[Triton-MLIR][FRONTEND] [BACKEND] fix atomics (#879)
minor fix to backend and frontend of atomics, we can pass 1 test without mask and the shape aligned with CTA size now Co-authored-by: dongdongl <dongdongl@nvidia.com>
This commit is contained in:
@@ -1086,9 +1086,7 @@ void init_triton_ir(py::module &&m) {
|
||||
mlir::Value &ptr, mlir::Value &val,
|
||||
mlir::Value &mask) -> mlir::Value {
|
||||
auto loc = self.getUnknownLoc();
|
||||
auto ptrType = mlir::getElementTypeOrSelf(ptr)
|
||||
.cast<mlir::triton::PointerType>();
|
||||
mlir::Type dstType = ptrType.getPointeeType();
|
||||
mlir::Type dstType = val.getType();
|
||||
return self.create<mlir::triton::AtomicRMWOp>(loc, dstType, rmwOp,
|
||||
ptr, val, mask);
|
||||
})
|
||||
|
Reference in New Issue
Block a user