Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Minor blocksparse refactoring, update block size restrictions, relax power of two constraint #277

Merged
merged 3 commits into from
Apr 20, 2022
Merged

Minor blocksparse refactoring, update block size restrictions, relax power of two constraint #277

merged 3 commits into from
Apr 20, 2022

Conversation

colehawkins
Copy link
Contributor

@colehawkins colehawkins commented Apr 20, 2022 •
edited
Loading

What does this PR do?

Minor blocksparse refactoring, update block size restrictions, relax power of two constraint.

Before submitting

  • [ X] Did you have fun?
    • Make sure you had fun coding 🙃
  • [ X] Did you read the contributor guideline?
  • Was this discussed/approved via a Github issue? (no need for typos, doc improvements)
    • N/A
  • Did you make sure to update the docs?
    • N/A
  • [ X] Did you write any new necessary tests?
    • N/A
  • Did you update the changelog? (if needed)
    • N/A

PR review

Anyone in the community is free to review the PR once the tests have passed.
If we didn't discuss your PR in Github issues there's a high chance it will not be merged.

@facebook-github-bot facebook-github-bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Apr 20, 2022
@colehawkins colehawkins mentioned this pull request Apr 20, 2022
Merged
15 tasks
blefaudeux
blefaudeux reviewed Apr 20, 2022
View reviewed changes
Copy link
Contributor

@blefaudeux blefaudeux left a comment •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good to me, if you can run black to fix the lint/format ? Else we can merge and I run that on the other branch ?

tests/test_triton_blocksparse.py
def test_attention_fwd_bwd(
block,
input_scale=1.0,
scale=1 / 8.0,
n_ctx=256,
n_ctx=384,
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Not too bad for the unit test time of execution? We're trying to keep these small on purpose

tests/test_triton_blocksparse.py
query, key, value = [x.clone() for x in qkvs]
query.retain_grad()
key.retain_grad()
value.retain_grad()
if block not in [16, 32, 64]:
if block not in [16, 32, 64, 128]:
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

👍

xformers/components/attention/blocksparse.py
self.block_size,
device=q.device,
)
self.create_triton_kernels(q.device)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

👍

xformers/components/attention/blocksparse.py
q.shape[-2], 2
).is_integer(), (
"For now blocksparse only works on power-of-two sequence lengths"
assert (
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Nice catch!

@blefaudeux
Copy link
Contributor

Thanks a bunch @colehawkins !

@blefaudeux
Copy link
Contributor

Looks good to me, if you can run black to fix the lint/format ? Else we can merge and I run that on the other branch ?

Actually I can merge and format, don't bother

blefaudeux
blefaudeux approved these changes Apr 20, 2022
View reviewed changes
Copy link
Contributor

@blefaudeux blefaudeux left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks again @colehawkins ! merging

@blefaudeux blefaudeux merged commit b212063 into facebookresearch:triton-2 Apr 20, 2022
blefaudeux pushed a commit that referenced this pull request Apr 20, 2022
@colehawkins @amazon-colehawk @blefaudeux
Minor blocksparse refactoring, update block size restrictions, relax …
72fb5c7 
…power of two constraint (#277)

* Relax device size restrictions

* Refactor device creation and run all tests

* linting

Co-authored-by: Cole Hawkins <colehawk@amazon.com>
blefaudeux pushed a commit that referenced this pull request Apr 21, 2022
@colehawkins @amazon-colehawk @blefaudeux
Minor blocksparse refactoring, update block size restrictions, relax …
7ec9726 
…power of two constraint (#277)

* Relax device size restrictions

* Refactor device creation and run all tests

* linting

Co-authored-by: Cole Hawkins <colehawk@amazon.com>
blefaudeux added a commit that referenced this pull request Apr 21, 2022
@blefaudeux @kashif
@colehawkins @amazon-colehawk
[backend] 3/3 Triton 2 update (#272)
4ecbec1 
* parent be72b26
author Kashif Rasul <kashif.rasul@gmail.com> 1648069860 +0100
committer Benjamin Lefaudeux <benjamin.lefaudeux@pm.me> 1650256563 -0700

Move to Triton 2

Author:    Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: Benjamin Lefaudeux <benjamin.lefaudeux@pm.me>

Tentatively fixing layernorm

- faster all around
- bugfix

better take on sparse tensors, put layout on the correct device
update the pip packages, minor cleanup

* catering for triton blocksparse being probably more reliable in fp16

* faster layernorm

* Minor blocksparse refactoring, update block size restrictions, relax power of two constraint (#277)

* Relax device size restrictions

* Refactor device creation and run all tests

* linting

Co-authored-by: Cole Hawkins <colehawk@amazon.com>

* code review, thanks @fmassa !

Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: colepshawkins <31542048+colehawkins@users.noreply.github.com>
Co-authored-by: Cole Hawkins <colehawk@amazon.com>
blefaudeux added a commit that referenced this pull request Apr 21, 2022
@blefaudeux @kashif
@colehawkins @amazon-colehawk
[RFC] 1/3 Moving the CI to conda, picking a more modern cuda + pytorc…
498e009 
…h combo (#271)

* testing using conda to get the pytorch nightlies and matching cuda

* [fix] Making it explicit whether the attention mechanism supports an attention mask or not (#266)

check the assert

* [backend] 3/3 Triton 2 update (#272)

* parent be72b26
author Kashif Rasul <kashif.rasul@gmail.com> 1648069860 +0100
committer Benjamin Lefaudeux <benjamin.lefaudeux@pm.me> 1650256563 -0700

Move to Triton 2

Author:    Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: Benjamin Lefaudeux <benjamin.lefaudeux@pm.me>

Tentatively fixing layernorm

- faster all around
- bugfix

better take on sparse tensors, put layout on the correct device
update the pip packages, minor cleanup

* catering for triton blocksparse being probably more reliable in fp16

* faster layernorm

* Minor blocksparse refactoring, update block size restrictions, relax power of two constraint (#277)

* Relax device size restrictions

* Refactor device creation and run all tests

* linting

Co-authored-by: Cole Hawkins <colehawk@amazon.com>

* code review, thanks @fmassa !

Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: colepshawkins <31542048+colehawkins@users.noreply.github.com>
Co-authored-by: Cole Hawkins <colehawk@amazon.com>

Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: colepshawkins <31542048+colehawkins@users.noreply.github.com>
Co-authored-by: Cole Hawkins <colehawk@amazon.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@blefaudeux blefaudeux blefaudeux approved these changes

Assignees
No one assigned
Labels
CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@colehawkins @blefaudeux @facebook-github-bot @amazon-colehawk