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.

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

[CUDA] Report that devices with cc >= sm_70 support seq_cst #1291

Merged
merged 1 commit into from
Mar 18, 2024

Conversation

JackAKirk
Copy link
Contributor

Report that devices with cc >= sm_70 support seq_cst memory consistency capabilities. Tested with intel/llvm#12516

Copy link
Contributor

@GeorgeWeb GeorgeWeb left a comment

Choose a reason for hiding this comment

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

LGTM! 🥳

Btw, small aside, I think at some point it may be good we cache stuff like CU_DEVICE_ATTRIBUTE_COMPUTE_CAPABILITY_MAJOR in the device on init, as we seem to use it for a bunch of device cap queries, so while not much it may be useful to remove the overhead (or rather move it to a one-time on init).

@JackAKirk JackAKirk added the ready to merge Added to PR's which are ready to merge label Feb 2, 2024
@JackAKirk JackAKirk removed the ready to merge Added to PR's which are ready to merge label Feb 14, 2024
@JackAKirk
Copy link
Contributor Author

JackAKirk commented Feb 14, 2024

I've also now marked fence seq_cst supported by sm70. This is already the situation in the latest intel/llvm, see intel/llvm#12713 for a description.
I've unmarked this as ready to merge just to get an updated review for this small update. I think it makes sense to do both changes in the one PR.

@JackAKirk JackAKirk requested a review from GeorgeWeb February 14, 2024 15:31
@JackAKirk JackAKirk added the ready to merge Added to PR's which are ready to merge label Feb 15, 2024
@kbenzie kbenzie added the v0.9.x Include in the v0.9.x release label Mar 11, 2024
@kbenzie
Copy link
Contributor

kbenzie commented Mar 11, 2024

Please pull in the main branch to have up to date testing, also update the tag in the intel/llvm PR.

cuda ptx 6.0 memory consistency model is implemented in dpc++
for devices > sm_70.

Signed-off-by: JackAKirk <jack.kirk@codeplay.com>
@codecov-commenter
Copy link

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 12.49%. Comparing base (9b97a5f) to head (4f5d005).
Report is 471 commits behind head on main.

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files
@@            Coverage Diff             @@
##             main    #1291      +/-   ##
==========================================
- Coverage   15.46%   12.49%   -2.98%     
==========================================
  Files         238      239       +1     
  Lines       33883    36003    +2120     
  Branches     3747     4086     +339     
==========================================
- Hits         5240     4498     -742     
- Misses      28593    31501    +2908     
+ Partials       50        4      -46     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

@kbenzie kbenzie merged commit 29ee45c into oneapi-src:main Mar 18, 2024
50 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
ready to merge Added to PR's which are ready to merge v0.9.x Include in the v0.9.x release
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants