[PATCH] D154797: [CUDA][HIP] Rename and fix `-fcuda-approx-transcendentals`
yaxunl added inline comments.
Comment at: clang/lib/Driver/ToolChains/Clang.cpp:7228
+ } else {
+Args.ClaimAllArgs(options::OPT_fgpu_approx_transcendentals);
+Args.ClaimAllArgs(options::OPT_fno_gpu_approx_transcendentals);
MaskRay wrote:
> You can use `Args.claimAllArgs(options::OPT_fgpu_approx_transcendentals,
> options::OPT_fno_gpu_approx_transcendentals);`
will do
Comment at: clang/lib/Frontend/InitPreprocessor.cpp:1294
+if (!LangOpts.HIP)
+ Builder.defineMacro("__CLANG_CUDA_APPROX_TRANSCENDENTALS__");
+Builder.defineMacro("__CLANG_GPU_APPROX_TRANSCENDENTALS__");
tra wrote:
> I think we can remove it. I don't think we need to keep the old one around.
> Internal headers have been changed and the macro was never intended for
> public use.
will remove
Comment at: clang/test/Driver/hip-options.hip:209
+
+// APPROXNEG-NOT: warning
MaskRay wrote:
> If `%t` happens to be in a path with `warning` as a substring, this will
> spuriously fail.
>
> Suggest `%clang -fdriver-only -Werror... 2>&1 | count 0` to test that there
> is no warning/error.
will do
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D154797/new/
https://reviews.llvm.org/D154797
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D154797: [CUDA][HIP] Rename and fix `-fcuda-approx-transcendentals`
This revision was landed with ongoing or failed builds. This revision was automatically updated to reflect the committed changes. yaxunl marked 4 inline comments as done. Closed by commit rGe17882430e5c: [CUDA][HIP] Rename and fix `-fcuda-approx-transcendentals` (authored by yaxunl). Herald added a project: clang. Changed prior to commit: https://reviews.llvm.org/D154797?vs=538787&id=544001#toc Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D154797/new/ https://reviews.llvm.org/D154797 Files: clang/include/clang/Basic/LangOptions.def clang/include/clang/Driver/Options.td clang/lib/Driver/ToolChains/Clang.cpp clang/lib/Driver/ToolChains/Cuda.cpp clang/lib/Driver/ToolChains/HIPAMD.cpp clang/lib/Driver/ToolChains/HIPSPV.cpp clang/lib/Frontend/InitPreprocessor.cpp clang/lib/Headers/__clang_cuda_math.h clang/lib/Headers/__clang_cuda_runtime_wrapper.h clang/lib/Headers/__clang_hip_math.h clang/test/Driver/hip-macros.hip clang/test/Driver/hip-options.hip clang/test/Headers/__clang_hip_math.hip clang/test/Headers/nvptx_device_math_sin.c clang/test/Headers/nvptx_device_math_sin.cpp clang/test/Preprocessor/cuda-approx-transcendentals.cu Index: clang/test/Preprocessor/cuda-approx-transcendentals.cu === --- clang/test/Preprocessor/cuda-approx-transcendentals.cu +++ clang/test/Preprocessor/cuda-approx-transcendentals.cu @@ -3,6 +3,6 @@ // RUN: %clang -fcuda-approx-transcendentals --cuda-device-only -nocudainc -nocudalib -target i386-unknown-linux-gnu -x cuda -E -dM -o - /dev/null | FileCheck --check-prefix DEVICE-FAST %s // RUN: %clang -ffast-math --cuda-device-only -nocudainc -nocudalib -target i386-unknown-linux-gnu -x cuda -E -dM -o - /dev/null | FileCheck --check-prefix DEVICE-FAST %s -// HOST-NOT: __CLANG_CUDA_APPROX_TRANSCENDENTALS__ -// DEVICE-NOFAST-NOT: __CLANG_CUDA_APPROX_TRANSCENDENTALS__ -// DEVICE-FAST: __CLANG_CUDA_APPROX_TRANSCENDENTALS__ +// HOST-NOT: __CLANG_GPU_APPROX_TRANSCENDENTALS__ +// DEVICE-NOFAST-NOT: __GPU_CUDA_APPROX_TRANSCENDENTALS__ +// DEVICE-FAST: __CLANG_GPU_APPROX_TRANSCENDENTALS__ Index: clang/test/Headers/nvptx_device_math_sin.cpp === --- clang/test/Headers/nvptx_device_math_sin.cpp +++ clang/test/Headers/nvptx_device_math_sin.cpp @@ -2,7 +2,7 @@ // RUN: %clang_cc1 -x c++ -internal-isystem %S/Inputs/include -fopenmp -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o %t-ppc-host.bc // RUN: %clang_cc1 -x c++ -include __clang_openmp_device_functions.h -internal-isystem %S/../../lib/Headers/openmp_wrappers -internal-isystem %S/Inputs/include -fopenmp -triple nvptx64-nvidia-cuda -aux-triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix=SLOW // RUN: %clang_cc1 -x c++ -internal-isystem %S/Inputs/include -fopenmp -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o %t-ppc-host.bc -ffast-math -ffp-contract=fast -// RUN: %clang_cc1 -x c++ -include __clang_openmp_device_functions.h -internal-isystem %S/../../lib/Headers/openmp_wrappers -internal-isystem %S/Inputs/include -fopenmp -triple nvptx64-nvidia-cuda -aux-triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - -ffast-math -ffp-contract=fast | FileCheck %s --check-prefix=FAST +// RUN: %clang_cc1 -x c++ -include __clang_openmp_device_functions.h -internal-isystem %S/../../lib/Headers/openmp_wrappers -internal-isystem %S/Inputs/include -fopenmp -triple nvptx64-nvidia-cuda -aux-triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - -fgpu-approx-transcendentals -ffast-math -ffp-contract=fast | FileCheck %s --check-prefix=FAST // expected-no-diagnostics #include Index: clang/test/Headers/nvptx_device_math_sin.c === --- clang/test/Headers/nvptx_device_math_sin.c +++ clang/test/Headers/nvptx_device_math_sin.c @@ -2,7 +2,7 @@ // RUN: %clang_cc1 -x c -internal-isystem %S/Inputs/include -fopenmp -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o %t-ppc-host.bc // RUN: %clang_cc1 -x c -include __clang_openmp_device_functions.h -internal-isystem %S/../../lib/Headers/openmp_wrappers -internal-isystem %S/Inputs/include -fopenmp -triple nvptx64-nvidia-cuda -aux-triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix=SLOW // RUN: %clang_cc1 -x c -internal-isystem %S/Inputs/include
[PATCH] D154797: [CUDA][HIP] Rename and fix `-fcuda-approx-transcendentals`
MaskRay accepted this revision. MaskRay added a comment. Some nits about testing CHANGES SINCE LAST ACTION https://reviews.llvm.org/D154797/new/ https://reviews.llvm.org/D154797 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D154797: [CUDA][HIP] Rename and fix `-fcuda-approx-transcendentals`
MaskRay added inline comments.
Comment at: clang/lib/Driver/ToolChains/Clang.cpp:7228
+ } else {
+Args.ClaimAllArgs(options::OPT_fgpu_approx_transcendentals);
+Args.ClaimAllArgs(options::OPT_fno_gpu_approx_transcendentals);
You can use `Args.claimAllArgs(options::OPT_fgpu_approx_transcendentals,
options::OPT_fno_gpu_approx_transcendentals);`
Comment at: clang/test/Driver/hip-options.hip:184
+
+// APPROX: clang{{.*}} "-triple" "amdgcn-amd-amdhsa" {{.*}}
"-fgpu-approx-transcendentals"
+// APPROX: clang{{.*}} "-triple" "x86_64-unknown-linux-gnu" {{.*}}
"-fgpu-approx-transcendentals"
Just test `-cc1`: `// APPROX: "-cc1"{{.*}} "-triple" "amdgcn-amd-amdhsa" {{.*}}
"-fgpu-approx-transcendentals"`
Testing `clang` requires `-no-canonical-prefixes`
https://maskray.me/blog/2021-03-28-compiler-driver-and-cross-compilation#misc
Comment at: clang/test/Driver/hip-options.hip:209
+
+// APPROXNEG-NOT: warning
If `%t` happens to be in a path with `warning` as a substring, this will
spuriously fail.
