Bug: MinGW build fails to load models with "error loading model: PrefetchVirtualMemory unavailable" #9311

Activity

1. jeberger

   added

   bug-unconfirmed

   high severityUsed to report high severity bugs in llama.cpp (Malfunctioning hinder important workflow)Used to report high severity bugs in llama.cpp (Malfunctioning hinder important workflow)

   on Sep 4, 2024
2. jeberger commented on Sep 4, 2024

   

   jeberger

   on Sep 4, 2024

   Author

   More actions

   I believe the issue was introduced when the source was reorganized into folders. I'm able to work around it by adding the following to the top level CMakeLists.txt:

   if (MINGW)
       add_compile_definitions(_WIN32_WINNT=0x602)
   endif()
   

   so I believe the equivalent code is missing from some CMakeLists.txt files in subfolders (I tried adding it to examples/llama-bench/CMakeLists.txt but that wasn't enough to fix the issue).
3. homenkovo commented on Sep 20, 2024

   

   homenkovo

   on Sep 20, 2024

   More actions

   I had a similar problem. I just replaced #define _WIN32_WINNT 0x0601 with #define _WIN32_WINNT 0x0A00 in the <_mingw.h> file and it worked.

   ❤️React with ❤️2Reacted by osadchi and Imsisukon
4. github-actions

   added

   stale

   on Oct 21, 2024
5. github-actions commented on Nov 4, 2024

   

   github-actionsBot

   on Nov 4, 2024 – with GitHub Actions

   Contributor

   More actions

   This issue was closed because it has been inactive for 14 days since being marked as stale.
6. github-actions

   closed this as completedon Nov 4, 2024
7. theSoberSobber commented on Jan 4, 2025

   

   theSoberSobber

   on Jan 4, 2025

   More actions

   Why was this closed, I suffered the same error as this in current latest commit and fixed it using @homenkovo's solution
8. watkyns commented on Feb 4, 2025

   

   watkyns

   on Feb 4, 2025

   More actions

   having same issue with b4628 and homenkovo's solution can fix it
9. 0dminnimda commented on Feb 5, 2025

   

   0dminnimda

   on Feb 5, 2025

   More actions

   No need to modify no files, a simple command line argument is enough:

   -DCMAKE_CXX_FLAGS=-D_WIN32_WINNT=0x602
   

   👍React with 👍3Reacted by Jeff Zhou, Pu Zhang and julmich

   ❤️React with ❤️3Reacted by Vishal Ahirwar, Pu Zhang and julmich
10. robbyvids commented on Feb 12, 2025

    

    robbyvids

    on Feb 12, 2025

    More actions

    not worked on win 11 for me, using python script. Memory more than enough, it happen when loading to gpu or cpu.

    llama_model_load: error loading model: PrefetchVirtualMemory unavailable
11. vishal-ahirwar commented on Feb 17, 2025

    

    vishal-ahirwar

    on Feb 17, 2025

    More actions

    No need to modify no files, a simple command line argument is enough:

    -DCMAKE_CXX_FLAGS=-D_WIN32_WINNT=0x602
    

    thanks it worked for me
12. Pigoomy commented on May 4, 2025

    

    Pigoomy

    on May 4, 2025

    More actions

    I encountered the same PrefetchVirtualMemory unavailable error with llama-cpp-python on Windows. Disabling memory mapping via use_mmap=False resolved it instantly!
    Hope this saves someone hours of debugging🛠️

    🎉React with 🎉1Reacted by Pigoomy
13. julmich commented on Jun 4, 2025

    

    julmich

    on Jun 4, 2025

    More actions

    PrefetchVirtualMemory unavailable error with llama-cpp on Windows - b5514 sources - compiling with llvm-mingw_64

    this CMAKE line share from 0dminnimda above fix the issue
    -DCMAKE_CXX_FLAGS=-D_WIN32_WINNT=0x602
14. slaren

    reopened this on Jun 4, 2025
15. slaren

    added

    bugSomething isn't workingSomething isn't working

    and removed

    bug-unconfirmed

    stale

    on Jun 4, 2025
16. slaren

    closed this as completedin #14013on Jun 5, 2025