llama.cpp
3fe81781 - CUDA: faster q8_0 -> f16 dequantization (#4895)

Commit

1 year ago

CUDA: faster q8_0 -> f16 dequantization (#4895)

References

#4895 - CUDA: faster q8_0 -> f16 dequantization

Author

JohannesGaessler

JohannesGaessler

Parents

Loading