT(_146)]._m0[0u].z) + _169.z) + _182.z) + _203.value.z) + _232.z) + _258.z) + _279.value.z) + _306) + _327) + _353) + uintBitsToFloat(_368.value.z)) + _397.value) + _419;
-    vec4 _428 = _427;
-    _428.w = ((((((((((((((((_67.w + _80.w) + _99.w) + _119.w) + _32[nonuniformEXT(_53)]._m0[0u].w) + _32[nonuniformEXT(_146)]._m0[0u].w) + _169.w) + _182.w) + _203.value.w) + _232.w) + _258.w) + _279.value.w) + _306) + _328) + _353) + uintBitsToFloat(_368.value.w)) + _397.value) + _419;
-    payload._m0 = _428;
-    AddCarry _433;
-    _433._m0 = uaddCarry(SBT._m3.x, (_58 * 4u) + 0u, _433._m1);
-    PhysicalPointerFloat(uvec2(_433._m0, SBT._m3.y + _433._m1)).value = _420;
-    AddCarry _445;
-    _445._m0 = uaddCarry(SBT._m5.x, _290, _445._m1);
-    PhysicalPointerFloat(uvec2(_445._m0, SBT._m5.y + _445._m1)).value = _421;
+    PhysicalPointerFloat4NonWriteCBVArray _194 = PhysicalPointerFloat4NonWriteCBVArray(SBT._m6);
+    vec4 _223 = textureLod(nonuniformEXT(sampler2D(_21[registers._m0 + _59], _36[(SBT._m10.x >> 5u) + 13u])), vec2(0.5), 0.0);
+    vec4 _249 = textureLod(nonuniformEXT(sampler2D(_21[registers._m0 + _58], _36[((SBT._m10.x >> 5u) + 14u) + (_58 ^ 1u)])), vec2(0.5), 0.0);
+    PhysicalPointerFloat4NonWriteArray _264 = PhysicalPointerFloat4NonWriteArray(SBT._m2);
+    float _285 = uintBitsToFloat(PhysicalPointerUintNonWriteArray(SBT._m4).value[_58]);
+    PhysicalPointerUint2NonWriteArray _294 = PhysicalPointerUint2NonWriteArray(SBT._m4);
+    float _300 = uintBitsToFloat(_294.value[_58].x);
+    float _301 = uintBitsToFloat(_294.value[_58].y);
+    PhysicalPointerUint3NonWriteArray _311 = PhysicalPointerUint3NonWriteArray(SBT._m4);
+    float _320 = uintBitsToFloat(_311.value[_58].z);
+    PhysicalPointerUint4NonWriteArray _329 = PhysicalPointerUint4NonWriteArray(SBT._m4);
+    PhysicalPointerFloatArray _350 = PhysicalPointerFloatArray(SBT._m3);
+    float _366 = uintBitsToFloat(PhysicalPointerUintArray(SBT._m5).value[_58]);
+    float _367 = ((((((((((((((((_67.x + _80.x) + _99.x) + _119.x) + _32[nonuniformEXT(_53)]._m0[0u].x) + _32[nonuniformEXT(_146)]._m0[0u].x) + _169.x) + _182.x) + _194.value[1u].x) + _223.x) + _249.x) + _264.value[_58].x) + _285) + _300) + uintBitsToFloat(_311.value[_58].x)) + uintBitsToFloat(_329.value[_58].x)) + _350.value[_58]) + _366;
+    float _368 = ((((((((((((((((_67.y + _80.y) + _99.y) + _119.y) + _32[nonuniformEXT(_53)]._m0[0u].y) + _32[nonuniformEXT(_146)]._m0[0u].y) + _169.y) + _182.y) + _194.value[1u].y) + _223.y) + _249.y) + _264.value[_58].y) + _285) + _301) + uintBitsToFloat(_311.value[_58].y)) + uintBitsToFloat(_329.value[_58].y)) + _350.value[_58]) + _366;
+    vec4 _371;
+    _371.x = _367;
+    _371.y = _368;
+    _371.z = ((((((((((((((((_67.z + _80.z) + _99.z) + _119.z) + _32[nonuniformEXT(_53)]._m0[0u].z) + _32[nonuniformEXT(_146)]._m0[0u].z) + _169.z) + _182.z) + _194.value[1u].z) + _223.z) + _249.z) + _264.value[_58].z) + _285) + _300) + _320) + uintBitsToFloat(_329.value[_58].z)) + _350.value[_58]) + _366;
+    _371.w = ((((((((((((((((_67.w + _80.w) + _99.w) + _119.w) + _32[nonuniformEXT(_53)]._m0[0u].w) + _32[nonuniformEXT(_146)]._m0[0u].w) + _169.w) + _182.w) + _194.value[1u].w) + _223.w) + _249.w) + _264.value[_58].w) + _285) + _301) + _320) + uintBitsToFloat(_329.value[_58].w)) + _350.value[_58]) + _366;
+    payload._m0 = _371;
+    PhysicalPointerFloatArray(SBT._m3).value[_58] = _367;
+    PhysicalPointerFloatArray(SBT._m5).value[_58] = _368;
 }
 
 
@@ -165,7 +143,7 @@ void main()
 ; SPIR-V
 ; Version: 1.4
 ; Generator: Unknown(30017); 21022
-; Bound: 454
+; Bound: 385
 ; Schema: 0
 OpCapability Shader
 OpCapability UniformBufferArrayDynamicIndexing
@@ -192,21 +170,22 @@ OpName %16 "SBT"
 OpName %29 "BindlessCBV"
 OpName %37 ""
 OpName %39 "payload"
-OpName %195 "AddCarry"
-OpName %201 "PhysicalPointerFloat4NonWrite"
-OpMemberName %201 0 "value"
-OpName %293 "PhysicalPointerUintNonWrite"
-OpMemberName %293 0 "value"
-OpName %312 "PhysicalPointerUint2NonWrite"
-OpMemberName %312 0 "value"
-OpName %335 "PhysicalPointerUint3NonWrite"
-OpMemberName %335 0 "value"
-OpName %359 "PhysicalPointerUint4NonWrite"
-OpMemberName %359 0 "value"
-OpName %386 "PhysicalPointerFloat"
-OpMemberName %386 0 "value"
-OpName %407 "PhysicalPointerUint"
-OpMemberName %407 0 "value"
+OpName %192 "PhysicalPointerFloat4NonWriteCBVArray"
+OpMemberName %192 0 "value"
+OpName %262 "PhysicalPointerFloat4NonWriteArray"
+OpMemberName %262 0 "value"
+OpName %279 "PhysicalPointerUintNonWriteArray"
+OpMemberName %279 0 "value"
+OpName %292 "PhysicalPointerUint2NonWriteArray"
+OpMemberName %292 0 "value"
+OpName %309 "PhysicalPointerUint3NonWriteArray"
+OpMemberName %309 0 "value"
+OpName %327 "PhysicalPointerUint4NonWriteArray"
+OpMemberName %327 0 "value"
+OpName %348 "PhysicalPointerFloatArray"
+OpMemberName %348 0 "value"
+OpName %361 "PhysicalPointerUintArray"
+OpMemberName %361 0 "value"
 OpDecorate %6 Block
 OpMemberDecorate %6 0 Offset 0
 OpMemberDecorate %6 1 Offset 4
@@ -251,30 +230,41 @@ OpDecorate %130 NonUniform
 OpDecorate %146 NonUniform
 OpDecorate %140 NonUniform
 OpDecorate %147 NonUniform
-OpMemberDecorate %201 0 Offset 0
-OpDecorate %201 Block
-OpMemberDecorate %201 0 NonWritable
-OpDecorate %227 NonUniform
-OpDecorate %229 NonUniform
-OpDecorate %255 NonUniform
-OpDecorate %256 NonUniform
-OpDecorate %257 NonUniform
-OpMemberDecorate %293 0 Offset 0
-OpDecorate %293 Block
-OpMemberDecorate %293 0 NonWritable
-OpMemberDecorate %312 0 Offset 0
-OpDecorate %312 Block
-OpMemberDecorate %312 0 NonWritable
-OpMemberDecorate %335 0 Offset 0
-OpDecorate %335 Block
-OpMemberDecorate %335 0 NonWritable
-OpMemberDecorate %359 0 Offset 0
-OpDecorate %359 Block
-OpMemberDecorate %359 0 NonWritable
-OpMemberDecorate %386 0 Offset 0
-OpDecorate %386 Block
-OpMemberDecorate %407 0 Offset 0
-OpDecorate %407 Block
+OpDecorate %191 ArrayStride 16
+OpMemberDecorate %192 0 Offset 0
+OpDecorate %192 Block
+OpMemberDecorate %192 0 NonWritable
+OpDecorate %218 NonUniform
+OpDecorate %220 NonUniform
+OpDecorate %246 NonUniform
+OpDecorate %247 NonUniform
+OpDecorate %248 NonUniform
+OpDecorate %261 ArrayStride 16
+OpMemberDecorate %262 0 Offset 0
+OpDecorate %262 Block
+OpMemberDecorate %262 0 NonWritable
+OpDecorate %278 ArrayStride 4
+OpMemberDecorate %279 0 Offset 0
+OpDecorate %279 Block
+OpMemberDecorate %279 0 NonWritable
+OpDecorate %291 ArrayStride 8
+OpMemberDecorate %292 0 Offset 0
+OpDecorate %292 Block
+OpMemberDecorate %292 0 NonWritable
+OpDecorate %308 ArrayStride 12
+OpMemberDecorate %309 0 Offset 0
+OpDecorate %309 Block
+OpMemberDecorate %309 0 NonWritable
+OpDecorate %326 ArrayStride 16
+OpMemberDecorate %327 0 Offset 0
+OpDecorate %327 Block
+OpMemberDecorate %327 0 NonWritable
+OpDecorate %347 ArrayStride 4
+OpMemberDecorate %348 0 Offset 0
+OpDecorate %348 Block
+OpDecorate %360 ArrayStride 4
+OpMemberDecorate %361 0 Offset 0
+OpDecorate %361 Block
 %1 = OpTypeVoid
 %2 = OpTypeFunction %1
 %5 = OpTypeInt 32 0
@@ -335,42 +325,50 @@ OpDecorate %407 Block
 %165 = OpConstant %5 3
 %167 = OpTypeVector %5 4
 %179 = OpConstant %5 4
-%192 = OpConstant %5 16
-%195 = OpTypeStruct %5 %5
-%201 = OpTypeStruct %26
-%202 = OpTypePointer PhysicalStorageBuffer %201
-%204 = OpTypePointer PhysicalStorageBuffer %26
-%220 = OpTypePointer UniformConstant %33
-%223 = OpConstant %5 10
-%228 = OpTypeSampledImage %18
-%230 = OpConstant %17 0.5
-%231 = OpConstant %17 0
-%233 = OpTypeVector %17 2
-%254 = OpConstant %5 14
-%293 = OpTypeStruct %5
-%294 = OpTypePointer PhysicalStorageBuffer %293
-%303 = OpTypePointer PhysicalStorageBuffer %5
-%312 = OpTypeStruct %13
-%313 = OpTypePointer PhysicalStorageBuffer %312
-%322 = OpTypePointer PhysicalStorageBuffer %13
-%334 = OpTypeVector %5 3
-%335 = OpTypeStruct %334
-%336 = OpTypePointer PhysicalStorageBuffer %335
-%345 = OpTypePointer PhysicalStorageBuffer %334
-%359 = OpTypeStruct %167
-%360 = OpTypePointer PhysicalStorageBuffer %359
-%369 = OpTypePointer PhysicalStorageBuffer %167
-%386 = OpTypeStruct %17
-%387 = OpTypePointer PhysicalStorageBuffer %386
-%398 = OpTypePointer PhysicalStorageBuffer %17
-%407 = OpTypeStruct %5
-%408 = OpTypePointer PhysicalStorageBuffer %407
+%191 = OpTypeArray %26 %27
+%192 = OpTypeStruct %191
+%193 = OpTypePointer PhysicalStorageBuffer %192
+%195 = OpTypePointer PhysicalStorageBuffer %26
+%211 = OpTypePointer UniformConstant %33
+%214 = OpConstant %5 10
+%219 = OpTypeSampledImage %18
+%221 = OpConstant %17 0.5
+%222 = OpConstant %17 0
+%224 = OpTypeVector %17 2
+%245 = OpConstant %5 14
+%261 = OpTypeRuntimeArray %26
+%262 = OpTypeStruct %261
+%263 = OpTypePointer PhysicalStorageBuffer %262
+%278 = OpTypeRuntimeArray %5
+%279 = OpTypeStruct %278
+%280 = OpTypePointer PhysicalStorageBuffer %279
+%282 = OpTypePointer PhysicalStorageBuffer %5
+%291 = OpTypeRuntimeArray %13
+%292 = OpTypeStruct %291
+%293 = OpTypePointer PhysicalStorageBuffer %292
+%295 = OpTypePointer PhysicalStorageBuffer %13
+%307 = OpTypeVector %5 3
+%308 = OpTypeRuntimeArray %307
+%309 = OpTypeStruct %308
+%310 = OpTypePointer PhysicalStorageBuffer %309
+%312 = OpTypePointer PhysicalStorageBuffer %307
+%326 = OpTypeRuntimeArray %167
+%327 = OpTypeStruct %326
+%328 = OpTypePointer PhysicalStorageBuffer %327
+%330 = OpTypePointer PhysicalStorageBuffer %167
+%347 = OpTypeRuntimeArray %17
+%348 = OpTypeStruct %347
+%349 = OpTypePointer PhysicalStorageBuffer %348
+%351 = OpTypePointer PhysicalStorageBuffer %17
+%360 = OpTypeRuntimeArray %5
+%361 = OpTypeStruct %360
+%362 = OpTypePointer PhysicalStorageBuffer %361
 %3 = OpFunction %1 None %2
 %4 = OpLabel
-%425 = OpUndef %26
-%440 = OpUndef %17
-OpBranch %452
-%452 = OpLabel
+%372 = OpUndef %26
+%378 = OpUndef %17
+OpBranch %383
+%383 = OpLabel
 %41 = OpAccessChain %40 %16 %11
 %42 = OpLoad %13 %41
 %44 = OpAccessChain %43 %16 %45
@@ -499,238 +497,161 @@ OpBranch %452
 %188 = OpFAdd %17 %175 %184
 %189 = OpFAdd %17 %176 %185
 %190 = OpFAdd %17 %177 %186
-%191 = OpIMul %5 %57 %192
-%193 = OpCompositeExtract %5 %42 0
-%194 = OpCompositeExtract %5 %42 1
-%196 = OpIAddCarry %195 %193 %191
-%197 = OpCompositeExtract %5 %196 0
-%198 = OpCompositeExtract %5 %196 1
-%199 = OpIAdd %5 %194 %198
-%200 = OpCompositeConstruct %13 %197 %199
-%203 = OpBitcast %202 %200
-%205 = OpAccessChain %204 %203 %45
-%206 = OpLoad %26 %205 Aligned 16
-%207 = OpCompositeExtract %17 %206 0
-%208 = OpCompositeExtract %17 %206 1
-%209 = OpCompositeExtract %17 %206 2
-%210 = OpCompositeExtract %17 %206 3
-%211 = OpFAdd %17 %187 %207
-%212 = OpFAdd %17 %188 %208
-%213 = OpFAdd %17 %189 %209
-%214 = OpFAdd %17 %190 %210
-%216 = OpAccessChain %62 %8 %45
-%217 = OpLoad %5 %216
-%218 = OpIAdd %5 %217 %59
-%215 = OpAccessChain %60 %21 %218
-%219 = OpLoad %18 %215
-%222 = OpAccessChain %48 %16 %223 %45
-%224 = OpLoad %5 %222
-%225 = OpShiftRightLogical %5 %224 %9
-%226 = OpIAdd %5 %225 %145
-%221 = OpAccessChain %220 %36 %226
-%227 = OpLoad %33 %221
-%229 = OpSampledImage %228 %219 %227
-%234 = OpCompositeConstruct %233 %230 %230
-%232 = OpImageSampleExplicitLod %26 %229 %234 Lod %231
-%235 = OpCompositeExtract %17 %232 0
-%236 = OpCompositeExtract %17 %232 1
-%237 = OpCompositeExtract %17 %232 2
-%238 = OpCompositeExtract %17 %232 3
-%239 = OpFAdd %17 %211 %235
-%240 = OpFAdd %17 %212 %236
-%241 = OpFAdd %17 %213 %237
-%242 = OpFAdd %17 %214 %238
-%243 = OpBitwiseXor %5 %58 %57
-%245 = OpAccessChain %62 %8 %45
-%246 = OpLoad %5 %245
-%247 = OpIAdd %5 %246 %58
-%244 = OpAccessChain %60 %21 %247
-%248 = OpLoad %18 %244
-%250 = OpAccessChain %48 %16 %223 %45
-%251 = OpLoad %5 %250
-%252 = OpShiftRightLogical %5 %251 %9
-%253 = OpIAdd %5 %252 %254
-%255 = OpIAdd %5 %253 %243
-%249 = OpAccessChain %220 %36 %255
-%256 = OpLoad %33 %249
-%257 = OpSampledImage %228 %248 %256
-%259 = OpCompositeConstruct %233 %230 %230
-%258 = OpImageSampleExplicitLod %26 %257 %259 Lod %231
-%260 = OpCompositeExtract %17 %258 0
-%261 = OpCompositeExtract %17 %258 1
-%262 = OpCompositeExtract %17 %258 2
-%263 = OpCompositeExtract %17 %258 3
-%264 = OpFAdd %17 %239 %260
-%265 = OpFAdd %17 %240 %261
-%266 = OpFAdd %17 %241 %262
-%267 = OpFAdd %17 %242 %263
-%268 = OpAccessChain %40 %16 %162
-%269 = OpLoad %13 %268
-%270 = OpIMul %5 %58 %192
-%271 = OpIAdd %5 %270 %45
-%272 = OpCompositeExtract %5 %269 0
-%273 = OpCompositeExtract %5 %269 1
-%274 = OpIAddCarry %195 %272 %271
-%275 = OpCompositeExtract %5 %274 0
-%276 = OpCompositeExtract %5 %274 1
-%277 = OpIAdd %5 %273 %276
-%278 = OpCompositeConstruct %13 %275 %277
-%279 = OpBitcast %202 %278
-%280 = OpAccessChain %204 %279 %45
-%281 = OpLoad %26 %280 Aligned 4
-%282 = OpCompositeExtract %17 %281 0
-%283 = OpCompositeExtract %17 %281 1
-%284 = OpCompositeExtract %17 %281 2
-%285 = OpCompositeExtract %17 %281 3
-%286 = OpFAdd %17 %264 %282
-%287 = OpFAdd %17 %265 %283
-%288 = OpFAdd %17 %266 %284
-%289 = OpFAdd %17 %267 %285
-%290 = OpShiftLeftLogical %5 %58 %162
-%291 = OpAccessChain %40 %16 %179
-%292 = OpLoad %13 %291
-%295 = OpCompositeExtract %5 %292 0
-%296 = OpCompositeExtract %5 %292 1
-%297 = OpIAddCarry %195 %295 %290
+%194 = OpBitcast %193 %42
+%196 = OpInBoundsAccessChain %195 %194 %45 %57
+%197 = OpLoad %26 %196 Aligned 16
+%198 = OpCompositeExtract %17 %197 0
+%199 = OpCompositeExtract %17 %197 1
+%200 = OpCompositeExtract %17 %197 2
+%201 = OpCompositeExtract %17 %197 3
+%202 = OpFAdd %17 %187 %198
+%203 = OpFAdd %17 %188 %199
+%204 = OpFAdd %17 %189 %200
+%205 = OpFAdd %17 %190 %201
+%207 = OpAccessChain %62 %8 %45
+%208 = OpLoad %5 %207
+%209 = OpIAdd %5 %208 %59
+%206 = OpAccessChain %60 %21 %209
