llama.cpp
hexagon: add support for SSM_SCAN op
#31921
Merged

hexagon: add support for SSM_SCAN op #31921

tboinovski1
github-actions github-actions added testing
github-actions github-actions added ggml
github-actions github-actions added Hexagon
tboinovski1 tboinovski1 force pushed from c708d31b to caa86004 6 days ago
tboinovski1 tboinovski1 force pushed from 9126417e to e6682309 6 days ago
tboinovski1 tboinovski1 marked this pull request as ready for review 6 days ago
tboinovski1 tboinovski1 requested a review from ggerganov ggerganov 6 days ago
tboinovski1 tboinovski1 requested a review 6 days ago
github-actions github-actions added documentation
tboinovski1 tboinovski1 force pushed from ae091cb2 to 865b7976 3 days ago
max-krasnyansky
tboinovski1 hexagon: add SSM_SCAN Op support
6db7318b
tboinovski1 hex-ssm-scan: fix multi-device spliting
5d246692
tboinovski1 hex-ssm-scan: remove redundant checks and make kernel use host config
ab4e5905
tboinovski1 hex-ssm-scan: remove direct ddr access
9d064764
tboinovski1 hex-ssm-scan: remove DMA ordering
6bbbd6cb
tboinovski1 hex-ops: update supported ops
a0ab382a
max-krasnyansky hex-ssm-scan: rewrite the dma pipeline to match other ops (dbl.buf, etc)
941f2514
max-krasnyansky hex-ssm-scan: remove all fallbacks, only dma/vtcm paths
b7aa84ff
max-krasnyansky hex-ssm-scan: add kernel_type to kparams and remove redundant recomputes
430c4356
max-krasnyansky hex-ssm-scan: avoid stalls in the overall pipeline (dma flush between…
5a0c3c2a
max-krasnyansky hex-ssm-scan: vectorize softplus
28047e37
max-krasnyansky hex-ssm-scan: optimize hvx loops
8b12aa4f
max-krasnyansky hex-ssm-scan: add HMX support
fe481938
max-krasnyansky hex-ssm-scan: add support for GGML_HEXAGON_SSM_SELECT
466f245e
max-krasnyansky hex-ssm-scan: cleanup kernel names
164c7ff1
max-krasnyansky hex-ssm-scan: simplify mdev checks
0c5684de
max-krasnyansky hex-mdev: improve mdev row-split solver
08a35a88
max-krasnyansky max-krasnyansky force pushed from 865b7976 to 08a35a88 2 days ago
max-krasnyansky
max-krasnyansky approved these changes on 2026-11-13
max-krasnyansky
lhez
lhez approved these changes on 2026-11-13
max-krasnyansky max-krasnyansky merged ded522f8 into master 1 day ago

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone