34 lines · plain
1// REQUIRES: nvptx-registered-target2// REQUIRES: x86-registered-target3 4// RUN: %clang_cc1 "-triple" "x86_64-unknown-linux-gnu" "-aux-triple" "nvptx64-nvidia-cuda" \5// RUN: "-target-cpu" "x86-64" -fsyntax-only -verify=scalar -Wno-unused %s6// RUN: %clang_cc1 "-aux-triple" "x86_64-unknown-linux-gnu" "-triple" "nvptx64-nvidia-cuda" \7// RUN: -fcuda-is-device "-aux-target-cpu" "x86-64" -fsyntax-only -verify=scalar -Wno-unused %s8 9#include "Inputs/cuda.h"10 11__device__ void test(bool b, __bf16 *out, __bf16 in) {12 __bf16 bf16 = in; // No error on using the type itself.13 14 bf16 + bf16;15 bf16 - bf16;16 bf16 * bf16;17 bf16 / bf16;18 19 __fp16 fp16;20 21 bf16 + fp16;22 fp16 + bf16;23 bf16 - fp16;24 fp16 - bf16;25 bf16 * fp16;26 fp16 * bf16;27 bf16 / fp16;28 fp16 / bf16;29 bf16 = fp16; // scalar-error {{assigning to '__bf16' from incompatible type '__fp16'}}30 fp16 = bf16; // scalar-error {{assigning to '__fp16' from incompatible type '__bf16'}}31 bf16 + (b ? fp16 : bf16);32 *out = bf16;33}34