Suggest `%clang -fdriver-only -Werror... 2>&1 | count 0` to test that there is
no warning/error.
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D154797/new/
https://reviews.llvm.org/D154797
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D154797: [CUDA][HIP] Rename and fix `-fcuda-approx-transcendentals`
tra accepted this revision.
tra added inline comments.
This revision is now accepted and ready to land.
Comment at: clang/lib/Frontend/InitPreprocessor.cpp:1294
+if (!LangOpts.HIP)
+ Builder.defineMacro("__CLANG_CUDA_APPROX_TRANSCENDENTALS__");
+Builder.defineMacro("__CLANG_GPU_APPROX_TRANSCENDENTALS__");
I think we can remove it. I don't think we need to keep the old one around.
Internal headers have been changed and the macro was never intended for public
use.
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D154797/new/
https://reviews.llvm.org/D154797
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D154797: [CUDA][HIP] Rename and fix `-fcuda-approx-transcendentals`
yaxunl updated this revision to Diff 538787. yaxunl marked 3 inline comments as done. yaxunl edited the summary of this revision. yaxunl added a comment. Herald added a reviewer: jdoerfert. Herald added subscribers: jplehr, asavonic, sstefan1. revised by comments CHANGES SINCE LAST ACTION https://reviews.llvm.org/D154797/new/ https://reviews.llvm.org/D154797 Files: clang/include/clang/Basic/LangOptions.def clang/include/clang/Driver/Options.td clang/lib/Driver/ToolChains/Clang.cpp clang/lib/Driver/ToolChains/Cuda.cpp clang/lib/Driver/ToolChains/HIPAMD.cpp clang/lib/Driver/ToolChains/HIPSPV.cpp clang/lib/Frontend/InitPreprocessor.cpp clang/lib/Headers/__clang_cuda_math.h clang/lib/Headers/__clang_cuda_runtime_wrapper.h clang/lib/Headers/__clang_hip_math.h clang/test/Driver/hip-macros.hip clang/test/Driver/hip-options.hip clang/test/Headers/__clang_hip_math.hip clang/test/Headers/nvptx_device_math_sin.c clang/test/Headers/nvptx_device_math_sin.cpp Index: clang/test/Headers/nvptx_device_math_sin.cpp === --- clang/test/Headers/nvptx_device_math_sin.cpp +++ clang/test/Headers/nvptx_device_math_sin.cpp @@ -2,7 +2,7 @@ // RUN: %clang_cc1 -x c++ -internal-isystem %S/Inputs/include -fopenmp -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o %t-ppc-host.bc // RUN: %clang_cc1 -x c++ -include __clang_openmp_device_functions.h -internal-isystem %S/../../lib/Headers/openmp_wrappers -internal-isystem %S/Inputs/include -fopenmp -triple nvptx64-nvidia-cuda -aux-triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix=SLOW // RUN: %clang_cc1 -x c++ -internal-isystem %S/Inputs/include -fopenmp -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o %t-ppc-host.bc -ffast-math -ffp-contract=fast -// RUN: %clang_cc1 -x c++ -include __clang_openmp_device_functions.h -internal-isystem %S/../../lib/Headers/openmp_wrappers -internal-isystem %S/Inputs/include -fopenmp -triple nvptx64-nvidia-cuda -aux-triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - -ffast-math -ffp-contract=fast | FileCheck %s --check-prefix=FAST +// RUN: %clang_cc1 -x c++ -include __clang_openmp_device_functions.h -internal-isystem %S/../../lib/Headers/openmp_wrappers -internal-isystem %S/Inputs/include -fopenmp -triple nvptx64-nvidia-cuda -aux-triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - -fgpu-approx-transcendentals -ffast-math -ffp-contract=fast | FileCheck %s --check-prefix=FAST // expected-no-diagnostics #include Index: clang/test/Headers/nvptx_device_math_sin.c === --- clang/test/Headers/nvptx_device_math_sin.c +++ clang/test/Headers/nvptx_device_math_sin.c @@ -2,7 +2,7 @@ // RUN: %clang_cc1 -x c -internal-isystem %S/Inputs/include -fopenmp -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o %t-ppc-host.bc // RUN: %clang_cc1 -x c -include __clang_openmp_device_functions.h -internal-isystem %S/../../lib/Headers/openmp_wrappers -internal-isystem %S/Inputs/include -fopenmp -triple nvptx64-nvidia-cuda -aux-triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix=SLOW // RUN: %clang_cc1 -x c -internal-isystem %S/Inputs/include -fopenmp -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o %t-ppc-host.bc -ffast-math -ffp-contract=fast -// RUN: %clang_cc1 -x c -include __clang_openmp_device_functions.h -internal-isystem %S/../../lib/Headers/openmp_wrappers -internal-isystem %S/Inputs/include -fopenmp -triple nvptx64-nvidia-cuda -aux-triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - -ffast-math -ffp-contract=fast | FileCheck %s --check-prefix=FAST +// RUN: %clang_cc1 -x c -include __clang_openmp_device_functions.h -internal-isystem %S/../../lib/Headers/openmp_wrappers -internal-isystem %S/Inputs/include -fopenmp -triple nvptx64-nvidia-cuda -aux-triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - -fgpu-approx-transcendentals -ffast-math -ffp-contract=fast | FileCheck %s --check-prefix=FAST // expected-no-diagnostics #include Index: clang/test/Headers/__clang_hip_math.hip === --- clang/test/Headers/__clang_hip_math.hip +++
[PATCH] D154797: [CUDA][HIP] Rename and fix `-fcuda-approx-transcendentals`
yaxunl marked 3 inline comments as done.
yaxunl added inline comments.
Comment at: clang/lib/Driver/ToolChains/Clang.cpp:7221-7223
+bool UseApproxTranscendentals = false;
+if (Args.hasFlag(options::OPT_ffast_math, options::OPT_fno_fast_math,
+ false))
tra wrote:
> ```
> bool UseApproxTranscendentals = Args.hasFlag(options::OPT_ffast_math,
> options::OPT_fno_fast_math, false));
> ```
fixed
Comment at: clang/lib/Frontend/InitPreprocessor.cpp:1292-1293
+ if (LangOpts.GPUDeviceApproxTranscendentals) {
+Builder.defineMacro(Twine("__CLANG_") + (LangOpts.HIP ? "HIP" : "CUDA") +
+"_APPROX_TRANSCENDENTALS__");
}
tra wrote:
> We may want to rename the macro to `__CLANG_GPU_APPROX_TRANSCENDENTALS__`,
> too.
>
will emit `__CLANG_GPU_APPROX_TRANSCENDENTALS__`
Comment at: clang/test/Driver/hip-options.hip:179
+// RUN: %clang -### --target=x86_64-unknown-linux-gnu -nogpuinc -nogpulib
-fgpu-approx-transcendentals \
+// RUN: --cuda-gpu-arch=gfx906 %s 2>&1 | FileCheck -check-prefixes=APPROX %s
+
MaskRay wrote:
> excess spaces before `%s`
>
> ditto below
>
> Prefer `--check-prefix=` when there is one single check.
fixed
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D154797/new/
https://reviews.llvm.org/D154797
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D154797: [CUDA][HIP] Rename and fix `-fcuda-approx-transcendentals`
MaskRay added inline comments. Comment at: clang/test/Driver/hip-options.hip:179 +// RUN: %clang -### --target=x86_64-unknown-linux-gnu -nogpuinc -nogpulib -fgpu-approx-transcendentals \ +// RUN: --cuda-gpu-arch=gfx906 %s 2>&1 | FileCheck -check-prefixes=APPROX %s + excess spaces before `%s` ditto below Prefer `--check-prefix=` when there is one single check. CHANGES SINCE LAST ACTION https://reviews.llvm.org/D154797/new/ https://reviews.llvm.org/D154797 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D154797: [CUDA][HIP] Rename and fix `-fcuda-approx-transcendentals`
tra added a comment.
