Skip to content

Windows single threaded version compiled with mingw64 does not work #1253

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
seechuenteck opened this issue Jul 27, 2017 · 8 comments
Closed

Comments

@seechuenteck
Copy link

I have used 'make USE_THREAD=0 USE_OPENMP=0' on 0.2.19 & 0.2.20 & compiled on mingw64 using msys2 & it does not work. Compiling using 'make' works fine on windows.

@martin-frbg
Copy link
Collaborator

How does "it does not work" manifest itself ? Do you get errors while compiling, or does the resulting library fail in any way ?

@brada4
Copy link
Contributor

brada4 commented Jul 27, 2017

Did you run 'make clean' between builds with different options?

@seechuenteck
Copy link
Author

Yes, I did a clean build. No compliation error. The program crashes when runned. In unix it is ok.

In addition, when I compiled openblas on mingw64 using 'make' & then runned using SET "OPENBLAS_NUM_THREADS=1", the program crashes. However, "OPENBLAS_NUM_THREADS=2" works fine.

@martin-frbg
Copy link
Collaborator

Which version of mingw, and did you specify any other options apart from USE_THREAD and USE_OPENMP ?

@seechuenteck
Copy link
Author

I have compiled using mSYS2. The version was mingw-w64-gcc 7.1.0 (x86_64). As described eariler, using 'make' & setting "OPENBLAS_NUM_THREADS=1", the program crashes. However, "OPENBLAS_NUM_THREADS=2" works fine.

@martin-frbg
Copy link
Collaborator

Can you tell if it is a specific function call that crashes ? (We do not know what your program is or does)

@seechuenteck
Copy link
Author

The single thread version on win hangs when dpotrf_(..) is called. Openblas is compiled on mingw64 as a dll & linked to program in msvc. > 1 thread works fine.

@brada4
Copy link
Contributor

brada4 commented Aug 1, 2017

Can you check with a debugger where exactly it hung?
Like short backtrace of calls (with arguments) and disasm around IP, and module file layout in RAM?
trmv called behind the scenes was corrected after 0.2.20 - can you try develop branch?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants