[Mlir-commits] [mlir] [mlir][spirv] Add bfloat16 support (PR #141458)

Jakub Kuderski llvmlistbot at llvm.org
Thu Jun 5 07:24:40 PDT 2025


================
@@ -532,7 +538,7 @@ void ScalarType::getExtensions(SPIRVType::ExtensionArrayRefVector &extensions,
     [[fallthrough]];
   case StorageClass::Input:
   case StorageClass::Output:
-    if (getIntOrFloatBitWidth() == 16) {
+    if (getIntOrFloatBitWidth() == 16 && !llvm::isa<BFloat16Type>(*this)) {
----------------
kuhar wrote:

nit: no need to prefix cast functions with `llvm::`. The existing code was automatically converted to the new casting style via clang tidy that added this namespace prefixes.

https://github.com/llvm/llvm-project/pull/141458


More information about the Mlir-commits mailing list