-
Notifications
You must be signed in to change notification settings - Fork 5.5k
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
[release] 2.8 release perf logs #40571
Conversation
w00h00 thank you, I'll leave this up for your team to review then. Also they normally try something like this #29615 to make it easier to review. |
|
Diff here: #40572 |
For |
|
Yeah, but given the release branch is just a prefix of the master branch, I belive it's more of flakiness. Could we rerun it? |
Will run again after cherry picks, should give more signal on those edge of distribution values |
The new ones from today. Still waiting on @GeneDer for 2.7.1 release logs to compare against (instead of 2.7.0)
|
Signed-off-by: vitsai <victoria@anyscale.com>
Signed-off-by: vitsai <vitsai@cs.stanford.edu>
This one is against 2.7.1
|
I believe there's some regression in The test is testing submitting of 1M tasks from the driver (which overloads the task backend buffer on the driver worker), given other more realistic tests like ![]() |
|
|
@rickyyx Still seeing 1000000_queued_time |
So on the same commit:
I think we should ignore this as variance. |
|
This seems regression since we had a run on 2.8.0 with non-regressed perf just before: https://buildkite.com/ray-project/release-tests-branch/builds/2365#018b843e-61ff-4d6d-880e-0a10b6b122fa |
And others metrics look like variance. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
119.28968447369434 | ||
], | ||
"1_n_actor_calls_async": [ | ||
9581.728569086026, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
regressed?
{ | ||
"perf_metric_name": "single_client_tasks_async", | ||
"perf_metric_type": "THROUGHPUT", | ||
"perf_metric_value": 8643.833466025399 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
this is pretty low
@@ -0,0 +1,13 @@ | |||
{ | |||
"broadcast_time": 82.940892212, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
this has a slight regression
{ | ||
"perf_metric_name": "stage_2_avg_iteration_time", | ||
"perf_metric_type": "LATENCY", | ||
"perf_metric_value": 68.77781887054444 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
regression
is this still relevant and a release-blocker for ray29? |
@vitsai should i merge this? |
Signed-off-by: vitsai <victoria@anyscale.com> Signed-off-by: vitsai <vitsai@cs.stanford.edu>
Why are these changes needed?
Related issue number
Checks
git commit -s
) in this PR.scripts/format.sh
to lint the changes in this PR.method in Tune, I've added it in
doc/source/tune/api/
under thecorresponding
.rst
file.