Respect matrix layout in uniform and in/out parameters for HLSL target. (#5013)

* Respect matrix layout in uniform and in/out parameters for HLSL target.

* Update test.

* Fix test.

* fix test.

* Fix metal layout calculation.

* Fix compile error.

* Fix compiler error.

---------

Co-authored-by: Yong He <yhe@nvidia.com>

1 files changed, 45 insertions, 48 deletions

diff --git a/tests/compute/constant-buffer-memory-packing.slang b/tests/compute/constant-buffer-memory-packing.slang
index 5246c4d33..fb7a78fc1 100644
--- a/tests/compute/constant-buffer-memory-packing.slang
+++ b/tests/compute/constant-buffer-memory-packing.slang
@@ -3,31 +3,31 @@
 // Metal/CPP/CUDA do not deal with packing currently, different results will occur.
 //TEST(compute):COMPARE_COMPUTE_EX(filecheck-buffer=BUF):-cpu -compute -xslang -DTARGET_WITHOUT_PACKING
 //TEST(compute):COMPARE_COMPUTE_EX(filecheck-buffer=BUF):-cuda -compute -xslang -DTARGET_WITHOUT_PACKING
-//TEST(compute):COMPARE_COMPUTE_EX(filecheck-buffer=BUF):-mtl -compute -xslang -DTARGET_WITHOUT_PACKING
 
-//TEST(compute):COMPARE_COMPUTE_EX(filecheck-buffer=BUF):-slang -compute -dx12
-//TEST(compute, vulkan):COMPARE_COMPUTE_EX(filecheck-buffer=BUF):-vk -compute
 //TEST(compute, vulkan):COMPARE_COMPUTE_EX(filecheck-buffer=BUF):-vk -compute -emit-spirv-via-glsl
+//TEST(compute):COMPARE_COMPUTE_EX(filecheck-buffer=BUF):-slang -compute
+//TEST(compute, vulkan):COMPARE_COMPUTE_EX(filecheck-buffer=BUF):-vk -compute
+//TEST(compute):COMPARE_COMPUTE_EX(filecheck-buffer=BUF):-mtl -compute
 
-// CPP/Metal/CUDA due to not having memory packing will recieve the following ROW matrix:
+// CPP/CUDA due to natural layout rules will recieve the following ROW matrix:
 // {1,2,3}
 // {0,4,5}
 // {6,0,7}
 
-// GLSL/SPIRV/HLSL due to having memory packing will recieve the following ROW/COL matrix:
+// GLSL/SPIRV/HLSL/Metal with cbuffer layout rules will recieve the following ROW/COL matrix:
 // {1,2,3}
-// {0,4,5}
-// {6,0,7}
+// {4,5,6}
+// {7,8,9}
 
-//TEST_INPUT:cbuffer(data=[1.0 2.0 3.0 0.0  4.0 5.0 6.0 0.0  7.0 8.0 9.0 0]):name matrixTestCBuf1
+//TEST_INPUT:cbuffer(data=[1.0 2.0 3.0 0.0  4.0 5.0 6.0 0.0  7.0 8.0 9.0 0.0]):name matrixTestCBuf1
 ConstantBuffer<row_major float3x3> matrixTestCBuf1;
 
 // CPP/Metal/CUDA due to not having memory packing will recieve the following COL matrix post-transpose:
-// {1,0,8}
-// {4,2,0}
-// {7,5,3}
+// {1,0,6}
+// {2,4,0}
+// {3,5,7}
 
-//TEST_INPUT:cbuffer(data=[1.0 4.0 7.0 0.0  2.0 5.0 8.0 0.0  3.0 6.0 9.0 0.0]):name matrixTestCBuf2
+//TEST_INPUT:cbuffer(data=[1.0 2.0 3.0 0.0  4.0 5.0 6.0 0.0  7.0 8.0 9.0 0.0]):name matrixTestCBuf2
 ConstantBuffer<column_major float3x3> matrixTestCBuf2;
 
