mirror of
https://github.com/ggerganov/llama.cpp.git
synced 2025-04-22 22:56:05 +00:00
0090950f67
q4_k and q5_k had a lot of redundant global loads where the same 16B of scale information is repeatedly loaded and decoded during each loop iteration. This change restructures the loops to more explicitly iterate over whole blocks in the outer loop (with unrolled inner loop) and to copy/decode the scale data into shared memory once at the start of each outer loop. The copy is pipelined so the scale load from global memory is relatively cheap. This improves q4_k/q5_k model prompt processing performance by around 5-7%. I briefly tried applying this to q6_k and q4_0, and it didn't help for q6_k and hurt for q4_0. The big "else" path in mul_mm_cm2.comp that had all the clamped/unclamped variants isn't used as often as it originally was (e.g. due to the padded_N change), so I trimmed it down to offset some of the new complexity of the semi-manual loop unrolling.