Drop stale param refs in _update_param_with_optimizer to cut VRAM spike#931
Open
Evggamgam wants to merge 3 commits intonerfstudio-project:mainfrom
Open
Drop stale param refs in _update_param_with_optimizer to cut VRAM spike#931Evggamgam wants to merge 3 commits intonerfstudio-project:mainfrom
Evggamgam wants to merge 3 commits intonerfstudio-project:mainfrom
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
_update_param_with_optimizerkeeps the oldParameteralive via three references (local var,optimizer.statekey,param_groups["params"]) while new Adam state tensors are allocated, causing a per-densify transient of ~1×size(largest param)above steady state. Rebind those refs tonew_paramanddelthe local beforeoptimizer_fn, so the old tensor is freed during the state reshape.Pure reordering — output
Parameterand Adam state are bit-identical to the pre-patch implementation; wall time unchanged.Max Gaussians in the densify-step budget on a 24 GB card (excluding render/backward overhead):