[Mlir-commits] [mlir] [AMDGPU] Add parameterization for optimized shared memory variables (PR #82508)
llvmlistbot at llvm.org
llvmlistbot at llvm.org
Tue Feb 27 17:13:53 PST 2024
================
@@ -37,10 +37,17 @@ def OptimizeSharedMemory : Pass<"amdgpu-optimize-shared-memory"> {
attempts to optimize reads/writes from a memref representing GPU shared
memory in order to avoid bank conflicts.
}];
-
let dependentDialects = [
"memref::MemRefDialect", "vector::VectorDialect"
];
+ let options = [
+ Option<"sharedMemoryLineSizeBytes", "shared-memory-line-size-bytes", "uint64_t",
----------------
harsh-amd wrote:
Would be nice if we had one source of truth for the pass and transform dialect op, but this is fine for now.
https://github.com/llvm/llvm-project/pull/82508
More information about the Mlir-commits
mailing list