 //TEST_INPUT:cbuffer(data=[1.0 2.0 3.0 0.0  4.0 5.0 6.0 0.0]):name NeedsPadding
@@ -55,34 +55,7 @@ void computeMain(uint3 tid : SV_DispatchThreadID)
     matrixTest2 = matrixTestCBuf2;
 
     output[0] = uint(true
-#ifndef TARGET_WITHOUT_PACKING
-            && floatCheck(matrixTest1[0][0], 1)
-            && floatCheck(matrixTest1[0][1], 2)
-            && floatCheck(matrixTest1[0][2], 3)
-            && floatCheck(matrixTest1[1][0], 4)
-            && floatCheck(matrixTest1[1][1], 5)
-            && floatCheck(matrixTest1[1][2], 6)
-            && floatCheck(matrixTest1[2][0], 7)
-            && floatCheck(matrixTest1[2][1], 8)
-            && floatCheck(matrixTest1[2][2], 9)
-
-            && floatCheck(matrixTest2[0][0], 1)
-            && floatCheck(matrixTest2[0][1], 2)
-            && floatCheck(matrixTest2[0][2], 3)
-            && floatCheck(matrixTest2[1][0], 4)
-            && floatCheck(matrixTest2[1][1], 5)
-            && floatCheck(matrixTest2[1][2], 6)
-            && floatCheck(matrixTest2[2][0], 7)
-            && floatCheck(matrixTest2[2][1], 8)
-            && floatCheck(matrixTest2[2][2], 9)
-
-            && floatCheck(data1[0], 1)
-            && floatCheck(data1[1], 2)
-            && floatCheck(data1[2], 3)
-            && floatCheck(data2[0], 4)
-            && floatCheck(data2[1], 5)
-            && floatCheck(data2[2], 6)
-#else
+#ifdef TARGET_WITHOUT_PACKING
             && floatCheck(matrixTest1[0][0], 1)
             && floatCheck(matrixTest1[0][1], 2)
             && floatCheck(matrixTest1[0][2], 3)
@@ -95,13 +68,13 @@ void computeMain(uint3 tid : SV_DispatchThreadID)
 
             && floatCheck(matrixTest2[0][0], 1)
             && floatCheck(matrixTest2[0][1], 0)
-            && floatCheck(matrixTest2[0][2], 8)
-            && floatCheck(matrixTest2[1][0], 4)
-            && floatCheck(matrixTest2[1][1], 2)
+            && floatCheck(matrixTest2[0][2], 6)
+            && floatCheck(matrixTest2[1][0], 2)
+            && floatCheck(matrixTest2[1][1], 4)
             && floatCheck(matrixTest2[1][2], 0)
-            && floatCheck(matrixTest2[2][0], 7)
+            && floatCheck(matrixTest2[2][0], 3)
             && floatCheck(matrixTest2[2][1], 5)
-            && floatCheck(matrixTest2[2][2], 3)
+            && floatCheck(matrixTest2[2][2], 7)
 
             && floatCheck(data1[0], 1)
             && floatCheck(data1[1], 2)
@@ -109,10 +82,34 @@ void computeMain(uint3 tid : SV_DispatchThreadID)
             && floatCheck(data2[0], 0)
             && floatCheck(data2[1], 4)
             && floatCheck(data2[2], 5)
+#else
+            && floatCheck(matrixTest1[0][0], 1)
+            && floatCheck(matrixTest1[0][1], 2)
+            && floatCheck(matrixTest1[0][2], 3)
+            && floatCheck(matrixTest1[1][0], 4)
+            && floatCheck(matrixTest1[1][1], 5)
+            && floatCheck(matrixTest1[1][2], 6)
+            && floatCheck(matrixTest1[2][0], 7)
+            && floatCheck(matrixTest1[2][1], 8)
+            && floatCheck(matrixTest1[2][2], 9)
+
+            && floatCheck(matrixTest2[0][0], 1)
+            && floatCheck(matrixTest2[0][1], 4)
+            && floatCheck(matrixTest2[0][2], 7)
+            && floatCheck(matrixTest2[1][0], 2)
+            && floatCheck(matrixTest2[1][1], 5)
+            && floatCheck(matrixTest2[1][2], 8)
+            && floatCheck(matrixTest2[2][0], 3)
+            && floatCheck(matrixTest2[2][1], 6)
+            && floatCheck(matrixTest2[2][2], 9)
+
+            && floatCheck(data1[0], 1)
+            && floatCheck(data1[1], 2)
+            && floatCheck(data1[2], 3)
+            && floatCheck(data2[0], 4)
+            && floatCheck(data2[1], 5)
+            && floatCheck(data2[2], 6)
 #endif
         );
-    output[1] = (uint)matrixTest2[0][0];
-    output[2] = (uint)matrixTest2[0][1];
-    output[3] = (uint)matrixTest2[0][2];
     //BUF: 1
 }