================ @@ -0,0 +1,97 @@ +// RUN: echo -n "GPU binary would be here." > %t + +// RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -fcuda-is-device -fclangir \ +// RUN: -x hip -emit-cir %s -o %t.dev.cir +// RUN: FileCheck --check-prefix=CIR-DEV --input-file=%t.dev.cir %s + +// RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -fcuda-is-device -fclangir \ +// RUN: -x hip -emit-llvm %s -o %t.dev.cir.ll +// RUN: FileCheck --check-prefix=DEV --input-file=%t.dev.cir.ll %s + +// RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -fcuda-is-device \ +// RUN: -x hip -emit-llvm %s -o %t.dev.ll +// RUN: FileCheck --check-prefix=DEV --input-file=%t.dev.ll %s + +// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -fclangir \ +// RUN: -x hip -fhip-new-launch-api -fcuda-include-gpubinary %t \ +// RUN: -emit-cir %s -o %t.host.cir +// RUN: FileCheck --check-prefix=CIR-HOST --input-file=%t.host.cir %s + +// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -fclangir \ +// RUN: -x hip -fhip-new-launch-api -fcuda-include-gpubinary %t \ +// RUN: -emit-llvm %s -o %t.host.cir.ll +// RUN: FileCheck --check-prefix=HOST --input-file=%t.host.cir.ll %s + +// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu \ +// RUN: -x hip -fhip-new-launch-api -fcuda-include-gpubinary %t \ +// RUN: -emit-llvm %s -o %t.host.ll +// RUN: FileCheck --check-prefix=HOST --input-file=%t.host.ll %s + +#include "../CodeGenCUDA/Inputs/cuda.h" ---------------- steffenlarsen wrote:
Hmm, that is indeed a little odd. Testing in places like clang/test/SemaCUDA/managed-var.cu seems to also only be HIP, but the [CUDA docs](https://docs.nvidia.com/cuda/cuda-programming-guide/05-appendices/cpp-language-extensions.html#memory-space-specifiers) mention the specifier as well. I'm not sure the relation here, but I can indeed not find any CUDA testing for it. https://github.com/llvm/llvm-project/pull/230910 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
