Repository navigation
Conversation
The stages before training (frames, select, mask, sfm) now run on an Apple silicon Mac, by hand and outside the queue. Training stays on CUDA. - scripts/sift_backend.py picks the SIFT backend (cuda, metal, cpu; SPLAT_SIFT overrides). The CUDA call is unchanged, so no cache key moves. On the Metal build extraction runs on the GPU and matching on the CPU. - scripts/setup_mac.sh builds pycolmap from pgodlews/colmap, branch metal-sift-lanxinger, pinned by commit, and a venv_gs with torch 2.14.1: torchvision 0.24.1's roi_align is unusable on MPS. - 81_fisheye_stitch.py builds its float64 grids on the CPU for MPS; 70_person_masks.py runs on MPS; frame decoding uses VideoToolbox on macOS (SPLAT_HWACCEL overrides). 30_run_sfm.py refuses to run without CUDA. - Docs: "Prep on Apple silicon" in how-it-works.md with the measurements (clips 0141 and 0005, trained against CUDA preps), troubleshooting #37 and #38, install and README pointers. Measured: trained PSNR 27.109 vs 27.010 (0141) and 21.990 vs 22.027 (0005) against the CUDA prep of the same job. Not done: queue scheduling on a Mac, and writing a handoff bundle from stages run by hand.
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The stages before training (frames, select, mask, sfm) now run on an Apple silicon Mac, by hand and outside the queue. Training stays on CUDA.
Measured: trained PSNR 27.109 vs 27.010 (0141) and 21.990 vs 22.027 (0005) against the CUDA prep of the same job. Not done: queue scheduling on a Mac, and writing a handoff bundle from stages run by hand.