Skip to content

hexagon: support I32 CPY and CONT - #29379

Merged
max-krasnyansky merged 1 commit into
ggml-org:masterfrom
qualcomm:hexagon-i32-cont
Sep 24, 2026
Merged

max-krasnyansky merged 1 commit into
ggml-org:masterfrom
qualcomm:hexagon-i32-cont

Conversation

@kurquhar

Copy link
Copy Markdown
Contributor

Overview

Add I32 CPY and CONT support to the hexagon backend

Additional information

Motivation is bringing the main graph ops of the LFM2.5-DSpark draft model onto hexagon, instead of using the CPU fallback.

Mostly this was an exercise of extending the current FP32 and FP16 support to also include INT32->INT32.

Validation commands:

python scripts/snapdragon/run.py --install-dir pkg-wos-i32-cont --devices HTP0 -- test-backend-ops -o CONT -b HTP0
python scripts/snapdragon/run.py --install-dir pkg-wos-i32-cont --devices HTP0 -- test-backend-ops -o CPY -b HTP0

Before: 12 CONT and 11 CPY test cases not supported
After: same 23 test cases [OK]

The corresponding test cases for FP32 and FP16 continue to pass, as before.

Requirements

  • I have read and agree with the contributing guidelines Yes
  • AI usage disclosure: Yes, for reasoning, code changes, and testing. All code was reviewed by me (human).

Assisted-by: OpenCode

Assisted-by: OpenCode
@kurquhar
kurquhar requested a review from a team as a code owner September 24, 2026 14:55
@max-krasnyansky

Copy link
Copy Markdown
Member

@lhez please review/ack. I'll wait for CI to clear and merge.

@github-actions github-actions Bot added ggml changes relating to the ggml tensor library for machine learning Hexagon labels Sep 24, 2026
@max-krasnyansky
max-krasnyansky merged commit 97a418b into ggml-org:master Sep 24, 2026
18 checks passed
frostyautumnleaf pushed a commit to frostyautumnleaf/llama.cpp that referenced this pull request Oct 5, 2026
edwardyoon pushed a commit to edwardyoon/focus-llama that referenced this pull request Oct 7, 2026
Assisted-by: OpenCode
(cherry picked from commit 97a418b)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ggml changes relating to the ggml tensor library for machine learning Hexagon

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants