v1.2.5

+ Add missing test file

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "slangtorch"
-version = "1.2.4"
+version = "1.2.5"
 dependencies = [
   "torch>=1.1.0",
   "hatchling>=1.11.0",

tests/multiply_half.slang

@@ -0,0 +1,24 @@
+static const half kFactor = 2.h;
+
+half computeOutputValue(TensorView<half> A, uint2 loc)
+{
+    return A[loc] * kFactor;
+}
+
+[CudaKernel]
+void mul_kernel(TensorView<half> A, TensorView<half> result)
+{
+    uint2 location = (cudaBlockDim() * cudaBlockIdx() + cudaThreadIdx()).xy;
+    result[location] = computeOutputValue(A, location);
+}
+
+[TorchEntryPoint]
+TorchTensor<half> multiply(TorchTensor<half> A)
+{
+    var result = TorchTensor<half>.zerosLike(A);
+    let blockCount = uint3(1);
+    let groupSize = uint3(A.size(0), A.size(1), 1);
+
+    __dispatch_kernel(mul_kernel, blockCount, groupSize)(A, result);
+    return result;
+}