1. 02 May, 2025 1 commit
    • Jesse Gross's avatar
      ollamarunner: Re-enable worst case graph preallocation. · c2f5d666
      Jesse Gross authored
      Worst case graph preallocation was disabled by a27462b7
      "ollamarunner: Temporarily disable worst case graph preallocation"
      since it caused crashes with large batches when not using the GPU.
      
      This backports upstream llama.cpp commit f057808
      "ggml: Don't assert fail when tensor data changes (#13222)", which
      fixes the underlying bug and allows reverting the previous workaround.
      c2f5d666