[HIPIFY][win] cannot find CUDA installation

Question

[HIPIFY][win] cannot find CUDA installation

JinchiDai opened this issue 9 months ago · 18 comments

Problem Description

I managed to generate the hipify-clang tool on win10, but it doesn't look good.

My installation path of CUDA is: C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6
When I type: " hipify-clang.exe D:\code\C++\CudaRuntime1\CudaRuntime1\kernel.cu --cuda-path=C:\Program Files\NVIDIA GPU Computing Toolkit "

The error is like:
error: no such file or directory: 'Files\NVIDIA'
error: no such file or directory: 'GPU'
error: no such file or directory: 'Computing'
error: no such file or directory: 'Toolkit\Files'
error: unsupported architecture 'nvptx64' for host compilation

error: cannot find CUDA installation; provide its path via '--cuda-path', or pass '-nocudainc' to build without CUDA includes
C:\Users\DaisyDIE\AppData\Local\Temp\kernel.cu-87d3d6.hip:2:10: fatal error: 'cuda_runtime.h' file not found
#include "cuda_runtime.h"

I want to know what I should do next.

Operating System

Windows 10 pro 19045.3803

CPU

Intel Core i7-10870H

ROCm Version

No response

ROCm Component

HIPIFY

Steps to Reproduce

cuda path: C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6
cudnn path: D:\software\cudnn-windows-x86_64-8.4.0.27_cuda11.6-archive
llvm path: D:\software\llvm-project-14.0.6.src\dist

by the way, I\ve added the installation path of the CUDA to the environment variable

(Optional for Linux users) Output of /opt/rocm/bin/rocminfo --support

No response

Additional Information

No response

Answer 1 · 2024-01-02T08:15:24.000Z

Specify CUDA path in double quotes as below
hipify-clang.exe D:\code\C++\CudaRuntime1\CudaRuntime1\kernel.cu --cuda-path="C:\Program Files\NVIDIA GPU Computing Toolkit"

Answer 2 · 2024-01-02T08:23:04.000Z

Specify CUDA path in double quotes as below hipify-clang.exe D:\code\C++\CudaRuntime1\CudaRuntime1\kernel.cu --cuda-path="C:\Program Files\NVIDIA GPU Computing Toolkit"

I've tried, but unfortunately, that doesn't work

Answer 3 · 2024-01-02T13:35:09.000Z

Please, use an exact path to your CUDA: --cuda-path="C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6".

Answer 4 · 2024-01-02T14:29:52.000Z

Please, use an exact path to your CUDA: --cuda-path="C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6".

It works! Thanks so much! However, I ran into a new problem.

The error is like:
error:
STL1002: Unexpected compiler version, expected CUDA 10.1 Update 2 or newer.
error: expected expression
addKernel << <1, size >> > (dev_c, dev_a, dev_b);

Is this because I didn't install CUDA properly? But cuda 11.6 works fine in my Visual Studio 2022.

Answer 5 · 2024-01-02T14:52:18.000Z

Could you provide an exact error log, please? Probably, the error is in space symbols in << < and >> >.

Answer 6 · 2024-01-02T15:21:40.000Z

Could you provide an exact error log, please? Probably, the error is in space symbols in << < and >> >.

Same as u think!

here is the exact error:
C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Tools\MSVC\14.30.30705\include\yvals_core.h:571:2: error:
STL1002: Unexpected compiler version, expected CUDA 10.1 Update 2 or newer.
#error STL1002: Unexpected compiler version, expected CUDA 10.1 Update 2 or newer.
^
C:\Users\DaisyDIE\AppData\Local\Temp\kernel.cu-51000a.hip:91:15: error: expected expression
addKernel << <1, size >> > (dev_c, dev_a, dev_b);
^
C:\Users\DaisyDIE\AppData\Local\Temp\kernel.cu-51000a.hip:91:27: error: expected expression
addKernel << <1, size >> > (dev_c, dev_a, dev_b);

Answer 7 · 2024-01-02T16:05:25.000Z

Could you please provide a whole hipify-clang output log (with the -v option) and its invocation command line?

Answer 8 · 2024-01-03T00:59:27.000Z

Could you please provide a whole hipify-clang output log (with the -v option) and its invocation command line?

