Stars
2
stars
written in Cuda
Clear filter
Mirage Persistent Kernel: Compiling LLMs into a MegaKernel
A CUDA implementation of SIFT for NVidia GPUs (1.2 ms on a GTX 1060)