[clang] [llvm] [AMDGPU][Clang] Allow amdgpu-waves-per-eu attribute to lower target occupancy range (PR #138284)

Matt Arsenault via llvm-commits llvm-commits at lists.llvm.org
Fri Jun 27 00:57:50 PDT 2025


================
@@ -46,7 +46,6 @@ __attribute__((amdgpu_num_sgpr(4294967296))) kernel void kernel_num_sgpr_L() {}
 __attribute__((amdgpu_num_vgpr(4294967296))) kernel void kernel_num_vgpr_L() {} // expected-error {{integer constant expression evaluates to value 4294967296 that cannot be represented in a 32-bit unsigned integer type}}
 
 __attribute__((amdgpu_flat_work_group_size(0, 64))) kernel void kernel_flat_work_group_size_0_64() {} // expected-error {{'amdgpu_flat_work_group_size' attribute argument is invalid: max must be 0 since min is 0}}
-__attribute__((amdgpu_waves_per_eu(0, 4))) kernel void kernel_waves_per_eu_0_4() {} // expected-error {{'amdgpu_waves_per_eu' attribute argument is invalid: max must be 0 since min is 0}}
----------------
arsenm wrote:

Missing clang codegen test changes that show the new accepted values. This is still not emitting minimums of 0 though, so this is just losing a test? 


https://github.com/llvm/llvm-project/pull/138284


More information about the llvm-commits mailing list