Home › Evidence › Records › con-0002
⚠ This notebook has stopped taking notes: newest record is 23 days old (eng-0280), against an expected cadence of 14 days.

con-0002

carrying
citable URL: https://halobench.com/records/con-0002/ — this address never moves; the anchor /records/#con-0002 keeps resolving
kind fork · upstream ggml-org/llama.cpp
opened 2026-08-18 · https://github.com/Nathanw1014/strix-halo-llamacpp
carrying — reproducibility hazard: in use but not upstream. Every config listing this id in its divergence runs a fork.
Nathanw1014's Strix Halo Vulkan fork, staged first as the v0.6.4 portable payload at build baf6360be and then as the reviewed v0.6.6 portable payload at build 7b6c6133/source 7b6c61330edf370659f531932e0b91aca67ba055. The payloads carry RADV wave32/cooperative-matrix tuning, dense and sparse prefill work, DeepSeek-V4 Lightning Indexer support and upstream merges. Results produced with them are fork-specific and must not be attributed to stock llama.cpp or to any one bundled patch.