llama.cpp

ClaudeBot/llama.cpp

Fork 1

mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-09-17 20:31:47 +02:00

Commit Graph

Author	SHA1	Message	Date
Marco ColomboandMax Krasnyansky	1ec8188094	hexagon: Support for K-Quants Q4_K and Q6_K (#28994 ) implement q6k/q4k kernels Squashed from: feat: implement q6k kernel hex-q6k: improve unpack accuracy hex-q4_k: add support for Q4_K kernels Co-authored-by: Max Krasnyansky <maxk@qti.qualcomm.com>	2026-09-16 09:00:31 -07:00
Marco Colombo	82324fc508	hexagon: accept the zeroed rope probe in supports_op (#28995 ) llama probes weight placement with a rope where all params are 0, so rejecting n_dims == 0 or freq_base == 0 puts rope_freqs on the CPU. That splits the decode graph at every full-attention layer (gemma-4-E2B: 5 splits instead of 2). Assisted-by: Claude Opus 5	2026-09-16 08:38:44 -07:00

Author

SHA1

Message

Date

Marco ColomboandMax Krasnyansky

1ec8188094

hexagon: Support for K-Quants Q4_K and Q6_K (#28994 )

implement q6k/q4k kernels

Squashed from:
  feat: implement q6k kernel
  hex-q6k: improve unpack accuracy
  hex-q4_k: add support for Q4_K kernels

Co-authored-by: Max Krasnyansky <maxk@qti.qualcomm.com>

2026-09-16 09:00:31 -07:00

Marco Colombo

82324fc508

hexagon: accept the zeroed rope probe in supports_op (#28995 )

llama probes weight placement with a rope where all params are 0, so rejecting
n_dims == 0 or freq_base == 0 puts rope_freqs on the CPU. That splits the decode
graph at every full-attention layer (gemma-4-E2B: 5 splits instead of 2).

Assisted-by: Claude Opus 5

2026-09-16 08:38:44 -07:00