+%210 = OpLoad %18 %206
+%213 = OpAccessChain %48 %16 %214 %45
+%215 = OpLoad %5 %213
+%216 = OpShiftRightLogical %5 %215 %9
+%217 = OpIAdd %5 %216 %145
+%212 = OpAccessChain %211 %36 %217
+%218 = OpLoad %33 %212
+%220 = OpSampledImage %219 %210 %218
+%225 = OpCompositeConstruct %224 %221 %221
+%223 = OpImageSampleExplicitLod %26 %220 %225 Lod %222
+%226 = OpCompositeExtract %17 %223 0
+%227 = OpCompositeExtract %17 %223 1
+%228 = OpCompositeExtract %17 %223 2
+%229 = OpCompositeExtract %17 %223 3
+%230 = OpFAdd %17 %202 %226
+%231 = OpFAdd %17 %203 %227
+%232 = OpFAdd %17 %204 %228
+%233 = OpFAdd %17 %205 %229
+%234 = OpBitwiseXor %5 %58 %57
+%236 = OpAccessChain %62 %8 %45
+%237 = OpLoad %5 %236
+%238 = OpIAdd %5 %237 %58
+%235 = OpAccessChain %60 %21 %238
+%239 = OpLoad %18 %235
+%241 = OpAccessChain %48 %16 %214 %45
+%242 = OpLoad %5 %241
+%243 = OpShiftRightLogical %5 %242 %9
+%244 = OpIAdd %5 %243 %245
+%246 = OpIAdd %5 %244 %234
+%240 = OpAccessChain %211 %36 %246
+%247 = OpLoad %33 %240
+%248 = OpSampledImage %219 %239 %247
+%250 = OpCompositeConstruct %224 %221 %221
+%249 = OpImageSampleExplicitLod %26 %248 %250 Lod %222
+%251 = OpCompositeExtract %17 %249 0
+%252 = OpCompositeExtract %17 %249 1
+%253 = OpCompositeExtract %17 %249 2
+%254 = OpCompositeExtract %17 %249 3
+%255 = OpFAdd %17 %230 %251
+%256 = OpFAdd %17 %231 %252
+%257 = OpFAdd %17 %232 %253
+%258 = OpFAdd %17 %233 %254
+%259 = OpAccessChain %40 %16 %162
+%260 = OpLoad %13 %259
+%264 = OpBitcast %263 %260
+%265 = OpAccessChain %195 %264 %45 %58
+%266 = OpLoad %26 %265 Aligned 4
+%267 = OpCompositeExtract %17 %266 0
+%268 = OpCompositeExtract %17 %266 1
+%269 = OpCompositeExtract %17 %266 2
+%270 = OpCompositeExtract %17 %266 3
+%271 = OpFAdd %17 %255 %267
+%272 = OpFAdd %17 %256 %268
+%273 = OpFAdd %17 %257 %269
+%274 = OpFAdd %17 %258 %270
+%275 = OpShiftLeftLogical %5 %58 %162
+%276 = OpAccessChain %40 %16 %179
+%277 = OpLoad %13 %276
+%281 = OpBitcast %280 %277
+%283 = OpAccessChain %282 %281 %45 %58
+%284 = OpLoad %5 %283 Aligned 4
+%285 = OpBitcast %17 %284
+%286 = OpFAdd %17 %271 %285
+%287 = OpFAdd %17 %272 %285
+%288 = OpFAdd %17 %273 %285
+%289 = OpFAdd %17 %274 %285
+%290 = OpShiftLeftLogical %5 %58 %165
+%294 = OpBitcast %293 %277
+%296 = OpAccessChain %295 %294 %45 %58
+%297 = OpLoad %13 %296 Aligned 4
 %298 = OpCompositeExtract %5 %297 0
 %299 = OpCompositeExtract %5 %297 1
-%300 = OpIAdd %5 %296 %299
-%301 = OpCompositeConstruct %13 %298 %300
-%302 = OpBitcast %294 %301
-%304 = OpAccessChain %303 %302 %45
-%305 = OpLoad %5 %304 Aligned 4
-%306 = OpBitcast %17 %305
-%307 = OpFAdd %17 %286 %306
-%308 = OpFAdd %17 %287 %306
-%309 = OpFAdd %17 %288 %306
-%310 = OpFAdd %17 %289 %306
-%311 = OpShiftLeftLogical %5 %58 %165
-%314 = OpCompositeExtract %5 %292 0
-%315 = OpCompositeExtract %5 %292 1
-%316 = OpIAddCarry %195 %314 %311
-%317 = OpCompositeExtract %5 %316 0
-%318 = OpCompositeExtract %5 %316 1
-%319 = OpIAdd %5 %315 %318
-%320 = OpCompositeConstruct %13 %317 %319
-%321 = OpBitcast %313 %320
-%323 = OpAccessChain %322 %321 %45
-%324 = OpLoad %13 %323 Aligned 4
-%325 = OpCompositeExtract %5 %324 0
-%326 = OpCompositeExtract %5 %324 1
-%327 = OpBitcast %17 %325
-%328 = OpBitcast %17 %326
-%329 = OpFAdd %17 %307 %327
-%330 = OpFAdd %17 %308 %328
-%331 = OpFAdd %17 %309 %327
-%332 = OpFAdd %17 %310 %328
-%333 = OpIMul %5 %58 %54
-%337 = OpCompositeExtract %5 %292 0
-%338 = OpCompositeExtract %5 %292 1
-%339 = OpIAddCarry %195 %337 %333
-%340 = OpCompositeExtract %5 %339 0
-%341 = OpCompositeExtract %5 %339 1
-%342 = OpIAdd %5 %338 %341
-%343 = OpCompositeConstruct %13 %340 %342
-%344 = OpBitcast %336 %343
-%346 = OpAccessChain %345 %344 %45
-%347 = OpLoad %334 %346 Aligned 4
-%348 = OpCompositeExtract %5 %347 0
-%349 = OpCompositeExtract %5 %347 1
-%350 = OpCompositeExtract %5 %347 2
-%351 = OpBitcast %17 %348
-%352 = OpBitcast %17 %349
-%353 = OpBitcast %17 %350
-%354 = OpFAdd %17 %329 %351
-%355 = OpFAdd %17 %330 %352
-%356 = OpFAdd %17 %331 %353
-%357 = OpFAdd %17 %332 %353
-%358 = OpShiftLeftLogical %5 %58 %179
-%361 = OpCompositeExtract %5 %292 0
-%362 = OpCompositeExtract %5 %292 1
-%363 = OpIAddCarry %195 %361 %358
-%364 = OpCompositeExtract %5 %363 0
-%365 = OpCompositeExtract %5 %363 1
-%366 = OpIAdd %5 %362 %365
-%367 = OpCompositeConstruct %13 %364 %366
-%368 = OpBitcast %360 %367
-%370 = OpAccessChain %369 %368 %45
-%371 = OpLoad %167 %370 Aligned 4
-%372 = OpCompositeExtract %5 %371 0
-%373 = OpCompositeExtract %5 %371 1
-%374 = OpCompositeExtract %5 %371 2
-%375 = OpCompositeExtract %5 %371 3
-%376 = OpBitcast %17 %372
-%377 = OpBitcast %17 %373
-%378 = OpBitcast %17 %374
-%379 = OpBitcast %17 %375
-%380 = OpFAdd %17 %354 %376
-%381 = OpFAdd %17 %355 %377
-%382 = OpFAdd %17 %356 %378
-%383 = OpFAdd %17 %357 %379
-%384 = OpAccessChain %40 %16 %165
-%385 = OpLoad %13 %384
-%388 = OpIMul %5 %58 %179
-%389 = OpIAdd %5 %388 %45
-%390 = OpCompositeExtract %5 %385 0
-%391 = OpCompositeExtract %5 %385 1
-%392 = OpIAddCarry %195 %390 %389
-%393 = OpCompositeExtract %5 %392 0
-%394 = OpCompositeExtract %5 %392 1
-%395 = OpIAdd %5 %391 %394
-%396 = OpCompositeConstruct %13 %393 %395
-%397 = OpBitcast %387 %396
-%399 = OpAccessChain %398 %397 %45
-%400 = OpLoad %17 %399 Aligned 4
-%401 = OpFAdd %17 %380 %400
-%402 = OpFAdd %17 %381 %400
-%403 = OpFAdd %17 %382 %400
-%404 = OpFAdd %17 %383 %400
-%405 = OpAccessChain %40 %16 %9
-%406 = OpLoad %13 %405
-%409 = OpCompositeExtract %5 %406 0
-%410 = OpCompositeExtract %5 %406 1
-%411 = OpIAddCarry %195 %409 %290
-%412 = OpCompositeExtract %5 %411 0
-%413 = OpCompositeExtract %5 %411 1
-%414 = OpIAdd %5 %410 %413
-%415 = OpCompositeConstruct %13 %412 %414
-%416 = OpBitcast %408 %415
-%417 = OpAccessChain %303 %416 %45
-%418 = OpLoad %5 %417 Aligned 4
-%419 = OpBitcast %17 %418
-%420 = OpFAdd %17 %401 %419
-%421 = OpFAdd %17 %402 %419
-%422 = OpFAdd %17 %403 %419
-%423 = OpFAdd %17 %404 %419
-%424 = OpCompositeInsert %26 %420 %425 0
-%426 = OpCompositeInsert %26 %421 %424 1
-%427 = OpCompositeInsert %26 %422 %426 2
-%428 = OpCompositeInsert %26 %423 %427 3
-OpStore %74 %428
-%429 = OpIMul %5 %58 %179
-%430 = OpIAdd %5 %429 %45
-%431 = OpCompositeExtract %5 %385 0
-%432 = OpCompositeExtract %5 %385 1
-%433 = OpIAddCarry %195 %431 %430
-%434 = OpCompositeExtract %5 %433 0
-%435 = OpCompositeExtract %5 %433 1
-%436 = OpIAdd %5 %432 %435
-%437 = OpCompositeConstruct %13 %434 %436
-%438 = OpBitcast %387 %437
-%439 = OpAccessChain %398 %438 %45
-OpStore %439 %420 Aligned 4
-%441 = OpAccessChain %40 %16 %9
-%442 = OpLoad %13 %441
-%443 = OpCompositeExtract %5 %442 0
-%444 = OpCompositeExtract %5 %442 1
-%445 = OpIAddCarry %195 %443 %290
-%446 = OpCompositeExtract %5 %445 0
-%447 = OpCompositeExtract %5 %445 1
-%448 = OpIAdd %5 %444 %447
-%449 = OpCompositeConstruct %13 %446 %448
-%450 = OpBitcast %387 %449
-%451 = OpAccessChain %398 %450 %45
-OpStore %451 %421 Aligned 4
+%300 = OpBitcast %17 %298
+%301 = OpBitcast %17 %299
+%302 = OpFAdd %17 %286 %300
+%303 = OpFAdd %17 %287 %301
+%304 = OpFAdd %17 %288 %300
+%305 = OpFAdd %17 %289 %301
+%306 = OpIMul %5 %58 %54
+%311 = OpBitcast %310 %277
+%313 = OpAccessChain %312 %311 %45 %58
+%314 = OpLoad %307 %313 Aligned 4
+%315 = OpCompositeExtract %5 %314 0
+%316 = OpCompositeExtract %5 %314 1
+%317 = OpCompositeExtract %5 %314 2
+%318 = OpBitcast %17 %315
+%319 = OpBitcast %17 %316
+%320 = OpBitcast %17 %317
+%321 = OpFAdd %17 %302 %318
+%322 = OpFAdd %17 %303 %319
+%323 = OpFAdd %17 %304 %320
+%324 = OpFAdd %17 %305 %320
+%325 = OpShiftLeftLogical %5 %58 %179
+%329 = OpBitcast %328 %277
+%331 = OpAccessChain %330 %329 %45 %58
+%332 = OpLoad %167 %331 Aligned 4
+%333 = OpCompositeExtract %5 %332 0
+%334 = OpCompositeExtract %5 %332 1
+%335 = OpCompositeExtract %5 %332 2
+%336 = OpCompositeExtract %5 %332 3
+%337 = OpBitcast %17 %333
+%338 = OpBitcast %17 %334
+%339 = OpBitcast %17 %335
+%340 = OpBitcast %17 %336
+%341 = OpFAdd %17 %321 %337
+%342 = OpFAdd %17 %322 %338
+%343 = OpFAdd %17 %323 %339
+%344 = OpFAdd %17 %324 %340
+%345 = OpAccessChain %40 %16 %165
+%346 = OpLoad %13 %345
+%350 = OpBitcast %349 %346
+%352 = OpAccessChain %351 %350 %45 %58
+%353 = OpLoad %17 %352 Aligned 4
+%354 = OpFAdd %17 %341 %353
+%355 = OpFAdd %17 %342 %353
+%356 = OpFAdd %17 %343 %353
+%357 = OpFAdd %17 %344 %353
+%358 = OpAccessChain %40 %16 %9
+%359 = OpLoad %13 %358
+%363 = OpBitcast %362 %359
+%364 = OpAccessChain %282 %363 %45 %58
+%365 = OpLoad %5 %364 Aligned 4
+%366 = OpBitcast %17 %365
+%367 = OpFAdd %17 %354 %366
+%368 = OpFAdd %17 %355 %366
+%369 = OpFAdd %17 %356 %366
+%370 = OpFAdd %17 %357 %366
+%371 = OpCompositeInsert %26 %367 %372 0
+%373 = OpCompositeInsert %26 %368 %371 1
+%374 = OpCompositeInsert %26 %369 %373 2
+%375 = OpCompositeInsert %26 %370 %374 3
+OpStore %74 %375
+%376 = OpBitcast %349 %346
+%377 = OpAccessChain %351 %376 %45 %58
+OpStore %377 %367 Aligned 4
+%379 = OpAccessChain %40 %16 %9
+%380 = OpLoad %13 %379
+%381 = OpBitcast %349 %380
+%382 = OpAccessChain %351 %381 %45 %58
+OpStore %382 %368 Aligned 4
 OpReturn
 OpFunctionEnd
 #endif
diff --git a/reference/shaders/resources/rt-resources.bindless.rmiss b/reference/shaders/resources/rt-resources.bindless.rmiss
index 845746d..10f0d7c 100644
--- a/reference/shaders/resources/rt-resources.bindless.rmiss
+++ b/reference/shaders/resources/rt-resources.bindless.rmiss
@@ -32,15 +32,12 @@ void main()
     vec4 _31 = texelFetch(_13[registers._m0 + (payload._m1 & 1u)], ivec2(uvec2(0u)), int(0u));
     vec4 _45 = texelFetch(_13[registers._m0 + payload._m1], ivec2(uvec2(0u)), int(0u));
     vec4 _62 = texelFetch(_13[(registers._m0 + 10u) + payload._m1], ivec2(uvec2(0u)), int(0u));
-    vec4 _72 = _73;
+    vec4 _72;
     _72.x = (_31.x + _45.x) + _62.x;
-    vec4 _74 = _72;
-    _74.y = (_31.y + _45.y) + _62.y;
-    vec4 _75 = _74;
-    _75.z = (_31.z + _45.z) + _62.z;
-    vec4 _76 = _75;
-    _76.w = (_31.w + _45.w) + _62.w;
-    payload._m0 = _76;
+    _72.y = (_31.y + _45.y) + _62.y;
+    _72.z = (_31.z + _45.z) + _62.z;
+    _72.w = (_31.w + _45.w) + _62.w;
+    payload._m0 = _72;
 }
 
