Cycles: Pack kernel textures into buffers for OpenCL

Image textures were being packed into a single buffer for OpenCL, which
limited the amount of memory available for images to the size of one
buffer (usually 4gb on AMD hardware). By packing textures into multiple
buffers that limit is removed, while simultaneously reducing the number
of buffers that need to be passed to each kernel.

Benchmarks were within 2%.

Fixes T51554.

Differential Revision: https://developer.blender.org/D2745

1 files changed, 67 insertions, 1 deletions

diff --git a/intern/cycles/kernel/kernel_globals.h b/intern/cycles/kernel/kernel_globals.h
index f95f0d98c52..c078f09e1d7 100644
--- a/intern/cycles/kernel/kernel_globals.h
+++ b/intern/cycles/kernel/kernel_globals.h
@@ -23,6 +23,10 @@
 #  include "util/util_vector.h"
 #endif
 
+#ifdef __KERNEL_OPENCL__
+#  include "util/util_atomic.h"
+#endif
+
 CCL_NAMESPACE_BEGIN
 
 /* On the CPU, we pass along the struct KernelGlobals to nearly everywhere in
@@ -109,11 +113,22 @@ typedef struct KernelGlobals {
 
 #ifdef __KERNEL_OPENCL__
 
+#  define KERNEL_TEX(type, ttype, name) \
+typedef type name##_t;
+#  include "kernel/kernel_textures.h"
+
+typedef struct tex_info_t {
+	uint buffer, padding;
+	ulong offset;
+	uint width, height, depth, options;
+} tex_info_t;
+
 typedef ccl_addr_space struct KernelGlobals {
 	ccl_constant KernelData *data;
+	ccl_global char *buffers[8];
 
 #  define KERNEL_TEX(type, ttype, name) \
-	ccl_global type *name;
+	tex_info_t name;
 #  include "kernel/kernel_textures.h"
 
 #  ifdef __SPLIT_KERNEL__
@@ -122,6 +137,57 @@ typedef ccl_addr_space struct KernelGlobals {
 #  endif
 } KernelGlobals;
 
+#define KERNEL_BUFFER_PARAMS \
+	ccl_global char *buffer0, \
+	ccl_global char *buffer1, \
+	ccl_global char *buffer2, \
+	ccl_global char *buffer3, \
+	ccl_global char *buffer4, \
+	ccl_global char *buffer5, \
+	ccl_global char *buffer6, \
+	ccl_global char *buffer7
+
+#define KERNEL_BUFFER_ARGS buffer0, buffer1, buffer2, buffer3, buffer4, buffer5, buffer6, buffer7
+
+ccl_device_inline void kernel_set_buffer_pointers(KernelGlobals *kg, KERNEL_BUFFER_PARAMS)
+{
+#ifdef __SPLIT_KERNEL__
+	if(ccl_local_id(0) + ccl_local_id(1) == 0)
+#endif
+	{
+		kg->buffers[0] = buffer0;
+		kg->buffers[1] = buffer1;
+		kg->buffers[2] = buffer2;
+		kg->buffers[3] = buffer3;
+		kg->buffers[4] = buffer4;
+		kg->buffers[5] = buffer5;
+		kg->buffers[6] = buffer6;
+		kg->buffers[7] = buffer7;
+	}
+
+#  ifdef __SPLIT_KERNEL__
+	ccl_barrier(CCL_LOCAL_MEM_FENCE);
+#  endif
+}
+
+ccl_device_inline void kernel_set_buffer_info(KernelGlobals *kg)
+{
+#  ifdef __SPLIT_KERNEL__
+	if(ccl_local_id(0) + ccl_local_id(1) == 0)
+#  endif
+	{
+		ccl_global tex_info_t *info = (ccl_global tex_info_t*)kg->buffers[0];
+
+#  define KERNEL_TEX(type, ttype, name) \
+		kg->name = *(info++);
+#  include "kernel/kernel_textures.h"
+	}
+
+#  ifdef __SPLIT_KERNEL__
+	ccl_barrier(CCL_LOCAL_MEM_FENCE);
+#  endif
+}
+
 #endif  /* __KERNEL_OPENCL__ */
 
 /* Interpolated lookup table access */