Of course！

D:\software\Hipify\dist\bin>hipify-clang.exe D:\code\C++\CudaRuntime1\CudaRuntime1\kernel.cu --cuda-path="C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6"
warning: CUDA version is newer than the latest supported version 11.5 [-Wunknown-cuda-version]
In file included from :1:
In file included from D:/software/llvm-project-14.0.6.src/dist/lib/clang/14.0.6\include__clang_cuda_runtime_wrapper.h:41:
In file included from C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Tools\MSVC\14.30.30705\include\cmath:9:
In file included from C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Tools\MSVC\14.30.30705\include\yvals.h:9:
C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Tools\MSVC\14.30.30705\include\yvals_core.h:571:2: error: STL1002: Unexpected compiler version, expected CUDA
10.1 Update 2 or newer.
#error STL1002: Unexpected compiler version, expected CUDA 10.1 Update 2 or newer.
^
C:\Users\DaisyDIE\AppData\Local\Temp\kernel.cu-8e4c6f.hip:91:15: error: expected expression
addKernel << <1, size >> > (dev_c, dev_a, dev_b);
^
C:\Users\DaisyDIE\AppData\Local\Temp\kernel.cu-8e4c6f.hip:91:27: error: expected expression
addKernel << <1, size >> > (dev_c, dev_a, dev_b);
^
C:\Users\DaisyDIE\AppData\Local\Temp\kernel.cu-8e4c6f.hip:91:30: warning: left operand of comma operator has no effect [-Wunused-value]
addKernel << <1, size >> > (dev_c, dev_a, dev_b);
^~~~~
C:\Users\DaisyDIE\AppData\Local\Temp\kernel.cu-8e4c6f.hip:91:37: warning: left operand of comma operator has no effect [-Wunused-value]
addKernel << <1, size >> > (dev_c, dev_a, dev_b);
^~~~~
2 warnings and 3 errors generated when compiling for host.
Error while processing C:\Users\DaisyDIE\AppData\Local\Temp\kernel.cu-8e4c6f.hip.

Answer 9 · 2024-01-03T11:29:26.000Z

Thanks, but could you provide a verbose log (-v), please?

Answer 10 · 2024-01-03T11:38:07.000Z

Thanks, but could you provide a verbose log (-v), please?

Sorry, I didn't understand it before, but now I do.
Did u mean this :

