[Mlir-commits] [mlir] [mlir][nvgpu] add `nvgpu.rcp` op (PR #100965)

llvmlistbot at llvm.org llvmlistbot at llvm.org
Mon Jul 29 02:05:13 PDT 2024


https://github.com/Observer007 updated https://github.com/llvm/llvm-project/pull/100965

>From 0749a595895feb7507baf3a9a2df093f7284849e Mon Sep 17 00:00:00 2001
From: jingzec <jingzec at nvidia.com>
Date: Fri, 26 Jul 2024 00:48:01 -0700
Subject: [PATCH 1/4] add rcp approxe op

---
 mlir/include/mlir/Dialect/NVGPU/IR/NVGPU.td   | 13 +++++++
 .../Conversion/NVGPUToNVVM/NVGPUToNVVM.cpp    | 38 ++++++++++++++++++-
 .../Conversion/NVGPUToNVVM/nvgpu-to-nvvm.mlir | 16 ++++++++
 3 files changed, 66 insertions(+), 1 deletion(-)

diff --git a/mlir/include/mlir/Dialect/NVGPU/IR/NVGPU.td b/mlir/include/mlir/Dialect/NVGPU/IR/NVGPU.td
index dda8f31e688fe..3501c5af16d8c 100644
--- a/mlir/include/mlir/Dialect/NVGPU/IR/NVGPU.td
+++ b/mlir/include/mlir/Dialect/NVGPU/IR/NVGPU.td
@@ -20,6 +20,7 @@
 #ifndef NVGPU
 #define NVGPU
 
+include "mlir/Interfaces/InferTypeOpInterface.td"
 include "mlir/Interfaces/SideEffectInterfaces.td"
 include "mlir/IR/AttrTypeBase.td"
 include "mlir/IR/OpBase.td"
@@ -802,4 +803,16 @@ def NVGPU_WarpgroupMmaInitAccumulatorOp : NVGPU_Op<"warpgroup.mma.init.accumulat
   let hasVerifier = 1;
 }
 
+def NVGPU_RcpApproxOp : NVGPU_Op<"rcp_approx", [
+  Pure, SameOperandsAndResultType
+]> {
+  let summary = "A wrapper of nvvm rcp.approx.ftz.f";
+  let description = [{
+    F32 vector reciprocal calculation using `nvvm.rcp.approx.ftz.f`.
+    The input and output are both F32 vector with same shape.
+  }];
+  let arguments = (ins VectorOf<[F32]>:$in);
+  let results = (outs VectorOf<[F32]>:$out);
+  let assemblyFormat = "$in attr-dict `:` type($out)";
+}
 #endif // NVGPU
diff --git a/mlir/lib/Conversion/NVGPUToNVVM/NVGPUToNVVM.cpp b/mlir/lib/Conversion/NVGPUToNVVM/NVGPUToNVVM.cpp
index 11d29754aa760..b7ea4aaeb7d8c 100644
--- a/mlir/lib/Conversion/NVGPUToNVVM/NVGPUToNVVM.cpp
+++ b/mlir/lib/Conversion/NVGPUToNVVM/NVGPUToNVVM.cpp
@@ -11,6 +11,7 @@
 #include "mlir/Conversion/GPUCommon/GPUCommonPass.h"
 #include "mlir/Conversion/LLVMCommon/ConversionTarget.h"
 #include "mlir/Conversion/LLVMCommon/Pattern.h"
+#include "mlir/Conversion/LLVMCommon/VectorPattern.h"
 #include "mlir/Dialect/Arith/IR/Arith.h"
 #include "mlir/Dialect/GPU/IR/GPUDialect.h"
 #include "mlir/Dialect/LLVMIR/LLVMDialect.h"
@@ -1666,6 +1667,41 @@ struct NVGPUTmaPrefetchOpLowering
   }
 };
 
+struct NVGPURcpApproxOpLowering
+    : public ConvertOpToLLVMPattern<nvgpu::RcpApproxOp> {
+  using ConvertOpToLLVMPattern<nvgpu::RcpApproxOp>::ConvertOpToLLVMPattern;
+  LogicalResult
+  matchAndRewrite(nvgpu::RcpApproxOp op, OpAdaptor adaptor,
+                  ConversionPatternRewriter &rewriter) const override {
+    ImplicitLocOpBuilder b(op->getLoc(), rewriter);
+    auto i64Ty = b.getI64Type();
+    auto f32Ty = b.getF32Type();
+    VectorType inTy = op.getIn().getType();
+    // apply rcp.approx.ftz.f on each element in vector.
+    auto convert1DVec = [&](Type llvm1DVectorTy, Value inVec) {
+      Value ret1DVec = b.create<LLVM::UndefOp>(llvm1DVectorTy);
+      int numElems = llvm::cast<VectorType>(llvm1DVectorTy).getNumElements();
+      for (int i = 0; i < numElems; i++) {
+        Value idx = b.create<LLVM::ConstantOp>(i64Ty, b.getI64IntegerAttr(i));
+        Value elem = b.create<LLVM::ExtractElementOp>(inVec, idx);
+        Value dst = b.create<NVVM::RcpApproxFtzF32Op>(f32Ty, elem);
+        ret1DVec = b.create<LLVM::InsertElementOp>(ret1DVec, dst, idx);
+      }
+      return ret1DVec;
+    };
+    if (inTy.getRank() == 1) {
+      rewriter.replaceOp(op, convert1DVec(inTy, adaptor.getIn()));
+      return success();
+    }
+    return LLVM::detail::handleMultidimensionalVectors(
+        op.getOperation(), adaptor.getOperands(), *(this->getTypeConverter()),
+        [&](Type llvm1DVectorTy, ValueRange operands) -> Value {
+          OpAdaptor adaptor(operands);
+          return convert1DVec(llvm1DVectorTy, adaptor.getIn());
+        },
+        rewriter);
+  }
+};
 } // namespace
 
 void mlir::populateNVGPUToNVVMConversionPatterns(LLVMTypeConverter &converter,
@@ -1688,5 +1724,5 @@ void mlir::populateNVGPUToNVVMConversionPatterns(LLVMTypeConverter &converter,
       NVGPUWarpgroupMmaInitAccumulatorOpLowering, // nvgpu.warpgroup.mma.init.accumulator
       MmaSyncOptoNVVM, MmaLdMatrixOpToNVVM, NVGPUAsyncCopyLowering,
       NVGPUAsyncCreateGroupLowering, NVGPUAsyncWaitLowering,
-      NVGPUMmaSparseSyncLowering>(converter);
+      NVGPUMmaSparseSyncLowering, NVGPURcpApproxOpLowering>(converter);
 }
diff --git a/mlir/test/Conversion/NVGPUToNVVM/nvgpu-to-nvvm.mlir b/mlir/test/Conversion/NVGPUToNVVM/nvgpu-to-nvvm.mlir
index 86a552c03a473..95cd7a0892b6c 100644
--- a/mlir/test/Conversion/NVGPUToNVVM/nvgpu-to-nvvm.mlir
+++ b/mlir/test/Conversion/NVGPUToNVVM/nvgpu-to-nvvm.mlir
@@ -1339,3 +1339,19 @@ module attributes {transform.with_named_sequence} {
     transform.yield
   }
 }
+
+// CHECK-LABEL: @rcp_approx_ftz_f32
+// CHECK-SAME:  %[[IN:.*]]: vector<32x16xf32>
+func.func @rcp_approx_ftz_f32(
+    %in: vector<32x16xf32>) {
+  // CHECK: %[[IN_LLVM:.*]] = builtin.unrealized_conversion_cast %[[IN]] : vector<32x16xf32> to !llvm.array<32 x vector<16xf32>>
+  // CHECK: %[[IN1DVEC:.*]] = llvm.extractvalue %[[IN_LLVM]][0] : !llvm.array<32 x vector<16xf32>>
+  // CHECK: %[[OUT1DVEC:.*]] = llvm.mlir.undef : vector<16xf32>
+  // CHECK: %[[IDX_0:.+]] = llvm.mlir.constant(0 : i64) : i64
+  // CHECK: %[[ELEM_0:.*]] = llvm.extractelement %[[IN1DVEC]][%[[IDX_0]] : i64]
+  // CHECK: %[[ELEM_RCP0:.*]] = nvvm.rcp.approx.ftz.f %[[ELEM_0]] : f32
+  // CHECK: llvm.insertelement %[[ELEM_RCP0]], %[[OUT1DVEC]][%[[IDX_0]] : i64] : vector<16xf32>
+  // CHECK-COUNT-511: nvvm.rcp.approx.ftz.f
+  %out = nvgpu.rcp_approx %in : vector<32x16xf32>
+  return
+}

>From 54d91e81723a4afa27f1267f63bee5f343cbf777 Mon Sep 17 00:00:00 2001
From: jingzec <jingzec at nvidia.com>
Date: Sun, 28 Jul 2024 21:12:31 -0700
Subject: [PATCH 2/4] fix

---
 mlir/include/mlir/Dialect/NVGPU/IR/NVGPUDialect.h | 1 +
 1 file changed, 1 insertion(+)

diff --git a/mlir/include/mlir/Dialect/NVGPU/IR/NVGPUDialect.h b/mlir/include/mlir/Dialect/NVGPU/IR/NVGPUDialect.h
index 19070f6f062a0..aad2ac6f4dd2b 100644
--- a/mlir/include/mlir/Dialect/NVGPU/IR/NVGPUDialect.h
+++ b/mlir/include/mlir/Dialect/NVGPU/IR/NVGPUDialect.h
@@ -17,6 +17,7 @@
 #include "mlir/IR/BuiltinTypes.h"
 #include "mlir/IR/Dialect.h"
 #include "mlir/IR/OpDefinition.h"
+#include "mlir/Interfaces/InferTypeOpInterface.h"
 #include "mlir/Interfaces/SideEffectInterfaces.h"
 
 #include "mlir/Dialect/NVGPU/IR/NVGPUEnums.h.inc"

>From 9d1e603b27810a60b313eb3fbf5c0cdb3e1988f2 Mon Sep 17 00:00:00 2001
From: jingzec <jingzec at nvidia.com>
Date: Mon, 29 Jul 2024 01:58:38 -0700
Subject: [PATCH 3/4] fix comment

---
 mlir/include/mlir/Dialect/NVGPU/IR/NVGPU.td   | 37 +++++++++++++++----
 .../mlir/Dialect/NVGPU/IR/NVGPUDialect.h      |  1 -
 .../Conversion/NVGPUToNVVM/NVGPUToNVVM.cpp    |  9 ++---
 mlir/lib/Dialect/NVGPU/IR/NVGPUDialect.cpp    | 17 ++++++++-
 .../Conversion/NVGPUToNVVM/nvgpu-to-nvvm.mlir |  2 +-
 mlir/test/Dialect/NVGPU/invalid.mlir          | 18 +++++++++
 6 files changed, 68 insertions(+), 16 deletions(-)

diff --git a/mlir/include/mlir/Dialect/NVGPU/IR/NVGPU.td b/mlir/include/mlir/Dialect/NVGPU/IR/NVGPU.td
index 3501c5af16d8c..8e0e5c0501c79 100644
--- a/mlir/include/mlir/Dialect/NVGPU/IR/NVGPU.td
+++ b/mlir/include/mlir/Dialect/NVGPU/IR/NVGPU.td
@@ -20,7 +20,6 @@
 #ifndef NVGPU
 #define NVGPU
 
-include "mlir/Interfaces/InferTypeOpInterface.td"
 include "mlir/Interfaces/SideEffectInterfaces.td"
 include "mlir/IR/AttrTypeBase.td"
 include "mlir/IR/OpBase.td"
@@ -110,10 +109,24 @@ def TensorMapInterleaveKind : I32EnumAttr<"TensorMapInterleaveKind",
   let cppNamespace = "::mlir::nvgpu";
 }
 