 
diff --git a/reference/shaders/resources/rt-resources.rmiss b/reference/shaders/resources/rt-resources.rmiss
index 7dcb965..81dfb27 100644
--- a/reference/shaders/resources/rt-resources.rmiss
+++ b/reference/shaders/resources/rt-resources.rmiss
@@ -19,15 +19,12 @@ void main()
 {
     vec4 _28 = texelFetch(Tex[payload._m1 & 1u], ivec2(uvec2(0u)), int(0u));
     vec4 _39 = texelFetch(TexUnsized[payload._m1], ivec2(uvec2(0u)), int(0u));
-    vec4 _49 = _50;
+    vec4 _49;
     _49.x = _28.x + _39.x;
-    vec4 _51 = _49;
-    _51.y = _28.y + _39.y;
-    vec4 _52 = _51;
-    _52.z = _28.z + _39.z;
-    vec4 _53 = _52;
-    _53.w = _28.w + _39.w;
-    payload._m0 = _53;
+    _49.y = _28.y + _39.y;
+    _49.z = _28.z + _39.z;
+    _49.w = _28.w + _39.w;
+    payload._m0 = _49;
 }
 
 
diff --git a/reference/shaders/resources/sm66/atomics-64bit.root-descriptor.sm66.comp b/reference/shaders/resources/sm66/atomics-64bit.root-descriptor.sm66.comp
index a4660a7..6a97050 100644
--- a/reference/shaders/resources/sm66/atomics-64bit.root-descriptor.sm66.comp
+++ b/reference/shaders/resources/sm66/atomics-64bit.root-descriptor.sm66.comp
@@ -1,10 +1,11 @@
 #version 460
 #extension GL_ARB_gpu_shader_int64 : require
 #extension GL_EXT_buffer_reference : require
+#extension GL_EXT_buffer_reference_uvec2 : require
 layout(local_size_x = 128, local_size_y = 1, local_size_z = 1) in;
 
 layout(buffer_reference) buffer PhysicalPointerUint64Array;
-layout(buffer_reference, std430) buffer PhysicalPointerUint64Array
+layout(buffer_reference, buffer_reference_align = 8, std430) buffer PhysicalPointerUint64Array
 {
     uint64_t value[];
 };
diff --git a/reference/shaders/resources/sm66/binding-range-selection.bindless.sm66.comp b/reference/shaders/resources/sm66/binding-range-selection.bindless.sm66.comp
index 4d353ba..7cd4ed4 100644
--- a/reference/shaders/resources/sm66/binding-range-selection.bindless.sm66.comp
+++ b/reference/shaders/resources/sm66/binding-range-selection.bindless.sm66.comp
@@ -4,7 +4,7 @@
 layout(local_size_x = 1, local_size_y = 1, local_size_z = 1) in;
 
 layout(buffer_reference) buffer AtomicCounter;
-layout(buffer_reference, std430) buffer AtomicCounter
+layout(buffer_reference, buffer_reference_align = 4, std430) buffer AtomicCounter
 {
     uint _m0;
 };
diff --git a/reference/shaders/resources/sm66/structured-buffer-heap.sm66.frag b/reference/shaders/resources/sm66/structured-buffer-heap.sm66.frag
index b5b2fa7..550183a 100644
--- a/reference/shaders/resources/sm66/structured-buffer-heap.sm66.frag
+++ b/reference/shaders/resources/sm66/structured-buffer-heap.sm66.frag
@@ -34,14 +34,9 @@ void main()
     vec4 _144 = uintBitsToFloat(uvec4(texelFetch(_9[_60], int(_131)).x, texelFetch(_9[_60], int(_131 + 1u)).x, texelFetch(_9[_60], int(_131 + 2u)).x, texelFetch(_9[_60], int(_131 + 3u)).x));
     uvec4 _152 = imageLoad(_13[INDEX + 4u], int(_37));
     uint _156 = _41 * 2u;
-    uvec4 _157 = imageLoad(_16[_68], int(_156));
-    uvec4 _159 = imageLoad(_16[_68], int(_156 + 1u));
-    vec2 _163 = uintBitsToFloat(uvec2(_157.x, _159.x));
+    vec2 _163 = uintBitsToFloat(uvec2(imageLoad(_16[_68], int(_156)).x, imageLoad(_16[_68], int(_156 + 1u)).x));
     uint _168 = _45 * 3u;
-    uvec4 _169 = imageLoad(_13[_72], int(_168));
-    uvec4 _171 = imageLoad(_13[_72], int(_168 + 1u));
-    uvec4 _174 = imageLoad(_13[_72], int(_168 + 2u));
-    vec3 _178 = uintBitsToFloat(uvec3(_169.x, _171.x, _174.x));
+    vec3 _178 = uintBitsToFloat(uvec3(imageLoad(_13[_72], int(_168)).x, imageLoad(_13[_72], int(_168 + 1u)).x, imageLoad(_13[_72], int(_168 + 2u)).x));
     uint _185 = _49 * 4u;
     uvec4 _186 = imageLoad(_13[_76], int(_185));
     uvec4 _188 = imageLoad(_13[_76], int(_185 + 1u));
diff --git a/reference/shaders/resources/sm66/structured-buffer-heap.ssbo.sm66.frag b/reference/shaders/resources/sm66/structured-buffer-heap.ssbo.sm66.frag
index 6ec773e..4fdcdf0 100644
--- a/reference/shaders/resources/sm66/structured-buffer-heap.ssbo.sm66.frag
+++ b/reference/shaders/resources/sm66/structured-buffer-heap.ssbo.sm66.frag
@@ -78,8 +78,7 @@ void main()
     vec3 _156 = uintBitsToFloat(_22[INDEX + 2u]._m0[_91]);
     vec4 _165 = uintBitsToFloat(_28[INDEX + 3u]._m0[_95]);
     uint _174 = _33[INDEX + 4u]._m0[_83];
-    uvec2 _178 = _38[_112]._m0[_87];
-    vec2 _179 = uintBitsToFloat(_178);
+    vec2 _179 = uintBitsToFloat(_38[_112]._m0[_87]);
     uvec3 _185 = _43[INDEX + 6u]._m0[_91];
     vec3 _186 = uintBitsToFloat(_185);
     uvec4 _194 = _48[INDEX + 7u]._m0[_95];
diff --git a/reference/shaders/resources/sm66/structured-buffer-heap.ssbo.ssbo-align.sm66.frag b/reference/shaders/resources/sm66/structured-buffer-heap.ssbo.ssbo-align.sm66.frag
index c921f2b..992a3be 100644
--- a/reference/shaders/resources/sm66/structured-buffer-heap.ssbo.ssbo-align.sm66.frag
+++ b/reference/shaders/resources/sm66/structured-buffer-heap.ssbo.ssbo-align.sm66.frag
@@ -96,14 +96,10 @@ void main()
     vec3 _224 = uintBitsToFloat(uvec3(_15[_106]._m0[_212], _15[_106]._m0[_212 + 1u], _15[_106]._m0[_212 + 2u]));
     vec4 _239 = uintBitsToFloat(_26[_112]._m0[(_88 < _118.y) ? (_88 + _118.x) : 268435455u]);
     uint _253 = _31[_121]._m0[(_76 < _127.y) ? (_76 + _127.x) : 1073741820u];
-    uvec2 _262 = _36[_128]._m0[(_80 < _135.y) ? (_80 + _135.x) : 536870911u];
-    vec2 _263 = uintBitsToFloat(_262);
+    vec2 _263 = uintBitsToFloat(_36[_128]._m0[(_80 < _135.y) ? (_80 + _135.x) : 536870911u]);
     uint _268 = _84 * 3u;
     uint _273 = (_268 < _142.y) ? (_268 + _142.x) : 1073741820u;
-    uint _275 = _31[_136]._m0[_273];
-    uint _278 = _31[_136]._m0[_273 + 1u];
-    uint _281 = _31[_136]._m0[_273 + 2u];
-    vec3 _283 = uintBitsToFloat(uvec3(_275, _278, _281));
+    vec3 _283 = uintBitsToFloat(uvec3(_31[_136]._m0[_273], _31[_136]._m0[_273 + 1u], _31[_136]._m0[_273 + 2u]));
     uvec4 _296 = _41[_143]._m0[(_88 < _150.y) ? (_88 + _150.x) : 268435455u];
     vec4 _297 = uintBitsToFloat(_296);
     uint _312 = _46[_151]._m0[(_76 < _158.y) ? (_76 + _158.x) : 1073741820u];
diff --git a/reference/shaders/resources/sm66/structured-buffer-heap.typed-buffer-offset.sm66.frag b/reference/shaders/resources/sm66/structured-buffer-heap.typed-buffer-offset.sm66.frag
index d4cb4a8..2a62d9d 100644
--- a/reference/shaders/resources/sm66/structured-buffer-heap.typed-buffer-offset.sm66.frag
+++ b/reference/shaders/resources/sm66/structured-buffer-heap.typed-buffer-offset.sm66.frag
@@ -71,9 +71,7 @@ void main()
     uvec4 _223 = imageLoad(_21[_84], int((_45 < _13._m0[_89].y) ? (_45 + _13._m0[_89].x) : 1073741820u));
     uint _227 = _49 * 2u;
     uint _232 = (_227 < _13._m0[_96].y) ? (_227 + _13._m0[_96].x) : 1073741820u;
-    uvec4 _233 = imageLoad(_24[_92], int(_232));
-    uvec4 _235 = imageLoad(_24[_92], int(_232 + 1u));
-    vec2 _239 = uintBitsToFloat(uvec2(_233.x, _235.x));
+    vec2 _239 = uintBitsToFloat(uvec2(imageLoad(_24[_92], int(_232)).x, imageLoad(_24[_92], int(_232 + 1u)).x));
     uint _244 = _53 * 3u;
     uint _249 = (_244 < _13._m0[_103].y) ? (_244 + _13._m0[_103].x) : 1073741820u;
     uvec4 _250 = imageLoad(_21[_99], int(_249));
diff --git a/reference/shaders/resources/ssbo-minprecision.sm60.ssbo.native-fp16.root-descriptor.frag b/reference/shaders/resources/ssbo-minprecision.sm60.ssbo.native-fp16.root-descriptor.frag
index c886c51..53e1ea3 100644
--- a/reference/shaders/resources/ssbo-minprecision.sm60.ssbo.native-fp16.root-descriptor.frag
+++ b/reference/shaders/resources/ssbo-minprecision.sm60.ssbo.native-fp16.root-descriptor.frag
@@ -9,29 +9,24 @@
 #error No extension available for FP16.
 #endif
 #extension GL_EXT_buffer_reference : require
+#extension GL_EXT_buffer_reference_uvec2 : require
 
-struct AddCarry
+layout(buffer_reference) buffer PhysicalPointerFloatNonWriteArray;
+layout(buffer_reference) buffer PhysicalPointerFloatArray;
+layout(buffer_reference) buffer PhysicalPointerUintArray;
+layout(buffer_reference, buffer_reference_align = 4, std430) readonly buffer PhysicalPointerFloatNonWriteArray
 {
-    uint _m0;
-    uint _m1;
+    float value[];
 };
 
-layout(buffer_reference) buffer PhysicalPointerFloatNonWrite;
-layout(buffer_reference) buffer PhysicalPointerFloat;
-layout(buffer_reference) buffer PhysicalPointerUint;
-layout(buffer_reference, std430) readonly buffer PhysicalPointerFloatNonWrite
+layout(buffer_reference, buffer_reference_align = 4, std430) buffer PhysicalPointerFloatArray
 {
-    float value;
+    float value[];
 };
 
-layout(buffer_reference, std430) buffer PhysicalPointerFloat
+layout(buffer_reference, buffer_reference_align = 4, std430) buffer PhysicalPointerUintArray
 {
-    float value;
-};
-
-layout(buffer_reference, std430) buffer PhysicalPointerUint
-{
-    uint value;
+    uint value[];
 };
 
 layout(set = 0, binding = 1, std430) restrict readonly buffer SSBO
@@ -50,24 +45,16 @@ layout(push_constant, std430) uniform RootConstants
 layout(location = 0) flat in mediump int A;
 layout(location = 0) out int SV_Target;
 