Looks good in general.
Comment at: clang/lib/Driver/ToolChains/Clang.cpp:7221-7223
+bool UseApproxTranscendentals = false;
+if (Args.hasFlag(options::OPT_ffast_math, options::OPT_fno_fast_math,
+ false))
```
bool UseApproxTranscendentals = Args.hasFlag(options::OPT_ffast_math,
options::OPT_fno_fast_math, false));
```
Comment at: clang/lib/Frontend/InitPreprocessor.cpp:1292-1293
+ if (LangOpts.GPUDeviceApproxTranscendentals) {
+Builder.defineMacro(Twine("__CLANG_") + (LangOpts.HIP ? "HIP" : "CUDA") +
+"_APPROX_TRANSCENDENTALS__");
}
We may want to rename the macro to `__CLANG_GPU_APPROX_TRANSCENDENTALS__`, too.
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D154797/new/
https://reviews.llvm.org/D154797
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D154797: [CUDA][HIP] Rename and fix `-fcuda-approx-transcendentals`
yaxunl created this revision.
yaxunl added reviewers: tra, MaskRay, b-sumner.
Herald added subscribers: mattd, carlosgalvezp.
Herald added a project: All.
yaxunl requested review of this revision.
Rename `-fcuda-approx-transcendentals` as
`-fgpu-approx-transcendentals` and pass it
to both device and host clang -cc1.
Fix its interaction with `-ffast-math` to allow
`-fno-gpu-approx-transcendentals` to override
the implicit `-fcuda-approx-transcendentals`
due to `-ffast-math`.
Rename the predefined macro to be
`__CLANG_[CUDA|HIP]_APPROX_TRANSCENDENTALS__`
depending on the language. Emit the macro for both
device and host compilation.
https://reviews.llvm.org/D154797
Files:
clang/include/clang/Basic/LangOptions.def
clang/include/clang/Driver/Options.td
clang/lib/Driver/ToolChains/Clang.cpp
clang/lib/Driver/ToolChains/Cuda.cpp
clang/lib/Driver/ToolChains/HIPAMD.cpp
clang/lib/Driver/ToolChains/HIPSPV.cpp
clang/lib/Frontend/InitPreprocessor.cpp
clang/lib/Headers/__clang_cuda_math.h
clang/lib/Headers/__clang_cuda_runtime_wrapper.h
clang/lib/Headers/__clang_hip_math.h
clang/test/Driver/hip-macros.hip
clang/test/Driver/hip-options.hip
clang/test/Headers/__clang_hip_math.hip
Index: clang/test/Headers/__clang_hip_math.hip
===
--- clang/test/Headers/__clang_hip_math.hip
+++ clang/test/Headers/__clang_hip_math.hip
@@ -22,7 +22,7 @@
// RUN: -internal-isystem %S/../../lib/Headers/cuda_wrappers \
// RUN: -internal-isystem %S/Inputs/include \
// RUN: -triple amdgcn-amd-amdhsa -aux-triple x86_64-unknown-unknown \
-// RUN: -target-cpu gfx906 -emit-llvm %s -fcuda-is-device -O1 -fcuda-approx-transcendentals -o - \
+// RUN: -target-cpu gfx906 -emit-llvm %s -fcuda-is-device -O1 -fgpu-approx-transcendentals -o - \
// RUN: -D__HIPCC_RTC__ | FileCheck -check-prefixes=CHECK,APPROX %s
#define BOOL_TYPE int
Index: clang/test/Driver/hip-options.hip
===
--- clang/test/Driver/hip-options.hip
+++ clang/test/Driver/hip-options.hip
@@ -169,3 +169,41 @@
// RUN: %clang -### -nogpuinc -nogpulib -fhip-fp32-correctly-rounded-divide-sqrt \
// RUN: --cuda-gpu-arch=gfx906 %s 2>&1 | FileCheck -check-prefixes=CRDS %s
// CRDS-NOT: "-f{{(no-)?}}hip-fp32-correctly-rounded-divide-sqrt"
+
+// Check -fgpu-approx-transcendentals is passed to clang -cc1 but
+// (default) -fno-gpu-approx-transcendentals is not.