+def RcpApprox : I32EnumAttrCase<"APPROX", 0, "approx">;
+def RcpRN     : I32EnumAttrCase<"RN", 1, "rn">;
+def RcpRZ     : I32EnumAttrCase<"RZ", 2, "rz">;
+def RcpRM     : I32EnumAttrCase<"RM", 3, "rm">;
+def RcpRP     : I32EnumAttrCase<"RP", 4, "rp">;
+def RcpRoundingMode   : I32EnumAttr<"RcpRoundingMode", "Rounding mode of rcp",
+  [RcpApprox, RcpRN, RcpRZ, RcpRM, RcpRP]> {
+  let genSpecializedAttr = 0;
+  let cppNamespace = "::mlir::nvgpu";
+}
+
 def TensorMapSwizzleAttr : EnumAttr<NVGPU_Dialect, TensorMapSwizzleKind, "swizzle">;
 def TensorMapL2PromoAttr : EnumAttr<NVGPU_Dialect, TensorMapL2PromoKind, "l2promo">;
 def TensorMapOOBAttr : EnumAttr<NVGPU_Dialect, TensorMapOOBKind, "oob">;
 def TensorMapInterleaveAttr : EnumAttr<NVGPU_Dialect, TensorMapInterleaveKind, "interleave">;
+def RcpRoundingModeAttr : EnumAttr<NVGPU_Dialect, RcpRoundingMode, "rcp_rounding_mode">;
+
+// rcp mode attribute
 
 //===----------------------------------------------------------------------===//
 // NVGPU Type Definitions
@@ -803,16 +816,24 @@ def NVGPU_WarpgroupMmaInitAccumulatorOp : NVGPU_Op<"warpgroup.mma.init.accumulat
   let hasVerifier = 1;
 }
 
-def NVGPU_RcpApproxOp : NVGPU_Op<"rcp_approx", [
-  Pure, SameOperandsAndResultType
-]> {
-  let summary = "A wrapper of nvvm rcp.approx.ftz.f";
+def NVGPU_RcpOp : NVGPU_Op<"rcp", [Pure,
+                                   SameOperandsAndResultShape,
+                                   PredOpTrait<"in and out have same element type",
+                                            TCresVTEtIsSameAsOp<0, 0>>]> {
+  let summary = "F32 rcp calculation for vector types using nvvm.rcp.* OP";
   let description = [{
-    F32 vector reciprocal calculation using `nvvm.rcp.approx.ftz.f`.
+    F32 vector reciprocal calculation using `nvvm.rcp.*`. Currently only
+    `nvvm.approx.ftz.f` is supported.
     The input and output are both F32 vector with same shape.
   }];
-  let arguments = (ins VectorOf<[F32]>:$in);
+  let arguments = (ins VectorOf<[F32]>:$in,
+                       DefaultValuedAttr<RcpRoundingModeAttr, "RcpRoundingMode::APPROX">:$rounding,
+                       UnitAttr:$ftz);
   let results = (outs VectorOf<[F32]>:$out);
-  let assemblyFormat = "$in attr-dict `:` type($out)";
+  let assemblyFormat = [{
+    $in `{` `rounding` `=` $rounding (`,` `ftz` $ftz^)? `}` 
+    attr-dict `:` type($in) `->` type($out)
+  }];
+  let hasVerifier = 1;
 }
 #endif // NVGPU
diff --git a/mlir/include/mlir/Dialect/NVGPU/IR/NVGPUDialect.h b/mlir/include/mlir/Dialect/NVGPU/IR/NVGPUDialect.h
index aad2ac6f4dd2b..19070f6f062a0 100644
--- a/mlir/include/mlir/Dialect/NVGPU/IR/NVGPUDialect.h
+++ b/mlir/include/mlir/Dialect/NVGPU/IR/NVGPUDialect.h
@@ -17,7 +17,6 @@
 #include "mlir/IR/BuiltinTypes.h"
 #include "mlir/IR/Dialect.h"
 #include "mlir/IR/OpDefinition.h"
-#include "mlir/Interfaces/InferTypeOpInterface.h"
 #include "mlir/Interfaces/SideEffectInterfaces.h"
 
 #include "mlir/Dialect/NVGPU/IR/NVGPUEnums.h.inc"
diff --git a/mlir/lib/Conversion/NVGPUToNVVM/NVGPUToNVVM.cpp b/mlir/lib/Conversion/NVGPUToNVVM/NVGPUToNVVM.cpp
index b7ea4aaeb7d8c..cf984ca3293c0 100644
--- a/mlir/lib/Conversion/NVGPUToNVVM/NVGPUToNVVM.cpp
+++ b/mlir/lib/Conversion/NVGPUToNVVM/NVGPUToNVVM.cpp
@@ -1667,11 +1667,10 @@ struct NVGPUTmaPrefetchOpLowering
   }
 };
 
-struct NVGPURcpApproxOpLowering
-    : public ConvertOpToLLVMPattern<nvgpu::RcpApproxOp> {
-  using ConvertOpToLLVMPattern<nvgpu::RcpApproxOp>::ConvertOpToLLVMPattern;
+struct NVGPURcpOpLowering : public ConvertOpToLLVMPattern<nvgpu::RcpOp> {
+  using ConvertOpToLLVMPattern<nvgpu::RcpOp>::ConvertOpToLLVMPattern;
   LogicalResult
-  matchAndRewrite(nvgpu::RcpApproxOp op, OpAdaptor adaptor,
+  matchAndRewrite(nvgpu::RcpOp op, OpAdaptor adaptor,
                   ConversionPatternRewriter &rewriter) const override {
     ImplicitLocOpBuilder b(op->getLoc(), rewriter);
     auto i64Ty = b.getI64Type();
@@ -1724,5 +1723,5 @@ void mlir::populateNVGPUToNVVMConversionPatterns(LLVMTypeConverter &converter,
       NVGPUWarpgroupMmaInitAccumulatorOpLowering, // nvgpu.warpgroup.mma.init.accumulator
       MmaSyncOptoNVVM, MmaLdMatrixOpToNVVM, NVGPUAsyncCopyLowering,
       NVGPUAsyncCreateGroupLowering, NVGPUAsyncWaitLowering,
-      NVGPUMmaSparseSyncLowering, NVGPURcpApproxOpLowering>(converter);
+      NVGPUMmaSparseSyncLowering, NVGPURcpOpLowering>(converter);
 }
diff --git a/mlir/lib/Dialect/NVGPU/IR/NVGPUDialect.cpp b/mlir/lib/Dialect/NVGPU/IR/NVGPUDialect.cpp
index 26f831f10a4e4..a320e893b59c7 100644
--- a/mlir/lib/Dialect/NVGPU/IR/NVGPUDialect.cpp
+++ b/mlir/lib/Dialect/NVGPU/IR/NVGPUDialect.cpp
@@ -10,9 +10,9 @@
 //
 //===----------------------------------------------------------------------===//
 
-#include "mlir/Dialect/NVGPU/IR/NVGPUDialect.h"
 #include "mlir/Dialect/GPU/IR/GPUDialect.h"
 #include "mlir/Dialect/LLVMIR/LLVMTypes.h"
+#include "mlir/Dialect/NVGPU/IR/NVGPUDialect.h"
 #include "mlir/IR/Builders.h"
 #include "mlir/IR/BuiltinAttributes.h"
 #include "mlir/IR/BuiltinTypes.h"
@@ -644,6 +644,21 @@ LogicalResult WarpgroupMmaInitAccumulatorOp::verify() {
   return success();
 }
 
+//===----------------------------------------------------------------------===//
+// RcpOp
+//===----------------------------------------------------------------------===//
+
+LogicalResult RcpOp::verify() {
+  RcpRoundingModeAttr rounding = getRoundingAttr();
+  bool ftz = getFtz();
+  // Currently, only `rcp_approx` and `ftz` is supported.
+  if (rounding.getValue() != RcpRoundingMode::APPROX || !ftz) {
+    return emitOpError() << "has a limitation. " << rounding
+                         << " or non-ftz is not supported yet.";
+  }
+  return success();
+}
+
 //===----------------------------------------------------------------------===//
 // TableGen'd dialect, type, and op definitions
 //===----------------------------------------------------------------------===//
diff --git a/mlir/test/Conversion/NVGPUToNVVM/nvgpu-to-nvvm.mlir b/mlir/test/Conversion/NVGPUToNVVM/nvgpu-to-nvvm.mlir
index 95cd7a0892b6c..464e11b79ae75 100644
--- a/mlir/test/Conversion/NVGPUToNVVM/nvgpu-to-nvvm.mlir
+++ b/mlir/test/Conversion/NVGPUToNVVM/nvgpu-to-nvvm.mlir
@@ -1352,6 +1352,6 @@ func.func @rcp_approx_ftz_f32(
   // CHECK: %[[ELEM_RCP0:.*]] = nvvm.rcp.approx.ftz.f %[[ELEM_0]] : f32
   // CHECK: llvm.insertelement %[[ELEM_RCP0]], %[[OUT1DVEC]][%[[IDX_0]] : i64] : vector<16xf32>
   // CHECK-COUNT-511: nvvm.rcp.approx.ftz.f
-  %out = nvgpu.rcp_approx %in : vector<32x16xf32>
+  %out = nvgpu.rcp %in {rounding = approx, ftz} : vector<32x16xf32> -> vector<32x16xf32>
   return
 }
diff --git a/mlir/test/Dialect/NVGPU/invalid.mlir b/mlir/test/Dialect/NVGPU/invalid.mlir
index c3aed35153241..726c245898101 100644
--- a/mlir/test/Dialect/NVGPU/invalid.mlir
+++ b/mlir/test/Dialect/NVGPU/invalid.mlir
@@ -336,3 +336,21 @@ func.func @tma_generate_descriptor_incorrect_last_dim(%desc: !desc,  %buffer2: m
   nvgpu.tma.async.load %desc[%c0, %c0], %mbarrier[%c0] to %buffer2 : !desc, !mbarrier -> memref<64x128xf32,3>
   return
 }
+// -----
+
+func.func @rcp_unsupported_rounding_0(%in : vector<16xf32>) {
+  // expected-error @+1 {{'nvgpu.rcp' op has a limitation. #nvgpu<rcp_rounding_mode rn> or non-ftz is not supported yet.}}
+  %out = nvgpu.rcp %in {rounding = rn, ftz} : vector<16xf32> -> vector<16xf32>
+}
+// -----
+
+func.func @rcp_unsupported_rounding_1(%in : vector<16xf32>) {
+  // expected-error @+1 {{'nvgpu.rcp' op has a limitation. #nvgpu<rcp_rounding_mode rz> or non-ftz is not supported yet.}}
+  %out = nvgpu.rcp %in {rounding = rz} : vector<16xf32> -> vector<16xf32>
+}
+// -----
+
+func.func @rcp_unsupported_ftz(%in : vector<16xf32>) {
+  // expected-error @+1 {{'nvgpu.rcp' op has a limitation. #nvgpu<rcp_rounding_mode approx> or non-ftz is not supported yet.}}
+  %out = nvgpu.rcp %in {rounding = approx} : vector<16xf32> -> vector<16xf32>
+}

>From 9f1829d93f178d3fa9745852e16ddeab7a55c3cc Mon Sep 17 00:00:00 2001
From: jingzec <jingzec at nvidia.com>
Date: Mon, 29 Jul 2024 02:04:34 -0700
Subject: [PATCH 4/4] drop useless

---
 mlir/include/mlir/Dialect/NVGPU/IR/NVGPU.td | 2 --
 1 file changed, 2 deletions(-)

diff --git a/mlir/include/mlir/Dialect/NVGPU/IR/NVGPU.td b/mlir/include/mlir/Dialect/NVGPU/IR/NVGPU.td
index 8e0e5c0501c79..d8caf2818c4f0 100644
--- a/mlir/include/mlir/Dialect/NVGPU/IR/NVGPU.td
+++ b/mlir/include/mlir/Dialect/NVGPU/IR/NVGPU.td
@@ -126,8 +126,6 @@ def TensorMapOOBAttr : EnumAttr<NVGPU_Dialect, TensorMapOOBKind, "oob">;
 def TensorMapInterleaveAttr : EnumAttr<NVGPU_Dialect, TensorMapInterleaveKind, "interleave">;
 def RcpRoundingModeAttr : EnumAttr<NVGPU_Dialect, RcpRoundingMode, "rcp_rounding_mode">;
 
-// rcp mode attribute
-
 //===----------------------------------------------------------------------===//
 // NVGPU Type Definitions
 //===----------------------------------------------------------------------===//



More information about the Mlir-commits mailing list