feat(python): gpu based ivf partition training #1361
Merged
Conversation
This file contains 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
eddyxu
requested review from
wjones127,
changhiskhan and
westonpace
and removed request for
wjones127 and
changhiskhan
python/python/lance/vector.py
Outdated
| else: | ||
| samples = dataset.sample(k * sample_rate)[column] | ||
|
|
||
| if accelerator in ["gpu", "cuda"]: |
There was a problem hiding this comment.
nit: gpu could be MPS or AMD as well, I don't think we should map gpu to cuda.
There was a problem hiding this comment.
Yea, so i was thinking that use gpu to call preferred_device() later to auto detect GPU on the machine
But it is fair that we don't need to do it now. As mps performance is not good at the moment.
| column: str, | ||
| k: int, | ||
| metric_type: str, | ||
| accelerator: str, |
There was a problem hiding this comment.
nit: allow device id here like cuda:0.
chebbyChefNEQ
approved these changes
Oct 5, 2023
westonpace
approved these changes
Oct 5, 2023
| *, | ||
| sample_rate: int = 256, | ||
| ) -> np.ndarray: | ||
| """Use accelerator (GPU or MPS) to train kmeans.""" |
There was a problem hiding this comment.
Is MPS actually supported currently?
There was a problem hiding this comment.
yes, we can run mps today, it is just not as fast as we desired.
python/python/lance/vector.py
Outdated
| sample_rate: int = 256, | ||
| ) -> np.ndarray: | ||
| """Use accelerator (GPU or MPS) to train kmeans.""" | ||
|
|
There was a problem hiding this comment.
nit: should we check for torch is installed before trying to do all the sampling?
Co-authored-by: Weston Pace <[email protected]>
Co-authored-by: Weston Pace <[email protected]>
Co-authored-by: Weston Pace <[email protected]>
wjones127
approved these changes
Oct 5, 2023
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.
Use pytorch to train IVF partitions on GPU
ds.create_index(..., accelerator="cuda")