[Mlir-commits] [mlir] [mlir][LLVMIR] Add support for tan intrinsic op (PR #125748)

[Paul Carabas]

https://github.com/PaulCarabas created https://github.com/llvm/llvm-project/pull/125748

This patch adds support for Tan trig. function intrinsic in LLVM dialect & adds missing import/export tests for Sin

>From 94fc09c7c9dd31dee5062cf5ac484b527e685af8 Mon Sep 17 00:00:00 2001
From: PaulCarabas <paulcaraa at gmail.com>
Date: Tue, 4 Feb 2025 21:24:19 +0200
Subject: [PATCH] [mlir][LLVMIR] Add support for tan intrinsic op

---
 .../mlir/Dialect/LLVMIR/LLVMIntrinsicOps.td   |  7 +++++--
 mlir/test/Target/LLVMIR/Import/intrinsic.ll   | 19 +++++++++++++++----
 .../test/Target/LLVMIR/llvmir-intrinsics.mlir | 18 ++++++++++++++----
 3 files changed, 34 insertions(+), 10 deletions(-)

diff --git a/mlir/include/mlir/Dialect/LLVMIR/LLVMIntrinsicOps.td b/mlir/include/mlir/Dialect/LLVMIR/LLVMIntrinsicOps.td
index a7d683438ee8ab0..72fae1bdbf461df 100644
--- a/mlir/include/mlir/Dialect/LLVMIR/LLVMIntrinsicOps.td
+++ b/mlir/include/mlir/Dialect/LLVMIR/LLVMIntrinsicOps.td
@@ -107,7 +107,6 @@ def LLVM_IsFPClass : LLVM_OneResultIntrOp<"is.fpclass", [], [0], [Pure],
 }
 
 def LLVM_CopySignOp : LLVM_BinarySameArgsIntrOpF<"copysign">;
-def LLVM_CosOp : LLVM_UnaryIntrOpF<"cos">;
 def LLVM_ExpOp : LLVM_UnaryIntrOpF<"exp">;
 def LLVM_Exp2Op : LLVM_UnaryIntrOpF<"exp2">;
 def LLVM_FAbsOp : LLVM_UnaryIntrOpF<"fabs">;
@@ -125,7 +124,6 @@ def LLVM_Prefetch : LLVM_ZeroResultIntrOp<"prefetch", [0],
 > {
   let arguments = (ins LLVM_AnyPointer:$addr, I32Attr:$rw, I32Attr:$hint, I32Attr:$cache);
 }
-def LLVM_SinOp : LLVM_UnaryIntrOpF<"sin">;
 def LLVM_RoundEvenOp : LLVM_UnaryIntrOpF<"roundeven">;
 def LLVM_RoundOp : LLVM_UnaryIntrOpF<"round">;
 def LLVM_FTruncOp : LLVM_UnaryIntrOpF<"trunc">;
@@ -167,6 +165,11 @@ def LLVM_SMaxOp : LLVM_BinarySameArgsIntrOpI<"smax">;
 def LLVM_SMinOp : LLVM_BinarySameArgsIntrOpI<"smin">;
 def LLVM_UMaxOp : LLVM_BinarySameArgsIntrOpI<"umax">;
 def LLVM_UMinOp : LLVM_BinarySameArgsIntrOpI<"umin">;
+
+def LLVM_SinOp : LLVM_UnaryIntrOpF<"sin">;
+def LLVM_CosOp : LLVM_UnaryIntrOpF<"cos">;
+def LLVM_TanOp : LLVM_UnaryIntrOpF<"tan">;
+
 def LLVM_SinhOp : LLVM_UnaryIntrOpF<"sinh">;
 def LLVM_CoshOp : LLVM_UnaryIntrOpF<"cosh">;
 def LLVM_TanhOp : LLVM_UnaryIntrOpF<"tanh">;
diff --git a/mlir/test/Target/LLVMIR/Import/intrinsic.ll b/mlir/test/Target/LLVMIR/Import/intrinsic.ll
index bd335323a2e1c93..249a0552c87f380 100644
--- a/mlir/test/Target/LLVMIR/Import/intrinsic.ll
+++ b/mlir/test/Target/LLVMIR/Import/intrinsic.ll
@@ -101,12 +101,23 @@ define void @floor_test(float %0, <8 x float> %1) {
   %4 = call <8 x float> @llvm.floor.v8f32(<8 x float> %1)
   ret void
 }
-; CHECK-LABEL:  llvm.func @cos_test
-define void @cos_test(float %0, <8 x float> %1) {
+; CHECK-LABEL:  llvm.func @trig_test
+define void @trig_test(float %0, <8 x float> %1) {
+  ; CHECK: llvm.intr.sin(%{{.*}}) : (f32) -> f32
+  %3 = call float @llvm.sin.f32(float %0)
+  ; CHECK: llvm.intr.sin(%{{.*}}) : (vector<8xf32>) -> vector<8xf32>
+  %4 = call <8 x float> @llvm.sin.v8f32(<8 x float> %1)
+
   ; CHECK: llvm.intr.cos(%{{.*}}) : (f32) -> f32
-  %3 = call float @llvm.cos.f32(float %0)
+  %5 = call float @llvm.cos.f32(float %0)
   ; CHECK: llvm.intr.cos(%{{.*}}) : (vector<8xf32>) -> vector<8xf32>
-  %4 = call <8 x float> @llvm.cos.v8f32(<8 x float> %1)
+  %6 = call <8 x float> @llvm.cos.v8f32(<8 x float> %1)
+
+  ; CHECK: llvm.intr.tan(%{{.*}}) : (f32) -> f32
+  %7 = call float @llvm.tan.f32(float %0)
+  ; CHECK: llvm.intr.tan(%{{.*}}) : (vector<8xf32>) -> vector<8xf32>
+  %8 = call <8 x float> @llvm.tan.v8f32(<8 x float> %1)
+
   ret void
 }
 ; CHECK-LABEL:  llvm.func @hyperbolic_trig_test
diff --git a/mlir/test/Target/LLVMIR/llvmir-intrinsics.mlir b/mlir/test/Target/LLVMIR/llvmir-intrinsics.mlir
index 382b2b9f3cd732e..2c208789e36ddbe 100644
--- a/mlir/test/Target/LLVMIR/llvmir-intrinsics.mlir
+++ b/mlir/test/Target/LLVMIR/llvmir-intrinsics.mlir
@@ -103,12 +103,22 @@ llvm.func @floor_test(%arg0: f32, %arg1: vector<8xf32>) {
   llvm.return
 }
 
-// CHECK-LABEL: @cos_test
-llvm.func @cos_test(%arg0: f32, %arg1: vector<8xf32>) {
+// CHECK-LABEL: @trig_test
+llvm.func @trig_test(%arg0: f32, %arg1: vector<8xf32>) {
+  // CHECK: call float @llvm.sin.f32
+  llvm.intr.sin(%arg0) : (f32) -> f32
+  // CHECK: call <8 x float> @llvm.sin.v8f32
+  llvm.intr.sin(%arg1) : (vector<8xf32>) -> vector<8xf32>
+
   // CHECK: call float @llvm.cos.f32
-  "llvm.intr.cos"(%arg0) : (f32) -> f32
+  llvm.intr.cos(%arg0) : (f32) -> f32
   // CHECK: call <8 x float> @llvm.cos.v8f32
-  "llvm.intr.cos"(%arg1) : (vector<8xf32>) -> vector<8xf32>
+  llvm.intr.cos(%arg1) : (vector<8xf32>) -> vector<8xf32>
+
+  // CHECK: call float @llvm.tan.f32
+  llvm.intr.tan(%arg0) : (f32) -> f32
+  // CHECK: call <8 x float> @llvm.tan.v8f32
+  llvm.intr.tan(%arg1) : (vector<8xf32>) -> vector<8xf32>
   llvm.return
 }