[Scheduled] Update Triton Commit Hash to 972d18aa0905dfb403091bd4c5746318d097df4d

backend/compiler.py

@@ -317,6 +317,7 @@ def add_stages(self, stages, options, language):
             stages["ttir"] = lambda src, metadata: self.make_ttir(src, metadata, options)
             stages["ttgir"] = lambda src, metadata: self.make_ttgir(src, metadata, options)
         elif language == Language.GLUON:
+            stages["glir"] = lambda src, metadata: src
             stages["ttgir"] = lambda src, metadata: self.gluon_to_ttgir(src, metadata, options)
         stages["llir"] = lambda src, metadata: self.make_llir(src, metadata, options)
         stages["asm"] = lambda src, metadata: self.make_asm(src, metadata, options)

cpu/lib/TritonCPUToLLVM/LowerDotMicrokernelToSME.cpp

@@ -421,9 +421,7 @@ static void emitContiguousStore(OpBuilder &rewriter, Location loc, Block *body,
       /*colOffset=*/body->getArgument(1), // col (M or N) offset
       /*rowStride=*/rowStride, /*colStride=*/1,
       cast<RankedTensorType>(value.getType()));
-  triton::StoreOp::create(rewriter, loc, ptrs, value,
-                          triton::CacheModifier::NONE,
-                          triton::EvictionPolicy::NORMAL);
+  triton::StoreOp::create(rewriter, loc, ptrs, value);
 }
 
 void rewriteExistingFunctionBody(DotDescriptor &desc, triton::FuncOp funcOp,
@@ -532,11 +530,9 @@ void rewriteExistingFunctionBody(DotDescriptor &desc, triton::FuncOp funcOp,
     Value cPtrs = buildContiguousPtrTensor(
         rewriter, loc, cArg, /*rowOffset=*/mOff, /*colOffset=*/nOff,
         /*rowStride=*/desc.blockN, /*colStride=*/1, cTileTy);
-    Value cTile = triton::LoadOp::create(rewriter, loc, cPtrs,
-                                         triton::CacheModifier::NONE,
-                                         triton::EvictionPolicy::NORMAL,
-                                         /*isVolatile=*/false)
-                      .getResult();
+    Value cTile =
+        triton::LoadOp::create(rewriter, loc, cPtrs, /*isVolatile=*/false)
+            .getResult();
     cpu::YieldOp::create(rewriter, loc, ValueRange{cTile});
   }
 

test/filtered_test_list.txt

@@ -6,6 +6,7 @@ test_atomic_cas
 test_tensor_atomic_cas
 test_tensor_atomic_use_result
 test_atomic_min_max_neg_zero
+test_atomic_load_store
 test_sum_dtype
 test_no_rematerialization_op
 test_abs_fp8

triton.txt

@@ -1 +1 @@
-20e6ba864830b88f56b907fef85cab5097cf3892
+972d18aa0905dfb403091bd4c5746318d097df4d