cuda: Add more formats for texture read/write (#7012)

* WiP: Add more formats for texture reads

* fix test

* format code

* add  float2/float4 versions for 1D and 3D as well

* fixed review comment

* fix review comments

---------

Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com>
Co-authored-by: Ellie Hermaszewska <ellieh@nvidia.com>

1 files changed, 86 insertions, 0 deletions

diff --git a/tests/compute/texture-subscript-uint-cuda.slang b/tests/compute/texture-subscript-uint-cuda.slang
new file mode 100644
index 000000000..d2b6cec59
--- /dev/null
+++ b/tests/compute/texture-subscript-uint-cuda.slang
@@ -0,0 +1,86 @@
+// Test for verifying subscript operator support with uint types in cuda.
+
+//TEST(compute):COMPARE_COMPUTE_EX(filecheck-buffer=CHECK):-cuda -compute -shaderobj
+//TEST_INPUT: Texture1D(size=4, content = one):name cudaT1D_u
+Texture1D<uint> cudaT1D_u;
+//TEST_INPUT: Texture2D(size=8, content = one):name cudaT2D_u
+Texture2D<uint> cudaT2D_u;
+//TEST_INPUT: Texture2D(size=8, content = one):name cudaT2D_u2
+Texture2D<uint2> cudaT2D_u2;
+//TEST_INPUT: Texture2D(size=8, content = one):name cudaT2D_u4
+Texture2D<uint4> cudaT2D_u4;
+//TEST_INPUT: Texture3D(size=8, content = one):name cudaT3D_u
+Texture3D<uint> cudaT3D_u;
+//TEST_INPUT: Texture3D(size=8, content = one):name cudaT3D_u2
+Texture3D<uint2> cudaT3D_u2;
+//TEST_INPUT: Texture3D(size=8, content = one):name cudaT3D_u4
+Texture3D<uint4> cudaT3D_u4;
+//TEST_INPUT: TextureCube(size=16, content = one):name cudaTCube_u
+TextureCube<uint> cudaTCube_u;
+//TEST_INPUT: Texture2D(size=16, content = one, arrayLength=3):name cudaT2DArray_u
+Texture2DArray<uint> cudaT2DArray_u;
+//TEST_INPUT: Texture2D(size=16, content = one, arrayLength=3):name cudaT2DArray_u2
+Texture2DArray<uint2> cudaT2DArray_u2;
+//TEST_INPUT: Texture2D(size=16, content = one, arrayLength=3):name cudaT2DArray_u4
+Texture2DArray<uint4> cudaT2DArray_u4;
+
+//TEST_INPUT: ubuffer(data=[0 0 0], stride=4):out,name cudaOutputBuffer
+RWStructuredBuffer<uint> cudaOutputBuffer;
+
+[numthreads(3, 1, 1)]
+[shader("compute")]
+void computeMain(int3 dispatchThreadID : SV_DispatchThreadID)
+{
+    int idx = dispatchThreadID.x;
+    
+    switch (idx)
+    {   
+    case 0:
+    {
+        int2 var = int2(1, 2);
+        uint result = cudaT2D_u[var];
+        cudaOutputBuffer[idx] = result;
+            
+        uint2 result2 = cudaT2D_u2[var];
+        cudaOutputBuffer[idx] += result2.x;
+
+        uint4 result4 = cudaT2D_u4[var];
+        cudaOutputBuffer[idx] += result4.x;
+        cudaOutputBuffer[idx] += result4.y;
+    }
+    break;
+
+    case 1:
+    {
+        int3 var = int3(1, 1, 1);
+        uint result = cudaT3D_u[var];
+        cudaOutputBuffer[idx] = result;
+            
+        uint2 result2 = cudaT3D_u2[var];
+        cudaOutputBuffer[idx] += result2.x;
+
+        uint4 result4 = cudaT3D_u4[var];
+        cudaOutputBuffer[idx] += result4.x;
+        cudaOutputBuffer[idx] += result4.w;
+    }
+    break;
+
+    case 2:
+    {
+        int3 var = int3(0, 0, 1);
+        uint result = cudaT2DArray_u[var];
+        cudaOutputBuffer[idx] = result;
+            
+        uint2 result2 = cudaT2DArray_u2[var];
+        cudaOutputBuffer[idx] += result2.y;
+
+        uint4 result4 = cudaT2DArray_u4[var];
+        cudaOutputBuffer[idx] += result4.x;
+        cudaOutputBuffer[idx] += result4.y;
+    }
+    break;
+    }
+} 
+// CHECK:FE000000
+// CHECK:FE000000
+// CHECK:FE000000
\ No newline at end of file