pre-commit

Author: plotfi

lib/Dialect/TritonGPU/Transforms/BF16DotTC.cpp

@@ -39,8 +39,8 @@ auto SplitF32(Value input, unsigned N, PatternRewriter &rewriter)
 
 Value IEEEDot(PatternRewriter &rewriter, Value lhs, Value rhs, Value acc) {
   return rewriter.create<DotOp>(lhs.getLoc(), lhs, rhs, acc,
-                               /*inputPrecision=*/InputPrecision::IEEE,
-                               /*maxNumImpreciseAcc=*/0);
+                                /*inputPrecision=*/InputPrecision::IEEE,
+                                /*maxNumImpreciseAcc=*/0);
 }
 
 auto getBF16Count(triton::InputPrecision precision) -> unsigned {
@@ -58,8 +58,10 @@ auto getBF16Count(triton::InputPrecision precision) -> unsigned {
 }
 
 // Implements 3xBF16 https://arxiv.org/abs/1904.06376
-// See also https://github.com/openxla/xla/blob/e33f93fb7220d408811afdc926cf10baaf49c64e/xla/backends/gpu/codegen/triton/dot_algorithms.cc#L152
-// As well as https://github.com/ROCm/rocm-libraries/blob/develop/projects/hipblaslt/tensilelite/Tensile/Components/LocalRead.py#L288-L330
+// See also
+// https://github.com/openxla/xla/blob/e33f93fb7220d408811afdc926cf10baaf49c64e/xla/backends/gpu/codegen/triton/dot_algorithms.cc#L152
+// As well as
+// https://github.com/ROCm/rocm-libraries/blob/develop/projects/hipblaslt/tensilelite/Tensile/Components/LocalRead.py#L288-L330
 struct BF16xN : public OpRewritePattern<DotOp> {
   using OpRewritePattern::OpRewritePattern;