Skip to content

Conversation

@AlexAUT
Copy link
Contributor

@AlexAUT AlexAUT commented Feb 20, 2025

This PR adds raw.ptr.buffer.load.lds op to ROCDL to expose buffer loads directly to LDS.
The op is converted to the corresponding intrinsic call during the translation from MLIR to LLVM IR.

@github-actions
Copy link

Thank you for submitting a Pull Request (PR) to the LLVM Project!

This PR will be automatically labeled and the relevant teams will be notified.

If you wish to, you can add reviewers by using the "Reviewers" section on this page.

If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using @ followed by their GitHub username.

If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers.

If you have further questions, they may be answered by the LLVM GitHub User Guide.

You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums.

@AlexAUT AlexAUT changed the title [MLIR][ROCDL] Add ops for raw.ptr.buffer.load.lds [MLIR][ROCDL] Add op for raw.ptr.buffer.load.lds Feb 20, 2025
@llvmbot
Copy link
Member

llvmbot commented Feb 20, 2025

@llvm/pr-subscribers-mlir

@llvm/pr-subscribers-mlir-llvm

Author: Alexander Weinrauch (AlexAUT)

Changes

This PR adds raw.ptr.buffer.load.lds op to ROCDL to expose buffer loads directly to LDS.
The op is converted to the corresponding intrinsic call during the translation from MLIR to LLVM IR.


Full diff: https://github.com/llvm/llvm-project/pull/127988.diff

3 Files Affected:

  • (modified) mlir/include/mlir/Dialect/LLVMIR/ROCDLOps.td (+19-1)
  • (modified) mlir/test/Dialect/LLVMIR/rocdl.mlir (+9)
  • (modified) mlir/test/Target/LLVMIR/rocdl.mlir (+12)
diff --git a/mlir/include/mlir/Dialect/LLVMIR/ROCDLOps.td b/mlir/include/mlir/Dialect/LLVMIR/ROCDLOps.td
index 7efa4ffa2aa6f..bdb7469433002 100644
--- a/mlir/include/mlir/Dialect/LLVMIR/ROCDLOps.td
+++ b/mlir/include/mlir/Dialect/LLVMIR/ROCDLOps.td
@@ -475,6 +475,24 @@ def ROCDL_RawPtrBufferLoadOp :
   }];
 }
 
+def ROCDL_RawPtrBufferLoadLdsOp :
+  ROCDL_IntrOp<"raw.ptr.buffer.load.lds", [], [], [], 0, 0, 1> {
+  dag args = (ins Arg<ROCDLBufferRsrc, "", [MemRead]>:$rsrc,
+                  Arg<ROCDLBufferLDS, "", [MemWrite]>:$ldsPtr,
+                  I32:$size,
+                  I32:$voffset,
+                  I32:$soffset,
+                  I32:$offset,
+                  I32:$aux);
+  let arguments = !con(args, aliasAttrs);
+  let assemblyFormat = "operands attr-dict";
+  let extraClassDefinition = [{
+    ::llvm::SmallVector<::mlir::Value> $cppClass::getAccessedOperands() {
+      return {getRsrc(), getLdsPtr()};
+    }
+  }];
+}
+
 def ROCDL_RawPtrBufferStoreOp :
   ROCDL_IntrOp<"raw.ptr.buffer.store", [], [0], [], 0, 0, 1> {
   dag args = (ins LLVM_Type:$vdata,
@@ -692,7 +710,7 @@ def ROCDL_CvtScaleF32PkFp8F32:
     attr-dict $srcA `,` $srcB `,` $scale `->` $old `[` $wordSel `]` `:` type($res)
   }];
 }
-    
+
 def ROCDL_CvtScaleF32PkBf8F32:
     ROCDL_IntrOp<"cvt.scalef32.pk.bf8.f32", [], [], [Pure], 1>,
     Arguments<(ins ROCDL_V2I16Type:$old, F32:$srcA, F32:$srcB, F32: $scale, I1:$wordSel)> {
diff --git a/mlir/test/Dialect/LLVMIR/rocdl.mlir b/mlir/test/Dialect/LLVMIR/rocdl.mlir
index 73e844b5f3e13..05469b64a8083 100644
--- a/mlir/test/Dialect/LLVMIR/rocdl.mlir
+++ b/mlir/test/Dialect/LLVMIR/rocdl.mlir
@@ -687,6 +687,15 @@ llvm.func @rocdl.raw.ptr.buffer.f32(%rsrc : !llvm.ptr<8>,
   llvm.return
 }
 
