MSL/HLSL: Support OpOuterProduct.

1 files changed, 48 insertions, 0 deletions

diff --git a/reference/shaders-hlsl/comp/outer-product.comp b/reference/shaders-hlsl/comp/outer-product.comp
new file mode 100644
index 00000000..71613d4f
--- /dev/null
+++ b/reference/shaders-hlsl/comp/outer-product.comp
@@ -0,0 +1,48 @@
+RWByteAddressBuffer _21 : register(u0);
+ByteAddressBuffer _26 : register(t1);
+
+void comp_main()
+{
+    float2x2 _32 = float2x2(asfloat(_26.Load2(0)) * asfloat(_26.Load2(0)).x, asfloat(_26.Load2(0)) * asfloat(_26.Load2(0)).y);
+    _21.Store2(0, asuint(_32[0]));
+    _21.Store2(8, asuint(_32[1]));
+    float2x3 _41 = float2x3(asfloat(_26.Load3(16)) * asfloat(_26.Load2(0)).x, asfloat(_26.Load3(16)) * asfloat(_26.Load2(0)).y);
+    _21.Store3(16, asuint(_41[0]));
+    _21.Store3(32, asuint(_41[1]));
+    float2x4 _50 = float2x4(asfloat(_26.Load4(32)) * asfloat(_26.Load2(0)).x, asfloat(_26.Load4(32)) * asfloat(_26.Load2(0)).y);
+    _21.Store4(48, asuint(_50[0]));
+    _21.Store4(64, asuint(_50[1]));
+    float3x2 _58 = float3x2(asfloat(_26.Load2(0)) * asfloat(_26.Load3(16)).x, asfloat(_26.Load2(0)) * asfloat(_26.Load3(16)).y, asfloat(_26.Load2(0)) * asfloat(_26.Load3(16)).z);
+    _21.Store2(80, asuint(_58[0]));
+    _21.Store2(88, asuint(_58[1]));
+    _21.Store2(96, asuint(_58[2]));
+    float3x3 _66 = float3x3(asfloat(_26.Load3(16)) * asfloat(_26.Load3(16)).x, asfloat(_26.Load3(16)) * asfloat(_26.Load3(16)).y, asfloat(_26.Load3(16)) * asfloat(_26.Load3(16)).z);
+    _21.Store3(112, asuint(_66[0]));
+    _21.Store3(128, asuint(_66[1]));
+    _21.Store3(144, asuint(_66[2]));
+    float3x4 _74 = float3x4(asfloat(_26.Load4(32)) * asfloat(_26.Load3(16)).x, asfloat(_26.Load4(32)) * asfloat(_26.Load3(16)).y, asfloat(_26.Load4(32)) * asfloat(_26.Load3(16)).z);
+    _21.Store4(160, asuint(_74[0]));
+    _21.Store4(176, asuint(_74[1]));
+    _21.Store4(192, asuint(_74[2]));
+    float4x2 _82 = float4x2(asfloat(_26.Load2(0)) * asfloat(_26.Load4(32)).x, asfloat(_26.Load2(0)) * asfloat(_26.Load4(32)).y, asfloat(_26.Load2(0)) * asfloat(_26.Load4(32)).z, asfloat(_26.Load2(0)) * asfloat(_26.Load4(32)).w);
+    _21.Store2(208, asuint(_82[0]));
+    _21.Store2(216, asuint(_82[1]));
+    _21.Store2(224, asuint(_82[2]));
+    _21.Store2(232, asuint(_82[3]));
+    float4x3 _90 = float4x3(asfloat(_26.Load3(16)) * asfloat(_26.Load4(32)).x, asfloat(_26.Load3(16)) * asfloat(_26.Load4(32)).y, asfloat(_26.Load3(16)) * asfloat(_26.Load4(32)).z, asfloat(_26.Load3(16)) * asfloat(_26.Load4(32)).w);
+    _21.Store3(240, asuint(_90[0]));
+    _21.Store3(256, asuint(_90[1]));
+    _21.Store3(272, asuint(_90[2]));
+    _21.Store3(288, asuint(_90[3]));
+    float4x4 _98 = float4x4(asfloat(_26.Load4(32)) * asfloat(_26.Load4(32)).x, asfloat(_26.Load4(32)) * asfloat(_26.Load4(32)).y, asfloat(_26.Load4(32)) * asfloat(_26.Load4(32)).z, asfloat(_26.Load4(32)) * asfloat(_26.Load4(32)).w);
+    _21.Store4(304, asuint(_98[0]));
+    _21.Store4(320, asuint(_98[1]));
+    _21.Store4(336, asuint(_98[2]));
+    _21.Store4(352, asuint(_98[3]));
+}
+
+[numthreads(1, 1, 1)]
+void main()
+{
+    comp_main();
+}