[Mlir-commits] [mlir] [MLIR][NVVM] Add support for st.bulk Op (PR #131727)

[Srinivasa Ravi]

================
@@ -125,6 +125,14 @@ llvm.func @convert_float_to_tf32_no_rnd_mode(%src : f32) -> i32 {
 
 // -----
 
+llvm.func @nvvm_st_bulk_initval_nonzero(%addr : !llvm.ptr, %size : i64) {
+  // expected-error @below {{only 0 is supported for initVal in st.bulk}}
+  nvvm.st.bulk %addr, %size, 1 : !llvm.ptr
----------------
Wolfram70 wrote:

Oh yes, that makes it more readable. I have changed the assembly format to be more explicit in the latest revision. Thanks!

https://github.com/llvm/llvm-project/pull/131727