================ @@ -0,0 +1,153 @@ +//===-- amdgpuintrin.h - AMDPGU intrinsic functions -----------------------===// +// +// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. +// See https://llvm.org/LICENSE.txt for license information. +// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception +// +//===----------------------------------------------------------------------===// + +#ifndef __AMDGPUINTRIN_H +#define __AMDGPUINTRIN_H + +#ifndef __AMDGPU__ +#error "This file is intended for AMDGPU targets or offloading to AMDGPU" +#endif + +#include <stdbool.h> +#include <stdint.h> + +#if defined(__HIP__) || defined(__CUDA__) +#define _DEFAULT_ATTRS __attribute__((device)) +#elif !defined(_DEFAULT_ATTRS) +#define _DEFAULT_ATTRS +#endif + +#pragma omp begin declare target device_type(nohost) +#pragma omp begin declare variant match(device = {arch(amdgcn)}) + +// Type aliases to the address spaces used by the AMDGPU backend. +#define _Private __attribute__((opencl_private)) +#define _Constant __attribute__((opencl_constant)) +#define _Local __attribute__((opencl_local)) +#define _Global __attribute__((opencl_global)) + +// Attribute to declare a function as a kernel. +#define _Kernel __attribute__((amdgpu_kernel, visibility("protected"))) ---------------- jhuber6 wrote:
The NVPTX target blatantly ignore visibility, so protected doesn't really make a huge difference. It does matter for LTO however which is why it's here. The GPU targets only ever go to an ELF target right now. We use protected for everything because this pretty much does exactly what we want, a symbol visible from the GPU ELF that can't be preempted so we don't need weird DSO things. https://github.com/llvm/llvm-project/pull/110179 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
