Commit Graph
8 Commits
Author SHA1 Message Date
Salvatore Sanfilippo fc8bf3c39c Add DeepSeek V4 Flash vision support 2026-09-01 00:11:58 +02:00
antirez b7e9f00911 Version DeepSeek Flash fixtures by checkpoint 2026-08-03 19:25:05 +02:00
antirez 005afedcd9 Add GLM 5.2 inference and quality fixtures 2026-07-18 18:48:45 +02:00
antirez 8384adf0f9 Fix Metal SSD streaming cache reuse
Track routed-expert cache entries referenced by encoded Metal command buffers and prevent cache eviction or buffer reuse until the owning command buffers have completed. This preserves layer-batched decode while avoiding stale buffer captures under SSD-streaming cache pressure.

Add a focused Metal SSD-streaming cache-pressure regression for issue #384.
2026-06-10 20:20:48 +02:00
antirez 17502b9af3 Add local golden inference drift test 2026-05-28 17:26:22 +02:00
Salvatore Sanfilippo 18c2d4bd79 NAX speedups and Metal4 cleanup
Squashes Salvatore Sanfilippo's Metal4/NAX speedups and cleanup work: direct-RHS dense Tensor kernels, routed MoE Tensor coverage, full-512 indexer preservation, NAX prefill indexer scores, indexed-attention half shadow cache, and removal of rejected experimental paths.
2026-05-23 11:43:29 +02:00
Ivan Fioravanti 63ceed6aca Metal Neural Acceleration initial implementation
Squashes Ivan Fioravanti's Metal4/M5 Neural Acceleration scaffold, benchmark tooling, drift diagnostics, eval trace regrading, and initial Tensor/MPP kernel work.
2026-05-23 11:42:59 +02:00
antirez d997b56c15 DS4 initial release 2026-05-07 17:16:28 +02:00