-float16_t _84;
-uint16_t _108;
+float16_t _57;
+uint16_t _73;
 
 void main()
 {
     uint16_t _31 = uint16_t(A);
     uint _32 = uint(int16_t(_31));
-    AddCarry _43;
-    _43._m0 = uaddCarry(registers._m1.x, (_32 * 4u) + 0u, _43._m1);
-    uint _56 = uint(int16_t(_31 + 1us));
-    AddCarry _61;
-    _61._m0 = uaddCarry(registers._m1.x, (_56 * 4u) + 0u, _61._m1);
-    AddCarry _77;
-    _77._m0 = uaddCarry(registers._m2.x, (_32 * 4u) + 0u, _77._m1);
-    PhysicalPointerFloat(uvec2(_77._m0, registers._m2.y + _77._m1)).value = float(float16_t(PhysicalPointerFloatNonWrite(uvec2(_61._m0, registers._m1.y + _61._m1)).value) + float16_t(PhysicalPointerFloatNonWrite(uvec2(_43._m0, registers._m1.y + _43._m1)).value));
-    AddCarry _100;
-    _100._m0 = uaddCarry(registers._m3.x, (_32 * 4u) + 0u, _100._m1);
-    PhysicalPointerUint(uvec2(_100._m0, registers._m3.y + _100._m1)).value = uint(uint16_t(_13._m0[_56]) + uint16_t(_13._m0[_32]));
+    uint _46 = uint(int16_t(_31 + 1us));
+    PhysicalPointerFloatArray(registers._m2).value[_32] = float(float16_t(PhysicalPointerFloatNonWriteArray(registers._m1).value[_46]) + float16_t(PhysicalPointerFloatNonWriteArray(registers._m1).value[_32]));
+    PhysicalPointerUintArray(registers._m3).value[_32] = uint(uint16_t(_13._m0[_46]) + uint16_t(_13._m0[_32]));
     SV_Target = int(10u);
 }
 
@@ -77,7 +64,7 @@ void main()
 ; SPIR-V
 ; Version: 1.3
 ; Generator: Unknown(30017); 21022
-; Bound: 114
+; Bound: 79
 ; Schema: 0
 OpCapability Shader
 OpCapability Float16
@@ -93,13 +80,12 @@ OpName %9 "registers"
 OpName %11 "SSBO"
 OpName %16 "A"
 OpName %18 "SV_Target"
-OpName %34 "PhysicalPointerFloatNonWrite"
-OpMemberName %34 0 "value"
-OpName %42 "AddCarry"
-OpName %71 "PhysicalPointerFloat"
-OpMemberName %71 0 "value"
-OpName %94 "PhysicalPointerUint"
-OpMemberName %94 0 "value"
+OpName %35 "PhysicalPointerFloatNonWriteArray"
+OpMemberName %35 0 "value"
+OpName %53 "PhysicalPointerFloatArray"
+OpMemberName %53 0 "value"
+OpName %68 "PhysicalPointerUintArray"
+OpMemberName %68 0 "value"
 OpDecorate %7 Block
 OpMemberDecorate %7 0 Offset 0
 OpMemberDecorate %7 1 Offset 8
@@ -116,13 +102,16 @@ OpDecorate %16 RelaxedPrecision
 OpDecorate %16 Flat
 OpDecorate %16 Location 0
 OpDecorate %18 Location 0
-OpMemberDecorate %34 0 Offset 0
-OpDecorate %34 Block
-OpMemberDecorate %34 0 NonWritable
-OpMemberDecorate %71 0 Offset 0
-OpDecorate %71 Block
-OpMemberDecorate %94 0 Offset 0
-OpDecorate %94 Block
+OpDecorate %34 ArrayStride 4
+OpMemberDecorate %35 0 Offset 0
+OpDecorate %35 Block
+OpMemberDecorate %35 0 NonWritable
+OpDecorate %52 ArrayStride 4
+OpMemberDecorate %53 0 Offset 0
+OpDecorate %53 Block
+OpDecorate %67 ArrayStride 4
+OpMemberDecorate %68 0 Offset 0
+OpDecorate %68 Block
 %1 = OpTypeVoid
 %2 = OpTypeFunction %1
 %5 = OpTypeInt 32 0
@@ -145,27 +134,28 @@ OpDecorate %94 Block
 %27 = OpConstant %5 1
 %30 = OpTypeInt 16 0
 %33 = OpTypeFloat 32
-%34 = OpTypeStruct %33
-%35 = OpTypePointer PhysicalStorageBuffer %34
-%36 = OpConstant %5 0
-%38 = OpConstant %5 4
-%42 = OpTypeStruct %5 %5
-%49 = OpTypePointer PhysicalStorageBuffer %33
-%52 = OpTypeFloat 16
-%55 = OpConstant %30 1
-%71 = OpTypeStruct %33
-%72 = OpTypePointer PhysicalStorageBuffer %71
-%86 = OpTypePointer StorageBuffer %5
-%94 = OpTypeStruct %5
-%95 = OpTypePointer PhysicalStorageBuffer %94
-%106 = OpTypePointer PhysicalStorageBuffer %5
-%110 = OpConstant %5 10
+%34 = OpTypeRuntimeArray %33
+%35 = OpTypeStruct %34
+%36 = OpTypePointer PhysicalStorageBuffer %35
+%38 = OpTypePointer PhysicalStorageBuffer %33
+%40 = OpConstant %5 0
+%42 = OpTypeFloat 16
+%45 = OpConstant %30 1
+%52 = OpTypeRuntimeArray %33
+%53 = OpTypeStruct %52
+%54 = OpTypePointer PhysicalStorageBuffer %53
+%59 = OpTypePointer StorageBuffer %5
+%67 = OpTypeRuntimeArray %5
+%68 = OpTypeStruct %67
+%69 = OpTypePointer PhysicalStorageBuffer %68
+%71 = OpTypePointer PhysicalStorageBuffer %5
+%75 = OpConstant %5 10
 %3 = OpFunction %1 None %2
 %4 = OpLabel
-%84 = OpUndef %52
-%108 = OpUndef %30
-OpBranch %112
-%112 = OpLabel
+%57 = OpUndef %42
+%73 = OpUndef %30
+OpBranch %77
+%77 = OpLabel
 %20 = OpAccessChain %19 %9 %21
 %22 = OpLoad %6 %20
 %23 = OpAccessChain %19 %9 %24
@@ -175,70 +165,34 @@ OpBranch %112
 %29 = OpLoad %14 %16
 %31 = OpSConvert %30 %29
 %32 = OpSConvert %5 %31
-%37 = OpIMul %5 %32 %38
-%39 = OpIAdd %5 %37 %36
-%40 = OpCompositeExtract %5 %28 0
-%41 = OpCompositeExtract %5 %28 1
-%43 = OpIAddCarry %42 %40 %39
-%44 = OpCompositeExtract %5 %43 0
-%45 = OpCompositeExtract %5 %43 1
-%46 = OpIAdd %5 %41 %45
-%47 = OpCompositeConstruct %6 %44 %46
-%48 = OpBitcast %35 %47
-%50 = OpAccessChain %49 %48 %36
-%51 = OpLoad %33 %50 Aligned 4
-%53 = OpFConvert %52 %51
-%54 = OpIAdd %30 %31 %55
-%56 = OpSConvert %5 %54
-%57 = OpIMul %5 %56 %38
-%58 = OpIAdd %5 %57 %36
-%59 = OpCompositeExtract %5 %28 0
-%60 = OpCompositeExtract %5 %28 1
-%61 = OpIAddCarry %42 %59 %58
-%62 = OpCompositeExtract %5 %61 0
-%63 = OpCompositeExtract %5 %61 1
-%64 = OpIAdd %5 %60 %63
-%65 = OpCompositeConstruct %6 %62 %64
-%66 = OpBitcast %35 %65
-%67 = OpAccessChain %49 %66 %36
-%68 = OpLoad %33 %67 Aligned 4
-%69 = OpFConvert %52 %68
-%70 = OpFAdd %52 %69 %53
-%73 = OpIMul %5 %32 %38
-%74 = OpIAdd %5 %73 %36
-%75 = OpCompositeExtract %5 %25 0
-%76 = OpCompositeExtract %5 %25 1
-%77 = OpIAddCarry %42 %75 %74
-%78 = OpCompositeExtract %5 %77 0
-%79 = OpCompositeExtract %5 %77 1
-%80 = OpIAdd %5 %76 %79
-%81 = OpCompositeConstruct %6 %78 %80
-%82 = OpBitcast %72 %81
-%83 = OpAccessChain %49 %82 %36
-%85 = OpFConvert %33 %70
-OpStore %83 %85 Aligned 4
-%87 = OpAccessChain %86 %13 %36 %32
-%88 = OpLoad %5 %87
-%89 = OpUConvert %30 %88
-%90 = OpAccessChain %86 %13 %36 %56
-%91 = OpLoad %5 %90
-%92 = OpUConvert %30 %91
-%93 = OpIAdd %30 %92 %89
-%96 = OpIMul %5 %32 %38
-%97 = OpIAdd %5 %96 %36
-%98 = OpCompositeExtract %5 %22 0
-%99 = OpCompositeExtract %5 %22 1
-%100 = OpIAddCarry %42 %98 %97
-%101 = OpCompositeExtract %5 %100 0
-%102 = OpCompositeExtract %5 %100 1
-%103 = OpIAdd %5 %99 %102
-%104 = OpCompositeConstruct %6 %101 %103
-%105 = OpBitcast %95 %104
-%107 = OpAccessChain %106 %105 %36
-%109 = OpUConvert %5 %93
-OpStore %107 %109 Aligned 4
-%111 = OpBitcast %14 %110
-OpStore %18 %111
+%37 = OpBitcast %36 %28
+%39 = OpAccessChain %38 %37 %40 %32
+%41 = OpLoad %33 %39 Aligned 4
+%43 = OpFConvert %42 %41
+%44 = OpIAdd %30 %31 %45
+%46 = OpSConvert %5 %44
+%47 = OpBitcast %36 %28
+%48 = OpAccessChain %38 %47 %40 %46
+%49 = OpLoad %33 %48 Aligned 4
+%50 = OpFConvert %42 %49
+%51 = OpFAdd %42 %50 %43
+%55 = OpBitcast %54 %25
+%56 = OpAccessChain %38 %55 %40 %32
+%58 = OpFConvert %33 %51
+OpStore %56 %58 Aligned 4
+%60 = OpAccessChain %59 %13 %40 %32
+%61 = OpLoad %5 %60
+%62 = OpUConvert %30 %61
+%63 = OpAccessChain %59 %13 %40 %46
+%64 = OpLoad %5 %63
+%65 = OpUConvert %30 %64
+%66 = OpIAdd %30 %65 %62
+%70 = OpBitcast %69 %22
+%72 = OpAccessChain %71 %70 %40 %32
+%74 = OpUConvert %5 %66
+OpStore %72 %74 Aligned 4
+%76 = OpBitcast %14 %75
+OpStore %18 %76
 OpReturn
 OpFunctionEnd
 #endif
diff --git a/reference/shaders/resources/ssbo-minprecision.sm60.ssbo.root-descriptor.frag b/reference/shaders/resources/ssbo-minprecision.sm60.ssbo.root-descriptor.frag
index 8cb4a89..dd70b90 100644
--- a/reference/shaders/resources/ssbo-minprecision.sm60.ssbo.root-descriptor.frag
+++ b/reference/shaders/resources/ssbo-minprecision.sm60.ssbo.root-descriptor.frag
@@ -1,22 +1,17 @@
 #version 460
 #extension GL_EXT_buffer_reference : require
+#extension GL_EXT_buffer_reference_uvec2 : require
 