+// -ffast-math implies -fgpu-approx-transcendentals, which can be overridden
+// by -fno-gpu-approx-transcendentals.
+
+// RUN: %clang -### --target=x86_64-unknown-linux-gnu -nogpuinc -nogpulib -fgpu-approx-transcendentals \
+// RUN: --cuda-gpu-arch=gfx906 %s 2>&1 | FileCheck -check-prefixes=APPROX %s
+
+// RUN: %clang -### --target=x86_64-unknown-linux-gnu -nogpuinc -nogpulib -ffast-math \
+// RUN: --cuda-gpu-arch=gfx906 %s 2>&1 | FileCheck -check-prefixes=APPROX %s
+
+// APPROX: clang{{.*}} "-triple" "amdgcn-amd-amdhsa" {{.*}} "-fgpu-approx-transcendentals"
+// APPROX: clang{{.*}} "-triple" "x86_64-unknown-linux-gnu" {{.*}} "-fgpu-approx-transcendentals"
+
+// RUN: %clang -### -nogpuinc -nogpulib -fno-gpu-approx-transcendentals \
+// RUN: --cuda-gpu-arch=gfx906 %s 2>&1 | FileCheck -check-prefixes=NOAPPROX %s
+
+// RUN: %clang -### -nogpuinc -nogpulib \
+// RUN: --cuda-gpu-arch=gfx906 %s 2>&1 | FileCheck -check-prefixes=NOAPPROX %s
+
+// RUN: %clang -### -nogpuinc -nogpulib -ffast-math -fno-fast-math \
+// RUN: --cuda-gpu-arch=gfx906 %s 2>&1 | FileCheck -check-prefixes=NOAPPROX %s
+
+// RUN: %clang -### -nogpuinc -nogpulib -ffast-math -fno-gpu-approx-transcendentals \
+// RUN: --cuda-gpu-arch=gfx906 %s 2>&1 | FileCheck -check-prefixes=NOAPPROX %s
+
+// NOAPPROX-NOT: "-f{{(no-)?}}gpu-approx-transcendentals"
+
+// Check no warnings for -fgpu-approx-transcendentals.
+
+// RUN: %clang -### --target=x86_64-unknown-linux-gnu -nogpuinc -nogpulib -fgpu-approx-transcendentals \
+// RUN: --cuda-gpu-arch=gfx906 %s 2>&1 | FileCheck -check-prefixes=APPROXNEG %s
+
+// RUN: %clang -### --target=x86_64-unknown-linux-gnu -nostdinc -nostdlib -fgpu-approx-transcendentals \
+// RUN: -x c++ %s 2>&1 | FileCheck -check-prefixes=APPROXNEG %s
+
+// APPROXNEG-NOT: warning
Index: clang/test/Driver/hip-macros.hip
===
--- clang/test/Driver/hip-macros.hip
+++ clang/test/Driver/hip-macros.hip
@@ -63,3 +63,11 @@
// WARN: warning: feature flag '{{[+|-]}}image-insts' is ignored since the feature is read only [-Winvalid-command-line-argument]
// IMAGE-NOT: #define __HIP_NO_IMAGE_SUPPORT
// NOIMAGE: #define __HIP_NO_IMAGE_SUPPORT 1
+
+// RUN: %clang -E -dM --offload-arch=gfx906 -nogpuinc -nogpulib \
+// RUN: %s 2>&1 | FileCheck --check-prefix=NOAPPROX %s
+// RUN: %clang -E -dM --offload-arch=gfx906 -nogpuinc -nogpulib -fgpu-approx-transcendentals \
+// RUN: %s 2>&1 | FileCheck --check-prefix=APPROX %s
+// NOAPPR
