llama.cpp
hexagon: support for UPSCALE op
#31331
Merged

hexagon: support for UPSCALE op #31331

max-krasnyansky
max-krasnyansky hex-upscale: basic upscale ops (not fully vectorized)
5d609a71
max-krasnyansky hex-upscale: optimize nearest 2x, 4x and bilinear 4x
4e61c2c7
max-krasnyansky hex-upscale: add support for non-contig rows and dims collapsing
8c945696
max-krasnyansky hex-upscale: use vgather instead of scalar staging
413a40d3
max-krasnyansky hex-upscale: move all kernel select logic to the host, review comment…
0804f941
max-krasnyansky tests: add permuted UPSCALE test to cover qwen3-vl image->text scenarios
137be776
max-krasnyansky max-krasnyansky requested a review from ggerganov ggerganov 11 days ago
max-krasnyansky max-krasnyansky requested a review 11 days ago
github-actions github-actions added testing
github-actions github-actions added ggml
github-actions github-actions added Hexagon
lhez
lhez approved these changes on 2026-10-28
max-krasnyansky max-krasnyansky added merge ready
max-krasnyansky
jhen0409
jhen0409 approved these changes on 2026-10-29
max-krasnyansky max-krasnyansky merged 8a199ad3 into master 10 days ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
Milestone