D:\software\Hipify\dist\bin>hipify-clang.exe D:\code\C++\CudaRuntime1\CudaRuntime1\kernel.cu --cuda-path="C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6" -v
clang version 14.0.6
Target: x86_64-pc-windows-msvc
Thread model: posix
InstalledDir: D:\software\Hipify\dist\bin
Found CUDA installation: C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6, version
warning: CUDA version is newer than the latest supported version 11.5 [-Wunknown-cuda-version]
clang Invocation:
"D:\software\Hipify\dist\bin\clang-tool" "-cc1" "-triple" "x86_64-pc-windows-msvc19.30.30705" "-target-sdk-version=11.5" "-aux-triple" "nvptx64-nvidia-cuda" "-fsyntax-only" "-disable-free" "-clear-ast-before-backend" "-disable-llvm-verifier" "-discard-value-names" "-main-file-name" "kernel.cu-ffafab.hip" "-mrelocation-model" "pic" "-pic-level" "2" "-mframe-pointer=none" "-fmath-errno" "-ffp-contract=on" "-fno-rounding-math" "-mconstructor-aliases" "-funwind-tables=2" "-target-cpu" "x86-64" "-tune-cpu" "generic" "-mllvm" "-treat-scalable-fixed-error-as-warning" "-v" "-fcoverage-compilation-dir=D:\software\Hipify\dist\bin" "-resource-dir" "D:/software/llvm-project-14.0.6.src/dist/lib/clang/14.0.6" "-internal-isystem" "D:/software/llvm-project-14.0.6.src/dist/lib/clang/14.0.6\include\cuda_wrappers" "-include" "__clang_cuda_runtime_wrapper.h" "-I" "D:\code\C++\CudaRuntime1\CudaRuntime1" "-internal-isystem" "D:/software/llvm-project-14.0.6.src/dist/lib/clang/14.0.6\include" "-internal-isystem" "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Tools\MSVC\14.30.30705\include" "-internal-isystem" "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Tools\MSVC\14.30.30705\atlmfc\include" "-internal-isystem" "C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\ucrt" "-internal-isystem" "C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\shared" "-internal-isystem" "C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\um" "-internal-isystem" "C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\winrt" "-internal-isystem" "C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\cppwinrt" "-internal-isystem" "D:/software/llvm-project-14.0.6.src/dist/lib/clang/14.0.6\include" "-internal-isystem" "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Tools\MSVC\14.30.30705\include" "-internal-isystem" "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Tools\MSVC\14.30.30705\atlmfc\include" "-internal-isystem" "C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\ucrt" "-internal-isystem" "C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\shared" "-internal-isystem" "C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\um" "-internal-isystem" "C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\winrt" "-internal-isystem" "C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\cppwinrt" "-internal-isystem" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6/include" "-Wno-pragma-once-outside-header" "-std=c++14" "-fdeprecated-macro" "-fdebug-compilation-dir=D:\software\Hipify\dist\bin" "-ferror-limit" "19" "-fmessage-length=188" "-fno-use-cxa-atexit" "-fms-extensions" "-fms-compatibility" "-fms-compatibility-version=19.30.30705" "-fcxx-exceptions" "-fexceptions" "-fcolor-diagnostics" "-internal-isystem" "D:\software\Hipify\dist\bin/../include/hipify/cuda_wrappers" "-internal-isystem" "D:\software\Hipify\dist\bin/../include/hipify" "-internal-isystem" "D:\software\Hipify\dist\bin/include/cuda_wrappers" "-internal-isystem" "D:\software\Hipify\dist\bin/include" "-faddrsig" "-x" "cuda" "C:\Users\DaisyDIE\AppData\Local\Temp\kernel.cu-ffafab.hip"

clang -cc1 version 14.0.6 based upon LLVM 14.0.6 default target x86_64-pc-windows-msvc
ignoring nonexistent directory "D:\software\Hipify\dist\bin/../include/hipify/cuda_wrappers"
ignoring nonexistent directory "D:\software\Hipify\dist\bin/../include/hipify"
ignoring nonexistent directory "D:\software\Hipify\dist\bin/include/cuda_wrappers"
ignoring nonexistent directory "D:\software\Hipify\dist\bin/include"
ignoring duplicate directory "D:/software/llvm-project-14.0.6.src/dist/lib/clang/14.0.6\include"
ignoring duplicate directory "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Tools\MSVC\14.30.30705\include"
ignoring duplicate directory "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Tools\MSVC\14.30.30705\atlmfc\include"
ignoring duplicate directory "C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\ucrt"
ignoring duplicate directory "C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\shared"
ignoring duplicate directory "C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\um"
ignoring duplicate directory "C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\winrt"
ignoring duplicate directory "C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\cppwinrt"
#include "..." search starts here:
#include <...> search starts here:
D:\code\C++\CudaRuntime1\CudaRuntime1
D:/software/llvm-project-14.0.6.src/dist/lib/clang/14.0.6\include\cuda_wrappers
D:/software/llvm-project-14.0.6.src/dist/lib/clang/14.0.6\include
C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Tools\MSVC\14.30.30705\include
C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Tools\MSVC\14.30.30705\atlmfc\include
C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\ucrt
C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\shared
C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\um
C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\winrt
C:\Program Files (x86)\Windows Kits\10\Include\10.0.22000.0\cppwinrt
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6/include
End of search list.
In file included from :1:
In file included from D:/software/llvm-project-14.0.6.src/dist/lib/clang/14.0.6\include__clang_cuda_runtime_wrapper.h:41:
In file included from C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Tools\MSVC\14.30.30705\include\cmath:9:
In file included from C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Tools\MSVC\14.30.30705\include\yvals.h:9:
C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Tools\MSVC\14.30.30705\include\yvals_core.h:571:2: error: STL1002: Unexpected compiler version, expected CUDA 10.1 Update 2 or
newer.
#error STL1002: Unexpected compiler version, expected CUDA 10.1 Update 2 or newer.
^
C:\Users\DaisyDIE\AppData\Local\Temp\kernel.cu-ffafab.hip:91:15: error: expected expression
addKernel << <1, size >> > (dev_c, dev_a, dev_b);
^
C:\Users\DaisyDIE\AppData\Local\Temp\kernel.cu-ffafab.hip:91:27: error: expected expression
addKernel << <1, size >> > (dev_c, dev_a, dev_b);
^
C:\Users\DaisyDIE\AppData\Local\Temp\kernel.cu-ffafab.hip:91:30: warning: left operand of comma operator has no effect [-Wunused-value]
addKernel << <1, size >> > (dev_c, dev_a, dev_b);
^~~~~
C:\Users\DaisyDIE\AppData\Local\Temp\kernel.cu-ffafab.hip:91:37: warning: left operand of comma operator has no effect [-Wunused-value]
addKernel << <1, size >> > (dev_c, dev_a, dev_b);
^~~~~
2 warnings and 3 errors generated when compiling for host.
Error while processing C:\Users\DaisyDIE\AppData\Local\Temp\kernel.cu-ffafab.hip.

