[Mlir-commits] [mlir] a6eb406 - [mlir][gpu] Increase default SM version from 35 to 50

[Guray Ozen]

Author: Guray Ozen
Date: 2023-07-24T15:11:30+02:00
New Revision: a6eb40692c795a9cc29266779ceca2e304141114

URL: https://github.com/llvm/llvm-project/commit/a6eb40692c795a9cc29266779ceca2e304141114
DIFF: https://github.com/llvm/llvm-project/commit/a6eb40692c795a9cc29266779ceca2e304141114.diff

LOG: [mlir][gpu] Increase default SM version from 35 to 50

Current SM version is 35 but it is deprecated long time ago. D155563 introduced ptxas compilations, using sm_35 causes failures in builtbot. This change increase default SM version to 50.

Differential Revision: https://reviews.llvm.org/D156098

Added: 
    

Modified: 
    mlir/lib/Dialect/GPU/Transforms/SerializeToCubin.cpp
    mlir/test/lib/Dialect/GPU/TestConvertGPUKernelToCubin.cpp

Removed: 
    


################################################################################
diff  --git a/mlir/lib/Dialect/GPU/Transforms/SerializeToCubin.cpp b/mlir/lib/Dialect/GPU/Transforms/SerializeToCubin.cpp
index ec0625c65d9f44..b61b8a9a6fae88 100644
--- a/mlir/lib/Dialect/GPU/Transforms/SerializeToCubin.cpp
+++ b/mlir/lib/Dialect/GPU/Transforms/SerializeToCubin.cpp
@@ -160,7 +160,7 @@ class SerializeToCubinPass
   MLIR_DEFINE_EXPLICIT_INTERNAL_INLINE_TYPE_ID(SerializeToCubinPass)
 
   SerializeToCubinPass(StringRef triple = "nvptx64-nvidia-cuda",
-                       StringRef chip = "sm_35", StringRef features = "+ptx60",
+                       StringRef chip = "sm_50", StringRef features = "+ptx60",
                        int optLevel = 2, bool dumpPtx = false,
                        bool usePtxas = true, StringRef ptxasParams = {});
 

diff  --git a/mlir/test/lib/Dialect/GPU/TestConvertGPUKernelToCubin.cpp b/mlir/test/lib/Dialect/GPU/TestConvertGPUKernelToCubin.cpp
index 1c442b0147c8b3..a056ed78ed171a 100644
--- a/mlir/test/lib/Dialect/GPU/TestConvertGPUKernelToCubin.cpp
+++ b/mlir/test/lib/Dialect/GPU/TestConvertGPUKernelToCubin.cpp
@@ -38,7 +38,7 @@ class TestSerializeToCubinPass
 
 TestSerializeToCubinPass::TestSerializeToCubinPass() {
   this->triple = "nvptx64-nvidia-cuda";
-  this->chip = "sm_35";
+  this->chip = "sm_50";
   this->features = "+ptx60";
 }