[Clang][AMDGPU] Add the missing builtin __builtin_amdgcn_sqrt_bf16 (#149447)

Co-authored-by: Mekhanoshin, Stanislav <[email protected]>

Author: shiltian

clang/include/clang/Basic/BuiltinsAMDGPU.def

@@ -673,6 +673,7 @@ TARGET_BUILTIN(__builtin_amdgcn_tanhf, "ff", "nc", "tanh-insts")
 TARGET_BUILTIN(__builtin_amdgcn_tanhh, "hh", "nc", "tanh-insts")
 TARGET_BUILTIN(__builtin_amdgcn_tanh_bf16, "yy", "nc", "bf16-trans-insts")
 TARGET_BUILTIN(__builtin_amdgcn_rcp_bf16, "yy", "nc", "bf16-trans-insts")
+TARGET_BUILTIN(__builtin_amdgcn_sqrt_bf16, "yy", "nc", "bf16-trans-insts")
 TARGET_BUILTIN(__builtin_amdgcn_rsq_bf16, "yy", "nc", "bf16-trans-insts")
 TARGET_BUILTIN(__builtin_amdgcn_log_bf16, "yy", "nc", "bf16-trans-insts")
 TARGET_BUILTIN(__builtin_amdgcn_exp2_bf16, "yy", "nc", "bf16-trans-insts")

clang/lib/CodeGen/TargetBuiltins/AMDGPU.cpp

@@ -416,6 +416,7 @@ Value *CodeGenFunction::EmitAMDGPUBuiltinExpr(unsigned BuiltinID,
   case AMDGPU::BI__builtin_amdgcn_sqrt:
   case AMDGPU::BI__builtin_amdgcn_sqrtf:
   case AMDGPU::BI__builtin_amdgcn_sqrth:
+  case AMDGPU::BI__builtin_amdgcn_sqrt_bf16:
     return emitBuiltinWithOneOverloadedType<1>(*this, E,
                                                Intrinsic::amdgcn_sqrt);
   case AMDGPU::BI__builtin_amdgcn_rsq:

clang/test/CodeGenOpenCL/builtins-amdgcn-gfx1250.cl

@@ -119,6 +119,25 @@ void test_rcp_bf16(global __bf16* out, __bf16 a)
   *out = __builtin_amdgcn_rcp_bf16(a);
 }
 
+// CHECK-LABEL: @test_sqrt_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[OUT_ADDR:%.*]] = alloca ptr addrspace(1), align 8, addrspace(5)
+// CHECK-NEXT:    [[A_ADDR:%.*]] = alloca bfloat, align 2, addrspace(5)
+// CHECK-NEXT:    [[OUT_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[OUT_ADDR]] to ptr
+// CHECK-NEXT:    [[A_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[A_ADDR]] to ptr
+// CHECK-NEXT:    store ptr addrspace(1) [[OUT:%.*]], ptr [[OUT_ADDR_ASCAST]], align 8
+// CHECK-NEXT:    store bfloat [[A:%.*]], ptr [[A_ADDR_ASCAST]], align 2
+// CHECK-NEXT:    [[TMP0:%.*]] = load bfloat, ptr [[A_ADDR_ASCAST]], align 2
+// CHECK-NEXT:    [[TMP1:%.*]] = call bfloat @llvm.amdgcn.sqrt.bf16(bfloat [[TMP0]])
+// CHECK-NEXT:    [[TMP2:%.*]] = load ptr addrspace(1), ptr [[OUT_ADDR_ASCAST]], align 8
+// CHECK-NEXT:    store bfloat [[TMP1]], ptr addrspace(1) [[TMP2]], align 2
+// CHECK-NEXT:    ret void
+//
+void test_sqrt_bf16(global __bf16* out, __bf16 a)
+{
+  *out = __builtin_amdgcn_sqrt_bf16(a);
+}
+
 // CHECK-LABEL: @test_rsq_bf16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[OUT_ADDR:%.*]] = alloca ptr addrspace(1), align 8, addrspace(5)