diff --git a/mlir/lib/Conversion/GPUToNVVM/LowerGpuOpsToNVVMOps.cpp b/mlir/lib/Conversion/GPUToNVVM/LowerGpuOpsToNVVMOps.cpp
--- a/mlir/lib/Conversion/GPUToNVVM/LowerGpuOpsToNVVMOps.cpp
+++ b/mlir/lib/Conversion/GPUToNVVM/LowerGpuOpsToNVVMOps.cpp
@@ -294,6 +294,8 @@
                                    "__nv_sqrt");
   populateOpPatterns<math::TanhOp>(converter, patterns, "__nv_tanhf",
                                    "__nv_tanh");
+  populateOpPatterns<math::TanOp>(converter, patterns, "__nv_tanf",
+                                  "__nv_tan");
 }
 
 std::unique_ptr<OperationPass<gpu::GPUModuleOp>>
diff --git a/mlir/test/Conversion/GPUToNVVM/gpu-to-nvvm.mlir b/mlir/test/Conversion/GPUToNVVM/gpu-to-nvvm.mlir
--- a/mlir/test/Conversion/GPUToNVVM/gpu-to-nvvm.mlir
+++ b/mlir/test/Conversion/GPUToNVVM/gpu-to-nvvm.mlir
@@ -343,6 +343,25 @@
 
 // -----
 
+gpu.module @test_module {
+  // CHECK: llvm.func @__nv_tanf(f32) -> f32
+  // CHECK: llvm.func @__nv_tan(f64) -> f64
+  // CHECK-LABEL: func @gpu_tan
+  func.func @gpu_tan(%arg_f16 : f16, %arg_f32 : f32, %arg_f64 : f64) -> (f16, f32, f64) {
+    %result16 = math.tan %arg_f16 : f16
+    // CHECK: llvm.fpext %{{.*}} : f16 to f32
+    // CHECK-NEXT: llvm.call @__nv_tanf(%{{.*}}) : (f32) -> f32
+    // CHECK-NEXT: llvm.fptrunc %{{.*}} : f32 to f16
+    %result32 = math.tan %arg_f32 : f32
+    // CHECK: llvm.call @__nv_tanf(%{{.*}}) : (f32) -> f32
+    %result64 = math.tan %arg_f64 : f64
+    // CHECK: llvm.call @__nv_tan(%{{.*}}) : (f64) -> f64
+    func.return %result16, %result32, %result64 : f16, f32, f64
+  }
+}
+
+// -----
+
 gpu.module @test_module {
   // CHECK: llvm.func @__nv_tanhf(f32) -> f32
   // CHECK: llvm.func @__nv_tanh(f64) -> f64