llama.cpp
141a908a - CUDA: mix virt/real CUDA archs for GGML_NATIVE=OFF (#13135)

Commit

135 days ago

CUDA: mix virt/real CUDA archs for GGML_NATIVE=OFF (#13135)

References

#13135 - CUDA: build archs as virtual for GGML_NATIVE=OFF

Author

JohannesGaessler

JohannesGaessler

Parents

Loading