llama.cpp
483bab2e - Avoid the transposed X branch in the Z = X * Y matrix multiplication (#439)

Commit

2 years ago

Avoid the transposed X branch in the Z = X * Y matrix multiplication (#439) Should make results reproducible for different number of threads and batch sizes

References

#439 - Avoid the "non-contiguous X" branch in the Z = X * Y matrix multiplication

Author

ggerganov

Parents

404e1da3

llama.cpp 483bab2e - Avoid the transposed X branch in the Z = X * Y matrix multiplication (#439)

llama.cpp
483bab2e - Avoid the transposed X branch in the Z = X * Y matrix multiplication (#439)