From 548612eddd9a56c1d9477b2f6c5ed97cee79f3a9 Mon Sep 17 00:00:00 2001
From: Whitney Tsang <whitney.tsang@intel.com>
Date: Thu, 25 Apr 2024 02:30:49 +0000
Subject: [PATCH] Replace GenISA ftof usage with arith::TruncFOp

Signed-off-by: Whitney Tsang <whitney.tsang@intel.com>
---
 .../ElementwiseOpToLLVM.cpp                   | 32 ++-----------------
 1 file changed, 2 insertions(+), 30 deletions(-)

diff --git a/third_party/intel/lib/TritonIntelGPUToLLVM/ElementwiseOpToLLVM.cpp b/third_party/intel/lib/TritonIntelGPUToLLVM/ElementwiseOpToLLVM.cpp
index 67d6da8aef..6d79c9606c 100644
--- a/third_party/intel/lib/TritonIntelGPUToLLVM/ElementwiseOpToLLVM.cpp
+++ b/third_party/intel/lib/TritonIntelGPUToLLVM/ElementwiseOpToLLVM.cpp
@@ -1,5 +1,4 @@
 #include "PatternTritonGPUOpToLLVM.h"
-#include "mlir/Conversion/ArithCommon/AttrToLLVMConverter.h"
 #include "mlir/Dialect/LLVMIR/LLVMDialect.h"
 #include "mlir/IR/MLIRContext.h"
 #include "triton/Conversion/TritonGPUToLLVM/TargetInfoBase.h"
@@ -1330,33 +1329,6 @@ struct FpToFpOpConversion
                                  ConversionPatternRewriter &rewriter,
                                  const Value &v, const RoundingMode rounding) {
     MLIRContext *ctx = rewriter.getContext();
-
-    LLVM::RoundingMode roundingMode;
-    switch (rounding) {
-    case RoundingMode::RTNE:
-      roundingMode = LLVM::RoundingMode::NearestTiesToEven;
-      break;
-    case RoundingMode::RTZ:
-      roundingMode = LLVM::RoundingMode::TowardZero;
-      break;
-    default:
-      llvm::errs() << "WARNING: unsupported rounding mode for f32->f16 "
-                      "conversion: "
-                   << stringifyRoundingMode(rounding) << "\n";
-      llvm_unreachable("");
-    }
-
-    NamedAttrList convertedAttr;
-    convertedAttr.set(LLVM::ConstrainedFPTruncIntr::getRoundingModeAttrName(),
-                      LLVM::RoundingModeAttr::get(ctx, roundingMode));
-    convertedAttr.set(
-        LLVM::ConstrainedFPTruncIntr::getFPExceptionBehaviorAttrName(),
-        arith::getLLVMDefaultFPExceptionBehavior(*ctx));
-    return rewriter.create<LLVM::ConstrainedFPTruncIntr>(loc, f16_ty, v,
-                                                         convertedAttr);
-
-#if 0
-    // FIXME: test_typeconvert_downcast fails when lower to arith::TruncFOp.
     arith::RoundingMode roundingMode;
     switch (rounding) {
     case RoundingMode::RTNE:
@@ -1373,7 +1345,6 @@ struct FpToFpOpConversion
     }
     return rewriter.create<arith::TruncFOp>(
         loc, f16_ty, v, arith::RoundingModeAttr::get(ctx, roundingMode));
-#endif
   }
 
   std::pair<ConverterT, size_t>
@@ -2023,9 +1994,10 @@ struct TruncFOpConversion
       return {// Trunc uses the default rounding mode: RTNE
               FpToFpOpConversion::convertFp32ToBf16(
                   loc, rewriter, operands[0][0], RoundingMode::RTNE)};
-    } else {
+    } else if (!op.getRoundingModeAttr()) {
       return {rewriter.create<LLVM::FPTruncOp>(loc, elemTy, operands[0][0])};
     }
+    return {};
   }
 };