ggml: add conv3d op
#15182
Merged
ggml: add conv3d op
#15182
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
github-actions
bot
added
testing
ggerganov
approved these changes
Aug 22, 2025
6 tasks
qnixsynapse
pushed a commit
to menloresearch/llama.cpp
that referenced
this pull request
Aug 25, 2025
3 tasks
Draft
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR adds
conv3dsupport for the CPU backend. This is required for tasks such as inferencing text to video models.Since we are limited by the number of tensor dimensions, my idea was to unroll the depth dimension. The implementation flattens 3D input patches into columns, then performs a single matrix multiplication against the flattened kernel weights.
I did not seek for optimizations, and I'm not sure it is worth it (unless there are obvious ones), as this implementation will essentially serve as a ground truth to test against when implementing this op on other backends.
The correctness was checked against PyTorch's native
conv3d