+llvm.func @rocdl.raw.ptr.buffer.load.lds(%rsrc : !llvm.ptr<8>, %dstLds : !llvm.ptr<3>,
+                       %size: i32, %voffset : i32, %soffset : i32, %offset : i32,
+                       %aux : i32) {
+  // CHECK-LABEL: rocdl.raw.ptr.buffer.load.lds
+  // CHECK: rocdl.raw.ptr.buffer.load.lds %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}}
+  rocdl.raw.ptr.buffer.load.lds %rsrc, %dstLds, %size, %voffset, %soffset, %offset, %aux
+
+  llvm.return
+}
 
 llvm.func @rocdl.raw.ptr.buffer.i32(%rsrc : !llvm.ptr<8>,
                        %offset : i32, %soffset : i32,
diff --git a/mlir/test/Target/LLVMIR/rocdl.mlir b/mlir/test/Target/LLVMIR/rocdl.mlir
index eac28c57e2ab4..196a15550e513 100644
--- a/mlir/test/Target/LLVMIR/rocdl.mlir
+++ b/mlir/test/Target/LLVMIR/rocdl.mlir
@@ -893,6 +893,18 @@ llvm.func @rocdl.raw.ptr.buffer(%rsrc : !llvm.ptr<8>,
   llvm.return
 }
 
+llvm.func @rocdl.raw.ptr.buffer.load.lds(%rsrc : !llvm.ptr<8>, %dstLds : !llvm.ptr<3>,
+                        %voffset : i32, %soffset : i32) {
+  %size = llvm.mlir.constant(4 : i32) : i32
+  %offset = llvm.mlir.constant(128 : i32) : i32
+  %aux = llvm.mlir.constant(1 : i32) : i32
+  // CHECK-LABEL: rocdl.raw.ptr.buffer.load.lds
+  // CHECK: call void @llvm.amdgcn.raw.ptr.buffer.load.lds(ptr addrspace(8) %{{.*}}, ptr addrspace(3) %{{.*}}, i32 4, i32 %{{.*}}, i32 %{{.*}}, i32 128, i32 1
+  rocdl.raw.ptr.buffer.load.lds %rsrc, %dstLds, %size, %voffset, %soffset, %offset, %aux
+
+  llvm.return
+}
+
 llvm.func @rocdl.raw.ptr.buffer.atomic.f32(%rsrc : !llvm.ptr<8>,
                         %offset : i32, %soffset : i32,
                         %vdata1 : f32) {

Copy link
Contributor

@krzysz00 krzysz00 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Approved

@jungpark-mlir jungpark-mlir merged commit fc2f31c into llvm:main Feb 21, 2025
11 checks passed
@github-actions
Copy link

@AlexAUT Congratulations on having your first Pull Request (PR) merged into the LLVM Project!

Your changes will be combined with recent changes from other authors, then tested by our build bots. If there is a problem with a build, you may receive a report in an email or a comment on this PR.

Please check whether problems have been caused by your change specifically, as the builds can include changes from many authors. It is not uncommon for your change to be included in a build that fails due to someone else's changes, or infrastructure issues.

How to do this, and the rest of the post-merge process, is covered in detail here.

If your change does cause a problem, it may be reverted, or you can revert it yourself. This is a normal part of LLVM development. You can fix your changes and open a new PR to merge them again.

If you don't get any reports, no action is required from you. Your changes are working as expected, well done!

antiagainst pushed a commit to triton-lang/triton that referenced this pull request Feb 21, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants