[llvm-branch-commits] [mlir] [mlir][GPU] Plumb range information through the NVVM lowterings (PR #107659)

[Krzysztof Drewniak via llvm-branch-commits]

================
@@ -699,9 +699,21 @@ gpu.module @test_module_32 {
 }
 
 gpu.module @test_module_33 {
-// CHECK-LABEL: func @kernel_with_block_size()
-// CHECK: attributes {gpu.kernel, gpu.known_block_size = array<i32: 128, 1, 1>, nvvm.kernel, nvvm.maxntid = array<i32: 128, 1, 1>}
-  gpu.func @kernel_with_block_size() kernel attributes {known_block_size = array<i32: 128, 1, 1>} {
+// CHECK-LABEL: func @kernel_with_block_size(
+// CHECK: attributes {gpu.kernel, gpu.known_block_size = array<i32: 32, 4, 2>, nvvm.kernel, nvvm.maxntid = array<i32: 32, 4, 2>}
+  gpu.func @kernel_with_block_size(%arg0: !llvm.ptr) kernel attributes {known_block_size = array<i32: 32, 4, 2>} {
----------------
krzysz00 wrote:

Yep, that code works generally

https://github.com/llvm/llvm-project/pull/107659