[Executorch][Llama] Dont memory plan for inputs #2155
Closed
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
For KV cache with IO tHis results in 1. allocating kv cache in the memory plan but also allocated by llama runner 2. Doing actual copy of kv cache Also we should really make plan_input = false by default. I dont imagine a case where this does not result in making copies. Planning for output is fine but still dangerous as people may assume having reference to output tensor is all good without realizing the underlying memory being shared. Differential Revision: [D54161288](https://our.internmc.facebook.com/intern/diff/D54161288/) [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/2155
Note: Links to docs will display an error until the docs builds have been completed. ❌ 4 New FailuresAs of commit 5682fe7 with merge base a78a07e ( NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
This was referenced Feb 28, 2024
|
This pull request was exported from Phabricator. Differential Revision: D54161288 |
mergennachin
approved these changes
Feb 28, 2024
For KV cache with IO tHis results in 1. allocating kv cache in the memory plan but also allocated by llama runner 2. Doing actual copy of kv cache Also we should really make plan_input = false by default. I dont imagine a case where this does not result in making copies. Planning for output is fine but still dangerous as people may assume having reference to output tensor is all good without realizing the underlying memory being shared. Differential Revision: [D54161288](https://our.internmc.facebook.com/intern/diff/D54161288/) [ghstack-poisoned]
For KV cache with IO tHis results in 1. allocating kv cache in the memory plan but also allocated by llama runner 2. Doing actual copy of kv cache Also we should really make plan_input = false by default. I dont imagine a case where this does not result in making copies. Planning for output is fine but still dangerous as people may assume having reference to output tensor is all good without realizing the underlying memory being shared. Differential Revision: [D54161288](https://our.internmc.facebook.com/intern/diff/D54161288/) [ghstack-poisoned]
For KV cache with IO tHis results in 1. allocating kv cache in the memory plan but also allocated by llama runner 2. Doing actual copy of kv cache Also we should really make plan_input = false by default. I dont imagine a case where this does not result in making copies. Planning for output is fine but still dangerous as people may assume having reference to output tensor is all good without realizing the underlying memory being shared. Differential Revision: [D54161288](https://our.internmc.facebook.com/intern/diff/D54161288/) [ghstack-poisoned]
For KV cache with IO tHis results in 1. allocating kv cache in the memory plan but also allocated by llama runner 2. Doing actual copy of kv cache Also we should really make plan_input = false by default. I dont imagine a case where this does not result in making copies. Planning for output is fine but still dangerous as people may assume having reference to output tensor is all good without realizing the underlying memory being shared. Differential Revision: [D54161288](https://our.internmc.facebook.com/intern/diff/D54161288/) [ghstack-poisoned]
|
This pull request has been merged in 76cbfb7. |
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.
Stack from ghstack (oldest at bottom):
For KV cache with IO tHis results in
Also we should really make plan_input = false by default. I dont imagine a case
where this does not result in making copies. Planning for output is fine but
still dangerous as people may assume having reference to output tensor is all
good without realizing the underlying memory being shared.
Differential Revision: D54161288