Answer 11 · 2024-01-03T12:16:05.000Z

Thanks. And what version of Microsoft Visual Studio 2022 are you using?

Answer 12 · 2024-01-03T12:22:24.000Z

Thanks. And what version of Microsoft Visual Studio 2022 are you using?

Thanks. I'm using Microsoft Visual Studio Enterprise 2022 (x64) - Current version 17.0.0

Answer 13 · 2024-01-03T13:45:16.000Z

So, it looks like your configuration of LLVM 14.0.6 + MSVC 17.0.0 is not quite supported: I see an STL mismatch, first. And this configuration has not been tested on our side: https://github.com/ROCm/HIPIFY/blob/amd-staging/docs/hipify-clang.md#hipify-clang-windows-testing.
Additionally, kernel invocation syntax is incorrect, as it contains space symbols.

The best solution here is to 1) fix the sources by removing the abovementioned spaces; 2) use a newer LLVM version (better, 17.x); 3) update your VS 2022 to the latest version (17.8.x).

Answer 14 · 2024-01-03T14:50:38.000Z

So, it looks like your configuration of LLVM 14.0.6 + MSVC 17.0.0 is not quite supported: I see an STL mismatch, first. And this configuration has not been tested on our side: https://github.com/ROCm/HIPIFY/blob/amd-staging/docs/hipify-clang.md#hipify-clang-windows-testing. Additionally, kernel invocation syntax is incorrect, as it contains space symbols.

The best solution here is to 1) fix the sources by removing the abovementioned spaces; 2) use a newer LLVM version (better, 17.x); 3) update your VS 2022 to the latest version (17.8.x).

Thanks so much! I'll get back to you after I've tried 2) and 3)!
I still don't quite understand 1) . could you please give me an example of how to use it correctly?

Answer 15 · 2024-01-04T11:18:53.000Z

addKernel << <1, size >> > (dev_c, dev_a, dev_b);

is incorrect; should be:

addKernel <<<1, size >>> (dev_c, dev_a, dev_b);

Answer 16 · 2024-01-04T12:09:26.000Z

addKernel << <1, size >> > (dev_c, dev_a, dev_b);

is incorrect; should be:

addKernel <<<1, size >>> (dev_c, dev_a, dev_b);

Thanks! I'll give it a try.

Answer 17 · 2024-01-04T14:56:24.000Z

So, it looks like your configuration of LLVM 14.0.6 + MSVC 17.0.0 is not quite supported: I see an STL mismatch, first. And this configuration has not been tested on our side: https://github.com/ROCm/HIPIFY/blob/amd-staging/docs/hipify-clang.md#hipify-clang-windows-testing. Additionally, kernel invocation syntax is incorrect, as it contains space symbols.

The best solution here is to 1) fix the sources by removing the abovementioned spaces; 2) use a newer LLVM version (better, 17.x); 3) update your VS 2022 to the latest version (17.8.x).

Thanks for your responsibility and kindness! Your suggestion worked!

Answer 18 · 2024-01-04T14:58:06.000Z

Thanks for the help!