Add cooperative matrix 1 support (#6565)

* initial wip for spirv

* working tiled example

* clean up store and load

* minor fixes

* fix loadAny name

* add initial tests, including broken/unimplemented intrinsics

* fix subscript

* run tests at 16x16, remove not supported arithmetic tests

* minor fixups on implementation

* rename CoopMatMatrixUse

* Update tests to pass validation layers locally

* Add mat-mul-add test and minor fixes

* Add more tests

* Remove dead code

* Add coopMatLoad function and tests, enforce constexpr for matrix layout

* Use getVectorOrCoopMatrixElementType in place of getVectorElementType

1 files changed, 32 insertions, 0 deletions

diff --git a/tests/cooperative-matrix/return.slang b/tests/cooperative-matrix/return.slang
new file mode 100644
index 000000000..339c9d04d
--- /dev/null
+++ b/tests/cooperative-matrix/return.slang
@@ -0,0 +1,32 @@
+//TEST(compute):COMPARE_COMPUTE(filecheck-buffer=CHECK):-vk -output-using-type
+
+// CHECK: type: float
+// CHECK-NEXT: 3.000000
+// CHECK-NEXT: 6.000000
+// CHECK-NEXT: 9.000000
+// CHECK-NEXT: 12.000000
+
+//TEST_INPUT:ubuffer(data=[1.0 2.0 3.0 4.0], stride=4, count=256):name=inputBuffer
+StructuredBuffer<float> inputBuffer;
+
+//TEST_INPUT:ubuffer(stride=4, count=256):out,name=outputBuffer
+RWStructuredBuffer<float> outputBuffer;
+
+typealias CoopMatType = CoopMat<float, CoopMatScope::Subgroup, 16, 16, CoopMatMatrixUse::MatrixAccumulator>;
+
+CoopMatType doubleCoopmat(CoopMatType mat)
+{
+    return mat * 3.0;
+}
+
+[numthreads(32, 1, 1)]
+void computeMain()
+{
+    let stride = 16;
+    let matrixLayout = CoopMatMatrixLayout::RowMajor;
+
+    let mat = CoopMatType.load(inputBuffer, 0, stride, matrixLayout);
+
+    let result = doubleCoopmat(mat);
+    result.store(outputBuffer, 0, 4, matrixLayout);
+}