Mozilla-Ocho / llamafile

Distribute and run LLMs with a single file.
https://llamafile.ai
Other
20.58k stars 1.04k forks source link

Bug: 'cmath' file not found on Windows with AMD Dedicated GPU #601

Open DK013 opened 4 weeks ago

DK013 commented 4 weeks ago

Contact Details

chakraborty.deep013@gmail.com

What happened?

When --gpu flag explicitly set for AMD the application crashes with the following error:

In file included from <built-in>:1:
In file included from C:\Program Files\AMD\ROCm\5.7\lib\clang\17.0.0\include\__clang_hip_runtime_wrapper.h:50:
C:\Program Files\AMD\ROCm\5.7\lib\clang\17.0.0\include\cuda_wrappers\cmath:27:15: fatal error: 'cmath' file not found
#include_next <cmath>
              ^~~~~~~
1 error generated when compiling for gfx1010.
fatal error: support for --gpu amd was explicitly requested, but it wasn't available

However, as my first instinct tell me to go on and check if the file actually exist or not, I can find it right there in C:\Program Files\AMD\ROCm\5.7\lib\clang\17.0.0\include\cuda_wrappers.

On my second instinct, I tried running the same on an elevated terminal instance, just to be sure that since the file is located under Program Files and may need Admin rights (though shouldn't need admin rights to read the file only), The same error pops up. So there may be something fundamentally wrong when the app is trying to read the file.

My System for relevance: OS: Windows 11 (build 22631.4391) GPU: AMD Radeon RX 5700 XT GPU Driver: 24.10.1 ROCm and clang versions are in the log

Version

llamafile v0.8.14

What operating system are you seeing the problem on?

Windows

Relevant log output

No response