-struct AddCarry
+layout(buffer_reference) buffer PhysicalPointerFloatNonWriteArray;
+layout(buffer_reference) buffer PhysicalPointerFloatArray;
+layout(buffer_reference, buffer_reference_align = 4, std430) readonly buffer PhysicalPointerFloatNonWriteArray
 {
-    uint _m0;
-    uint _m1;
+    float value[];
 };
 
-layout(buffer_reference) buffer PhysicalPointerFloatNonWrite;
-layout(buffer_reference) buffer PhysicalPointerFloat;
-layout(buffer_reference, std430) readonly buffer PhysicalPointerFloatNonWrite
+layout(buffer_reference, buffer_reference_align = 4, std430) buffer PhysicalPointerFloatArray
 {
-    float value;
-};
-
-layout(buffer_reference, std430) buffer PhysicalPointerFloat
-{
-    float value;
+    float value[];
 };
 
 layout(push_constant, std430) uniform RootConstants
@@ -30,18 +25,12 @@ layout(push_constant, std430) uniform RootConstants
 layout(location = 0) flat in mediump int A;
 layout(location = 0) out int SV_Target;
 
-float _70;
+float _43;
 
 void main()
 {
     uint _23 = uint(A);
-    AddCarry _34;
-    _34._m0 = uaddCarry(registers._m1.x, (_23 * 4u) + 0u, _34._m1);
-    AddCarry _48;
-    _48._m0 = uaddCarry(registers._m1.x, ((_23 + 1u) * 4u) + 0u, _48._m1);
-    AddCarry _63;
-    _63._m0 = uaddCarry(registers._m2.x, (_23 * 4u) + 0u, _63._m1);
-    PhysicalPointerFloat(uvec2(_63._m0, registers._m2.y + _63._m1)).value = PhysicalPointerFloatNonWrite(uvec2(_48._m0, registers._m1.y + _48._m1)).value + PhysicalPointerFloatNonWrite(uvec2(_34._m0, registers._m1.y + _34._m1)).value;
+    PhysicalPointerFloatArray(registers._m2).value[_23] = PhysicalPointerFloatNonWriteArray(registers._m1).value[_23 + 1u] + PhysicalPointerFloatNonWriteArray(registers._m1).value[_23];
     SV_Target = int(10u);
 }
 
@@ -51,7 +40,7 @@ void main()
 ; SPIR-V
 ; Version: 1.3
 ; Generator: Unknown(30017); 21022
-; Bound: 75
+; Bound: 48
 ; Schema: 0
 OpCapability Shader
 OpCapability PhysicalStorageBufferAddresses
@@ -64,11 +53,10 @@ OpName %7 "RootConstants"
 OpName %9 "registers"
 OpName %12 "A"
 OpName %14 "SV_Target"
-OpName %25 "PhysicalPointerFloatNonWrite"
-OpMemberName %25 0 "value"
-OpName %33 "AddCarry"
-OpName %57 "PhysicalPointerFloat"
-OpMemberName %57 0 "value"
+OpName %26 "PhysicalPointerFloatNonWriteArray"
+OpMemberName %26 0 "value"
+OpName %39 "PhysicalPointerFloatArray"
+OpMemberName %39 0 "value"
 OpDecorate %7 Block
 OpMemberDecorate %7 0 Offset 0
 OpMemberDecorate %7 1 Offset 8
@@ -78,11 +66,13 @@ OpDecorate %12 RelaxedPrecision
 OpDecorate %12 Flat
 OpDecorate %12 Location 0
 OpDecorate %14 Location 0
-OpMemberDecorate %25 0 Offset 0
-OpDecorate %25 Block
-OpMemberDecorate %25 0 NonWritable
-OpMemberDecorate %57 0 Offset 0
-OpDecorate %57 Block
+OpDecorate %25 ArrayStride 4
+OpMemberDecorate %26 0 Offset 0
+OpDecorate %26 Block
+OpMemberDecorate %26 0 NonWritable
+OpDecorate %38 ArrayStride 4
+OpMemberDecorate %39 0 Offset 0
+OpDecorate %39 Block
 %1 = OpTypeVoid
 %2 = OpTypeFunction %1
 %5 = OpTypeInt 32 0
@@ -99,66 +89,39 @@ OpDecorate %57 Block
 %17 = OpConstant %5 2
 %20 = OpConstant %5 1
 %24 = OpTypeFloat 32
-%25 = OpTypeStruct %24
-%26 = OpTypePointer PhysicalStorageBuffer %25
-%27 = OpConstant %5 0
-%29 = OpConstant %5 4
-%33 = OpTypeStruct %5 %5
-%40 = OpTypePointer PhysicalStorageBuffer %24
-%57 = OpTypeStruct %24
-%58 = OpTypePointer PhysicalStorageBuffer %57
-%71 = OpConstant %5 10
+%25 = OpTypeRuntimeArray %24
+%26 = OpTypeStruct %25
+%27 = OpTypePointer PhysicalStorageBuffer %26
+%29 = OpTypePointer PhysicalStorageBuffer %24
+%31 = OpConstant %5 0
+%38 = OpTypeRuntimeArray %24
+%39 = OpTypeStruct %38
+%40 = OpTypePointer PhysicalStorageBuffer %39
+%44 = OpConstant %5 10
 %3 = OpFunction %1 None %2
 %4 = OpLabel
-%70 = OpUndef %24
-OpBranch %73
-%73 = OpLabel
+%43 = OpUndef %24
+OpBranch %46
+%46 = OpLabel
 %16 = OpAccessChain %15 %9 %17
 %18 = OpLoad %6 %16
 %19 = OpAccessChain %15 %9 %20
 %21 = OpLoad %6 %19
 %22 = OpLoad %10 %12
 %23 = OpBitcast %5 %22
-%28 = OpIMul %5 %23 %29
-%30 = OpIAdd %5 %28 %27
-%31 = OpCompositeExtract %5 %21 0
-%32 = OpCompositeExtract %5 %21 1
-%34 = OpIAddCarry %33 %31 %30
-%35 = OpCompositeExtract %5 %34 0
-%36 = OpCompositeExtract %5 %34 1
-%37 = OpIAdd %5 %32 %36
-%38 = OpCompositeConstruct %6 %35 %37
-%39 = OpBitcast %26 %38
-%41 = OpAccessChain %40 %39 %27
-%42 = OpLoad %24 %41 Aligned 4
-%43 = OpIAdd %5 %23 %20
-%44 = OpIMul %5 %43 %29
-%45 = OpIAdd %5 %44 %27
-%46 = OpCompositeExtract %5 %21 0
-%47 = OpCompositeExtract %5 %21 1
-%48 = OpIAddCarry %33 %46 %45
-%49 = OpCompositeExtract %5 %48 0
-%50 = OpCompositeExtract %5 %48 1
-%51 = OpIAdd %5 %47 %50
-%52 = OpCompositeConstruct %6 %49 %51
-%53 = OpBitcast %26 %52
-%54 = OpAccessChain %40 %53 %27
-%55 = OpLoad %24 %54 Aligned 4
-%56 = OpFAdd %24 %55 %42
-%59 = OpIMul %5 %23 %29
-%60 = OpIAdd %5 %59 %27
-%61 = OpCompositeExtract %5 %18 0
-%62 = OpCompositeExtract %5 %18 1
-%63 = OpIAddCarry %33 %61 %60
-%64 = OpCompositeExtract %5 %63 0
-%65 = OpCompositeExtract %5 %63 1
-%66 = OpIAdd %5 %62 %65
-%67 = OpCompositeConstruct %6 %64 %66
-%68 = OpBitcast %58 %67
-%69 = OpAccessChain %40 %68 %27
-OpStore %69 %56 Aligned 4
-%72 = OpBitcast %10 %71
-OpStore %14 %72
+%28 = OpBitcast %27 %21
+%30 = OpAccessChain %29 %28 %31 %23
+%32 = OpLoad %24 %30 Aligned 4
+%33 = OpIAdd %5 %23 %20
+%34 = OpBitcast %27 %21
+%35 = OpAccessChain %29 %34 %31 %33
+%36 = OpLoad %24 %35 Aligned 4
+%37 = OpFAdd %24 %36 %32
+%41 = OpBitcast %40 %18
+%42 = OpAccessChain %29 %41 %31 %23
+OpStore %42 %37 Aligned 4
+%45 = OpBitcast %10 %44
+OpStore %14 %45
 OpReturn
 OpFunctionEnd
 #endif
diff --git a/reference/shaders/resources/typed-resources-16bit-sparse.frag b/reference/shaders/resources/typed-resources-16bit-sparse.frag
index f172beb..5bb8c77 100644
--- a/reference/shaders/resources/typed-resources-16bit-sparse.frag
+++ b/reference/shaders/resources/typed-resources-16bit-sparse.frag
@@ -120,16 +120,16 @@ void main()
     float _435;
     _434 = sparseTextureClampARB(sampler2DShadow(_8, _29), vec3(vec2(UV.x, UV.y), 0.5), 0.0, _435);
     SparseTexel_3 _200 = SparseTexel_3(_434, _435);
-    float _203 = _200._m1;
+    mediump float _203 = _200._m1;
     _204 _205 = _204(_203, _203, _203, _203, _200._m0);
-    float _206 = _205._m0;
+    mediump float _206 = _205._m0;
     uint _436;
     float _437;
     _436 = sparseTextureLodARB(sampler2DShadow(_8, _29), vec3(vec2(UV.x, UV.y), 0.5), 0.0, _437);
     SparseTexel_3 _222 = SparseTexel_3(_436, _437);
-    float _225 = _222._m1;
+    mediump float _225 = _222._m1;
     _204 _226 = _204(_225, _225, _225, _225, _222._m0);
-    float _227 = _226._m0;
+    mediump float _227 = _226._m0;
     vec2 _243 = vec2(UV.x, UV.y);
     uint _438;
     vec4 _439;
diff --git a/reference/shaders/resources/typed-resources-16bit.bindless.frag b/reference/shaders/resources/typed-resources-16bit.bindless.frag
index 1c7618f..e6be43d 100644
--- a/reference/shaders/resources/typed-resources-16bit.bindless.frag
+++ b/reference/shaders/resources/typed-resources-16bit.bindless.frag
@@ -67,10 +67,10 @@ void main()
     f16vec4 _225 = f16vec4(textureGather(sampler2D(_13[registers._m0], _62[registers._m2]), vec2(UV.x, UV.y)));
     u16vec4 _238 = u16vec4(textureGather(isampler2D(_18[_145], _62[registers._m2]), vec2(UV.x, UV.y), int(1u)));
     u16vec4 _251 = u16vec4(textureGather(usampler2D(_22[_139], _62[registers._m2]), vec2(UV.x, UV.y), int(2u)));
-    vec4 _266 = vec4(texture(sampler2DShadow(_13[registers._m0], _62[registers._m2 + 1u]), vec3(vec2(UV.x, UV.y), 0.5)));
-    float _267 = _266.x;
-    vec4 _282 = vec4(textureLod(sampler2DShadow(_13[registers._m0], _62[registers._m2 + 1u]), vec3(vec2(UV.x, UV.y), 0.5), 0.0));
-    float _283 = _282.x;
+    mediump vec4 _266 = vec4(texture(sampler2DShadow(_13[registers._m0], _62[registers._m2 + 1u]), vec3(vec2(UV.x, UV.y), 0.5)));
+    mediump float _267 = _266.x;
+    mediump vec4 _282 = vec4(textureLod(sampler2DShadow(_13[registers._m0], _62[registers._m2 + 1u]), vec3(vec2(UV.x, UV.y), 0.5), 0.0));
+    mediump float _283 = _282.x;
     vec2 _296 = vec2(UV.x, UV.y);
     f16vec4 _298 = f16vec4(textureGather(sampler2DShadow(_13[registers._m0], _62[registers._m2 + 1u]), _296, 0.5));
     f16vec4 _309 = f16vec4(textureLod(sampler2D(_13[registers._m0], _62[registers._m2]), vec2(UV.x, UV.y), 0.0));
diff --git a/reference/shaders/resources/typed-resources-16bit.frag b/reference/shaders/resources/typed-resources-16bit.frag
index 63ab00a..649a4d3 100644
--- a/reference/shaders/resources/typed-resources-16bit.frag
+++ b/reference/shaders/resources/typed-resources-16bit.frag
@@ -52,10 +52,10 @@ void main()
     f16vec4 _143 = f16vec4(textureGather(sampler2D(_8, _46), vec2(UV.x, UV.y)));
     u16vec4 _156 = u16vec4(textureGather(isampler2D(_12, _46), vec2(UV.x, UV.y), int(1u)));
     u16vec4 _169 = u16vec4(textureGather(usampler2D(_16, _46), vec2(UV.x, UV.y), int(2u)));
-    vec4 _184 = vec4(texture(sampler2DShadow(_8, _47), vec3(vec2(UV.x, UV.y), 0.5)));
-    float _185 = _184.x;
-    vec4 _200 = vec4(textureLod(sampler2DShadow(_8, _47), vec3(vec2(UV.x, UV.y), 0.5), 0.0));
-    float _201 = _200.x;
+    mediump vec4 _184 = vec4(texture(sampler2DShadow(_8, _47), vec3(vec2(UV.x, UV.y), 0.5)));
+    mediump float _185 = _184.x;
+    mediump vec4 _200 = vec4(textureLod(sampler2DShadow(_8, _47), vec3(vec2(UV.x, UV.y), 0.5), 0.0));
+    mediump float _201 = _200.x;
     vec2 _214 = vec2(UV.x, UV.y);
     f16vec4 _216 = f16vec4(textureGather(sampler2DShadow(_8, _47), _214, 0.5));
     f16vec4 _227 = f16vec4(textureLod(sampler2D(_8, _46), vec2(UV.x, UV.y), 0.0));
diff --git a/reference/shaders/resources/typed-resources-16bit.sm60.bindless.frag b/reference/shaders/resources/typed-resources-16bit.sm60.bindless.frag
index 75c9408..4a3719c 100644
--- a/reference/shaders/resources/typed-resources-16bit.sm60.bindless.frag
+++ b/reference/shaders/resources/typed-resources-16bit.sm60.bindless.frag
@@ -50,36 +50,68 @@ void main()
     imageStore(_50[registers._m4 + 3u], int(_167), vec4(8.0));
     imageStore(_54[registers._m4 + 4u], int(_167), ivec4(uvec4(4294967276u)));
     imageStore(_58[registers._m4 + 5u], int(_167), uvec4(80u));
-    vec4 _191 = texture(sampler2D(_13[registers._m0], _62[registers._m2]), vec2(UV.x, UV.y));
+    mediump vec4 _191 = texture(sampler2D(_13[registers._m0], _62[registers._m2]), vec2(UV.x, UV.y));
+    mediump float _193 = _191.x;
+    float hp_copy_193 = _193;
+    mediump float _194 = _191.y;
+    float hp_copy_194 = _194;
+    mediump float _195 = _191.z;
+    float hp_copy_195 = _195;
+    mediump float _196 = _191.w;
+    float hp_copy_196 = _196;
     uvec4 _199 = uvec4(texelFetch(_18[_145], ivec2(uvec2(1u, 2u)), int(3u)));
-    uvec4 _205 = texelFetch(_22[_139], ivec2(uvec2(4u, 5u)), int(6u));
-    vec4 _212 = textureGather(sampler2D(_13[registers._m0], _62[registers._m2]), vec2(UV.x, UV.y));
+    mediump uvec4 _205 = texelFetch(_22[_139], ivec2(uvec2(4u, 5u)), int(6u));
+    mediump uint _207 = _205.x;
+    uint hp_copy_207 = _207;
+    mediump uint _208 = _205.y;
+    uint hp_copy_208 = _208;
+    mediump uint _209 = _205.z;
+    uint hp_copy_209 = _209;
+    mediump uint _210 = _205.w;
+    uint hp_copy_210 = _210;
+    mediump vec4 _212 = textureGather(sampler2D(_13[registers._m0], _62[registers._m2]), vec2(UV.x, UV.y));
+    mediump float _213 = _212.x;
+    float hp_copy_213 = _213;
+    mediump float _214 = _212.y;
+    float hp_copy_214 = _214;
+    mediump float _215 = _212.z;
+    float hp_copy_215 = _215;
+    mediump float _216 = _212.w;
+    float hp_copy_216 = _216;
     uvec4 _225 = uvec4(textureGather(isampler2D(_18[_145], _62[registers._m2]), vec2(UV.x, UV.y), int(1u)));
-    uvec4 _237 = textureGather(usampler2D(_22[_139], _62[registers._m2]), vec2(UV.x, UV.y), int(2u));
-    vec4 _252 = vec4(texture(sampler2DShadow(_13[registers._m0], _62[registers._m2 + 1u]), vec3(vec2(UV.x, UV.y), 0.5)));
-    float _253 = _252.x;
-    vec4 _260 = vec4(textureLod(sampler2DShadow(_13[registers._m0], _62[registers._m2 + 1u]), vec3(vec2(UV.x, UV.y), 0.5), 0.0));
-    float _261 = _260.x;
+    mediump uvec4 _237 = textureGather(usampler2D(_22[_139], _62[registers._m2]), vec2(UV.x, UV.y), int(2u));
+    mediump uint _238 = _237.x;
+    uint hp_copy_238 = _238;
+    mediump uint _239 = _237.y;
+    uint hp_copy_239 = _239;
+    mediump uint _240 = _237.z;
+    uint hp_copy_240 = _240;
+    mediump uint _241 = _237.w;
+    uint hp_copy_241 = _241;
+    mediump vec4 _252 = vec4(texture(sampler2DShadow(_13[registers._m0], _62[registers._m2 + 1u]), vec3(vec2(UV.x, UV.y), 0.5)));
+    mediump float _253 = _252.x;
+    mediump vec4 _260 = vec4(textureLod(sampler2DShadow(_13[registers._m0], _62[registers._m2 + 1u]), vec3(vec2(UV.x, UV.y), 0.5), 0.0));
+    mediump float _261 = _260.x;
     vec2 _266 = vec2(UV.x, UV.y);
-    vec4 _267 = textureGather(sampler2DShadow(_13[registers._m0], _62[registers._m2 + 1u]), _266, 0.5);
-    vec4 _276 = textureLod(sampler2D(_13[registers._m0], _62[registers._m2]), vec2(UV.x, UV.y), 0.0);
-    vec4 _289 = textureGrad(sampler2D(_13[registers._m0], _62[registers._m2]), vec2(UV.x, UV.y), vec2(0.20000000298023223876953125, 0.300000011920928955078125), vec2(0.4000000059604644775390625, 0.5));
-    vec4 _301 = texture(sampler2D(_13[registers._m0], _62[registers._m2]), vec2(UV.x, UV.y), 0.5);
-    vec4 _311 = texelFetch(_26[registers._m1 + 3u], int(_167));
+    mediump vec4 _267 = textureGather(sampler2DShadow(_13[registers._m0], _62[registers._m2 + 1u]), _266, 0.5);
+    mediump vec4 _276 = textureLod(sampler2D(_13[registers._m0], _62[registers._m2]), vec2(UV.x, UV.y), 0.0);
+    mediump vec4 _289 = textureGrad(sampler2D(_13[registers._m0], _62[registers._m2]), vec2(UV.x, UV.y), vec2(0.20000000298023223876953125, 0.300000011920928955078125), vec2(0.4000000059604644775390625, 0.5));
+    mediump vec4 _301 = texture(sampler2D(_13[registers._m0], _62[registers._m2]), vec2(UV.x, UV.y), 0.5);
+    mediump vec4 _311 = texelFetch(_26[registers._m1 + 3u], int(_167));
     uvec4 _321 = uvec4(texelFetch(_30[registers._m1 + 4u], int(_167)));
-    uvec4 _330 = texelFetch(_34[registers._m1 + 5u], int(_167));
-    SV_Target.x = ((((_267.x + (_261 + (_253 + (_212.x + _191.x)))) + _276.x) + _289.x) + _301.x) + _311.x;
-    SV_Target.y = ((((_267.y + (_261 + (_253 + (_212.y + _191.y)))) + _276.y) + _289.y) + _301.y) + _311.y;
-    SV_Target.z = ((((_267.z + (_261 + (_253 + (_212.z + _191.z)))) + _276.z) + _289.z) + _301.z) + _311.z;
-    SV_Target.w = ((((_267.w + (_261 + (_253 + (_212.w + _191.w)))) + _276.w) + _289.w) + _301.w) + _311.w;
+    mediump uvec4 _330 = texelFetch(_34[registers._m1 + 5u], int(_167));
+    SV_Target.x = ((((_267.x + (_261 + (_253 + (hp_copy_213 + hp_copy_193)))) + _276.x) + _289.x) + _301.x) + _311.x;
+    SV_Target.y = ((((_267.y + (_261 + (_253 + (hp_copy_214 + hp_copy_194)))) + _276.y) + _289.y) + _301.y) + _311.y;
+    SV_Target.z = ((((_267.z + (_261 + (_253 + (hp_copy_215 + hp_copy_195)))) + _276.z) + _289.z) + _301.z) + _311.z;
+    SV_Target.w = ((((_267.w + (_261 + (_253 + (hp_copy_216 + hp_copy_196)))) + _276.w) + _289.w) + _301.w) + _311.w;
     SV_Target_1.x = int((_225.x + _199.x) + _321.x);
     SV_Target_1.y = int((_225.y + _199.y) + _321.y);
     SV_Target_1.z = int((_225.z + _199.z) + _321.z);
     SV_Target_1.w = int((_225.w + _199.w) + _321.w);
-    SV_Target_2.x = (_237.x + _205.x) + _330.x;
-    SV_Target_2.y = (_237.y + _205.y) + _330.y;
-    SV_Target_2.z = (_237.z + _205.z) + _330.z;
-    SV_Target_2.w = (_237.w + _205.w) + _330.w;
+    SV_Target_2.x = (hp_copy_238 + hp_copy_207) + _330.x;
+    SV_Target_2.y = (hp_copy_239 + hp_copy_208) + _330.y;
+    SV_Target_2.z = (hp_copy_240 + hp_copy_209) + _330.z;
+    SV_Target_2.w = (hp_copy_241 + hp_copy_210) + _330.w;
 }
 
 
diff --git a/reference/shaders/resources/typed-resources-16bit.sm60.frag b/reference/shaders/resources/typed-resources-16bit.sm60.frag
index c6b90d0..9305754 100644
--- a/reference/shaders/resources/typed-resources-16bit.sm60.frag
+++ b/reference/shaders/resources/typed-resources-16bit.sm60.frag
@@ -35,36 +35,68 @@ void main()
     imageStore(_37, int(_81), vec4(8.0));
     imageStore(_40, int(_81), ivec4(uvec4(4294967276u)));
     imageStore(_43, int(_81), uvec4(80u));
-    vec4 _105 = texture(sampler2D(_8, _46), vec2(UV.x, UV.y));
+    mediump vec4 _105 = texture(sampler2D(_8, _46), vec2(UV.x, UV.y));
+    mediump float _107 = _105.x;
+    float hp_copy_107 = _107;
+    mediump float _108 = _105.y;
+    float hp_copy_108 = _108;
+    mediump float _109 = _105.z;
+    float hp_copy_109 = _109;
+    mediump float _110 = _105.w;
+    float hp_copy_110 = _110;
     uvec4 _115 = uvec4(texelFetch(_12, ivec2(uvec2(1u, 2u)), int(3u)));
-    uvec4 _123 = texelFetch(_16, ivec2(uvec2(4u, 5u)), int(6u));
-    vec4 _130 = textureGather(sampler2D(_8, _46), vec2(UV.x, UV.y));
+    mediump uvec4 _123 = texelFetch(_16, ivec2(uvec2(4u, 5u)), int(6u));
+    mediump uint _125 = _123.x;
+    uint hp_copy_125 = _125;
+    mediump uint _126 = _123.y;
+    uint hp_copy_126 = _126;
+    mediump uint _127 = _123.z;
+    uint hp_copy_127 = _127;
+    mediump uint _128 = _123.w;
+    uint hp_copy_128 = _128;
+    mediump vec4 _130 = textureGather(sampler2D(_8, _46), vec2(UV.x, UV.y));
+    mediump float _131 = _130.x;
+    float hp_copy_131 = _131;
+    mediump float _132 = _130.y;
+    float hp_copy_132 = _132;
+    mediump float _133 = _130.z;
+    float hp_copy_133 = _133;
+    mediump float _134 = _130.w;
+    float hp_copy_134 = _134;
     uvec4 _143 = uvec4(textureGather(isampler2D(_12, _46), vec2(UV.x, UV.y), int(1u)));
-    uvec4 _155 = textureGather(usampler2D(_16, _46), vec2(UV.x, UV.y), int(2u));
-    vec4 _170 = vec4(texture(sampler2DShadow(_8, _47), vec3(vec2(UV.x, UV.y), 0.5)));
-    float _171 = _170.x;
-    vec4 _178 = vec4(textureLod(sampler2DShadow(_8, _47), vec3(vec2(UV.x, UV.y), 0.5), 0.0));
-    float _179 = _178.x;
+    mediump uvec4 _155 = textureGather(usampler2D(_16, _46), vec2(UV.x, UV.y), int(2u));
+    mediump uint _156 = _155.x;
+    uint hp_copy_156 = _156;
+    mediump uint _157 = _155.y;
+    uint hp_copy_157 = _157;
+    mediump uint _158 = _155.z;
+    uint hp_copy_158 = _158;
+    mediump uint _159 = _155.w;
+    uint hp_copy_159 = _159;
+    mediump vec4 _170 = vec4(texture(sampler2DShadow(_8, _47), vec3(vec2(UV.x, UV.y), 0.5)));
+    mediump float _171 = _170.x;
+    mediump vec4 _178 = vec4(textureLod(sampler2DShadow(_8, _47), vec3(vec2(UV.x, UV.y), 0.5), 0.0));
+    mediump float _179 = _178.x;
     vec2 _184 = vec2(UV.x, UV.y);
-    vec4 _185 = textureGather(sampler2DShadow(_8, _47), _184, 0.5);
-    vec4 _194 = textureLod(sampler2D(_8, _46), vec2(UV.x, UV.y), 0.0);
-    vec4 _207 = textureGrad(sampler2D(_8, _46), vec2(UV.x, UV.y), vec2(0.20000000298023223876953125, 0.300000011920928955078125), vec2(0.4000000059604644775390625, 0.5));
-    vec4 _219 = texture(sampler2D(_8, _46), vec2(UV.x, UV.y), 0.5);
-    vec4 _229 = texelFetch(_19, int(_81));
+    mediump vec4 _185 = textureGather(sampler2DShadow(_8, _47), _184, 0.5);
+    mediump vec4 _194 = textureLod(sampler2D(_8, _46), vec2(UV.x, UV.y), 0.0);
+    mediump vec4 _207 = textureGrad(sampler2D(_8, _46), vec2(UV.x, UV.y), vec2(0.20000000298023223876953125, 0.300000011920928955078125), vec2(0.4000000059604644775390625, 0.5));
+    mediump vec4 _219 = texture(sampler2D(_8, _46), vec2(UV.x, UV.y), 0.5);
+    mediump vec4 _229 = texelFetch(_19, int(_81));
     uvec4 _239 = uvec4(texelFetch(_22, int(_81)));
-    uvec4 _248 = texelFetch(_25, int(_81));
-    SV_Target.x = ((((_185.x + (_179 + (_171 + (_130.x + _105.x)))) + _194.x) + _207.x) + _219.x) + _229.x;
-    SV_Target.y = ((((_185.y + (_179 + (_171 + (_130.y + _105.y)))) + _194.y) + _207.y) + _219.y) + _229.y;
-    SV_Target.z = ((((_185.z + (_179 + (_171 + (_130.z + _105.z)))) + _194.z) + _207.z) + _219.z) + _229.z;
-    SV_Target.w = ((((_185.w + (_179 + (_171 + (_130.w + _105.w)))) + _194.w) + _207.w) + _219.w) + _229.w;
+    mediump uvec4 _248 = texelFetch(_25, int(_81));
+    SV_Target.x = ((((_185.x + (_179 + (_171 + (hp_copy_131 + hp_copy_107)))) + _194.x) + _207.x) + _219.x) + _229.x;
+    SV_Target.y = ((((_185.y + (_179 + (_171 + (hp_copy_132 + hp_copy_108)))) + _194.y) + _207.y) + _219.y) + _229.y;
+    SV_Target.z = ((((_185.z + (_179 + (_171 + (hp_copy_133 + hp_copy_109)))) + _194.z) + _207.z) + _219.z) + _229.z;
+    SV_Target.w = ((((_185.w + (_179 + (_171 + (hp_copy_134 + hp_copy_110)))) + _194.w) + _207.w) + _219.w) + _229.w;
     SV_Target_1.x = int((_143.x + _115.x) + _239.x);
     SV_Target_1.y = int((_143.y + _115.y) + _239.y);
     SV_Target_1.z = int((_143.z + _115.z) + _239.z);
     SV_Target_1.w = int((_143.w + _115.w) + _239.w);
-    SV_Target_2.x = (_155.x + _123.x) + _248.x;
-    SV_Target_2.y = (_155.y + _123.y) + _248.y;
-    SV_Target_2.z = (_155.z + _123.z) + _248.z;
-    SV_Target_2.w = (_155.w + _123.w) + _248.w;
+    SV_Target_2.x = (hp_copy_156 + hp_copy_125) + _248.x;
+    SV_Target_2.y = (hp_copy_157 + hp_copy_126) + _248.y;
+    SV_Target_2.z = (hp_copy_158 + hp_copy_127) + _248.z;
+    SV_Target_2.w = (hp_copy_159 + hp_copy_128) + _248.w;
 }
 
 
diff --git a/reference/shaders/resources/typed-resources-16bit.sm60.native-fp16.bindless.frag b/reference/shaders/resources/typed-resources-16bit.sm60.native-fp16.bindless.frag
index 1c7618f..e6be43d 100644
--- a/reference/shaders/resources/typed-resources-16bit.sm60.native-fp16.bindless.frag
+++ b/reference/shaders/resources/typed-resources-16bit.sm60.native-fp16.bindless.frag
@@ -67,10 +67,10 @@ void main()
     f16vec4 _225 = f16vec4(textureGather(sampler2D(_13[registers._m0], _62[registers._m2]), vec2(UV.x, UV.y)));
     u16vec4 _238 = u16vec4(textureGather(isampler2D(_18[_145], _62[registers._m2]), vec2(UV.x, UV.y), int(1u)));
     u16vec4 _251 = u16vec4(textureGather(usampler2D(_22[_139], _62[registers._m2]), vec2(UV.x, UV.y), int(2u)));
-    vec4 _266 = vec4(texture(sampler2DShadow(_13[registers._m0], _62[registers._m2 + 1u]), vec3(vec2(UV.x, UV.y), 0.5)));
-    float _267 = _266.x;
-    vec4 _282 = vec4(textureLod(sampler2DShadow(_13[registers._m0], _62[registers._m2 + 1u]), vec3(vec2(UV.x, UV.y), 0.5), 0.0));
-    float _283 = _282.x;
+    mediump vec4 _266 = vec4(texture(sampler2DShadow(_13[registers._m0], _62[registers._m2 + 1u]), vec3(vec2(UV.x, UV.y), 0.5)));
+    mediump float _267 = _266.x;
+    mediump vec4 _282 = vec4(textureLod(sampler2DShadow(_13[registers._m0], _62[registers._m2 + 1u]), vec3(vec2(UV.x, UV.y), 0.5), 0.0));
+    mediump float _283 = _282.x;
     vec2 _296 = vec2(UV.x, UV.y);
     f16vec4 _298 = f16vec4(textureGather(sampler2DShadow(_13[registers._m0], _62[registers._m2 + 1u]), _296, 0.5));
     f16vec4 _309 = f16vec4(textureLod(sampler2D(_13[registers._m0], _62[registers._m2]), vec2(UV.x, UV.y), 0.0));
diff --git a/reference/shaders/resources/typed-resources-16bit.sm60.native-fp16.frag b/reference/shaders/resources/typed-resources-16bit.sm60.native-fp16.frag
index 63ab00a..649a4d3 100644
--- a/reference/shaders/resources/typed-resources-16bit.sm60.native-fp16.frag
+++ b/reference/shaders/resources/typed-resources-16bit.sm60.native-fp16.frag
@@ -52,10 +52,10 @@ void main()
     f16vec4 _143 = f16vec4(textureGather(sampler2D(_8, _46), vec2(UV.x, UV.y)));
     u16vec4 _156 = u16vec4(textureGather(isampler2D(_12, _46), vec2(UV.x, UV.y), int(1u)));
     u16vec4 _169 = u16vec4(textureGather(usampler2D(_16, _46), vec2(UV.x, UV.y), int(2u)));
-    vec4 _184 = vec4(texture(sampler2DShadow(_8, _47), vec3(vec2(UV.x, UV.y), 0.5)));
-    float _185 = _184.x;
-    vec4 _200 = vec4(textureLod(sampler2DShadow(_8, _47), vec3(vec2(UV.x, UV.y), 0.5), 0.0));
-    float _201 = _200.x;
+    mediump vec4 _184 = vec4(texture(sampler2DShadow(_8, _47), vec3(vec2(UV.x, UV.y), 0.5)));
+    mediump float _185 = _184.x;
+    mediump vec4 _200 = vec4(textureLod(sampler2DShadow(_8, _47), vec3(vec2(UV.x, UV.y), 0.5), 0.0));
+    mediump float _201 = _200.x;
     vec2 _214 = vec2(UV.x, UV.y);
     f16vec4 _216 = f16vec4(textureGather(sampler2DShadow(_8, _47), _214, 0.5));
     f16vec4 _227 = f16vec4(textureLod(sampler2D(_8, _46), vec2(UV.x, UV.y), 0.0));
diff --git a/reference/shaders/resources/uav-counter.bindless.root-constant.comp b/reference/shaders/resources/uav-counter.bindless.root-constant.comp
index f56bafc..339c02c 100644
--- a/reference/shaders/resources/uav-counter.bindless.root-constant.comp
+++ b/reference/shaders/resources/uav-counter.bindless.root-constant.comp
@@ -4,7 +4,7 @@
 layout(local_size_x = 64, local_size_y = 1, local_size_z = 1) in;
 
 layout(buffer_reference) buffer AtomicCounter;
-layout(buffer_reference, std430) buffer AtomicCounter
+layout(buffer_reference, buffer_reference_align = 4, std430) buffer AtomicCounter
 {
     uint _m0;
 };
diff --git a/third_party/SPIRV-Cross b/third_party/SPIRV-Cross
-Subproject bab4e5911b1bfa5a86bc80006b7301ae4836384
+Subproject b3ff97d0feafd2b7ca72aec7215cfc3d0998fb7
diff --git a/third_party/SPIRV-Tools b/third_party/SPIRV-Tools
-Subproject 2d12367ced2dd34444822340070b4545ae7c02f
+Subproject 98340ec500e92a534dc8384d5c45d2f488e40f6
diff --git a/third_party/spirv-headers b/third_party/spirv-headers
-Subproject ae217c17809fadb232ec94b29304b4afcd417bb
+Subproject b765c355f488837ca4c77980ba69484f3ff277f