perf(cuda): the recurrent-state GET_ROWS folds into GATED_DELTA_NET's load, f32 or q8_0 cache (#220 plus -cts q8_0) - #31

perf(cuda): the recurrent-state GET_ROWS folds into GATED_DELTA_NET's load, f32 or q8_0 cache (#220 plus -cts q8_0)#31

marcospaulo merged 4 commits into

maintorad-labs/llama.cpp:mainfrom

perf/gdn-state-gather-foldtorad-labs/llama.cpp:perf/gdn-state-gather-foldCopy head branch name to clipboard