[Mlir-commits] [mlir] [MLIR][NVVM] Add `nvvm.cos` OP (PR #193792)

[Guray Ozen]

https://github.com/grypp updated https://github.com/llvm/llvm-project/pull/193792

>From aecb4ac0361967253d8a2f305063ce4223e3603d Mon Sep 17 00:00:00 2001
From: Guray Ozen <gozen at nvidia.com>
Date: Fri, 24 Apr 2026 12:38:52 +0200
Subject: [PATCH] [MLIR][NVVM] Add `nvvm.cos` OP

Implement `nvvm.cos` with ftz flag
---
 mlir/include/mlir/Dialect/LLVMIR/NVVMOps.td       | 15 +++++++++++++++
 mlir/lib/Dialect/LLVMIR/IR/NVVMDialect.cpp        | 10 ++++++++++
 .../test/Dialect/LLVMIR/nvvm-transcendentals.mlir | 15 +++++++++++++++
 mlir/test/Target/LLVMIR/nvvm/transcendentals.mlir | 14 ++++++++++++++
 4 files changed, 54 insertions(+)

diff --git a/mlir/include/mlir/Dialect/LLVMIR/NVVMOps.td b/mlir/include/mlir/Dialect/LLVMIR/NVVMOps.td
index 9e94477c3a60a..0a6c55d68a5e1 100644
--- a/mlir/include/mlir/Dialect/LLVMIR/NVVMOps.td
+++ b/mlir/include/mlir/Dialect/LLVMIR/NVVMOps.td
@@ -543,6 +543,21 @@ def NVVM_SinOp : NVVM_SingleResultIntrinsicOp<"sin",
   let assemblyFormat = "$src attr-dict `:` type($src)";
 }
 
+
+def NVVM_CosOp : NVVM_SingleResultIntrinsicOp<"cos",
+    [Pure, SameOperandsAndResultType]> {
+  let summary = "Cosine (fast approximation)";
+  let description = [{
+    Computes a fast approximation of the cosine of the input value (in
+    radians). The `ftz` attribute, when set, flushes subnormal inputs
+    and results to sign-preserving zero.
+  }];
+  let arguments = (ins F32:$src,
+                       DefaultValuedAttr<BoolAttr, "false">:$ftz);
+  let results = (outs F32:$res);
+  let assemblyFormat = "$src attr-dict `:` type($src)";
+}
+
 //===----------------------------------------------------------------------===//
 // NVVM redux op definitions
 //===----------------------------------------------------------------------===//
diff --git a/mlir/lib/Dialect/LLVMIR/IR/NVVMDialect.cpp b/mlir/lib/Dialect/LLVMIR/IR/NVVMDialect.cpp
index b0cebd45624a9..8b4f832997a0f 100644
--- a/mlir/lib/Dialect/LLVMIR/IR/NVVMDialect.cpp
+++ b/mlir/lib/Dialect/LLVMIR/IR/NVVMDialect.cpp
@@ -3464,6 +3464,16 @@ mlir::NVVM::IDArgPair NVVM::BarrierOp::getIntrinsicIDAndArgs(
   return {id, std::move(args)};
 }
 
+mlir::NVVM::IDArgPair
+CosOp::getIntrinsicIDAndArgs(Operation &op, LLVM::ModuleTranslation &mt,
+                             llvm::IRBuilderBase &builder) {
+  auto thisOp = cast<NVVM::CosOp>(op);
+  llvm::Intrinsic::ID id = thisOp.getFtz()
+                               ? llvm::Intrinsic::nvvm_cos_approx_ftz_f
+                               : llvm::Intrinsic::nvvm_cos_approx_f;
+  return {id, {mt.lookupValue(thisOp.getSrc())}};
+}
+
 mlir::NVVM::IDArgPair
 SinOp::getIntrinsicIDAndArgs(Operation &op, LLVM::ModuleTranslation &mt,
                              llvm::IRBuilderBase &builder) {
diff --git a/mlir/test/Dialect/LLVMIR/nvvm-transcendentals.mlir b/mlir/test/Dialect/LLVMIR/nvvm-transcendentals.mlir
index c773e3f06a612..777152e0fd797 100644
--- a/mlir/test/Dialect/LLVMIR/nvvm-transcendentals.mlir
+++ b/mlir/test/Dialect/LLVMIR/nvvm-transcendentals.mlir
@@ -1,5 +1,20 @@
 // RUN: mlir-opt %s -split-input-file | FileCheck %s
 
+// CHECK-LABEL: @nvvm_cos_f32
+func.func @nvvm_cos_f32(%arg0: f32) -> f32 {
+  // CHECK: nvvm.cos {{.*}} : f32
+  %0 = nvvm.cos %arg0 : f32
+  return %0 : f32
+}
+
+// CHECK-LABEL: @nvvm_cos_ftz_f32
+func.func @nvvm_cos_ftz_f32(%arg0: f32) -> f32 {
+  // CHECK: nvvm.cos {{.*}} {ftz = true} : f32
+  %0 = nvvm.cos %arg0 {ftz = true} : f32
+  return %0 : f32
+}
+// RUN: mlir-opt %s -split-input-file | FileCheck %s
+
 // CHECK-LABEL: @nvvm_sin_f32
 func.func @nvvm_sin_f32(%arg0: f32) -> f32 {
   // CHECK: nvvm.sin {{.*}} : f32
diff --git a/mlir/test/Target/LLVMIR/nvvm/transcendentals.mlir b/mlir/test/Target/LLVMIR/nvvm/transcendentals.mlir
index 65374c9bd5e87..06da78820b25f 100644
--- a/mlir/test/Target/LLVMIR/nvvm/transcendentals.mlir
+++ b/mlir/test/Target/LLVMIR/nvvm/transcendentals.mlir
@@ -13,3 +13,17 @@ llvm.func @nvvm_sin_ftz(%arg0: f32) -> f32 {
   %0 = nvvm.sin %arg0 {ftz = true} : f32
   llvm.return %0 : f32
 }
+
+// CHECK-LABEL: @nvvm_cos
+llvm.func @nvvm_cos(%arg0: f32) -> f32 {
+  // CHECK: call float @llvm.nvvm.cos.approx.f(float %{{.*}})
+  %0 = nvvm.cos %arg0 : f32
+  llvm.return %0 : f32
+}
+
+// CHECK-LABEL: @nvvm_cos_ftz
+llvm.func @nvvm_cos_ftz(%arg0: f32) -> f32 {
+  // CHECK: call float @llvm.nvvm.cos.approx.ftz.f(float %{{.*}})
+  %0 = nvvm.cos %arg0 {ftz = true} : f32
+  llvm.return %0 : f32
+}