Skip to content

[upstream] added per tier pool class rolling average latency - #45

Closed
guptask wants to merge 1 commit into
intel:mainfrom
guptask:upstream_rolling_stats
Closed

[upstream] added per tier pool class rolling average latency#45
guptask wants to merge 1 commit into
intel:mainfrom
guptask:upstream_rolling_stats

Conversation

@guptask

@guptaskguptask commented Dec 20, 2022

Copy link
Copy Markdown

This change is Reviewable

@guptaskguptask self-assigned this Dec 20, 2022
@guptask
guptaskforce-pushed the upstream_rolling_stats branch from 38b17a8 to caa1318CompareDecember 20, 2022 22:55

@igchorigchor left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

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

Reviewable status: 0 of 7 files reviewed, 3 unresolved discussions (waiting on @guptask)


cachelib/allocator/Cache.h line 89 at r1 (raw file):

// TODO: come up with some reasonable numberstaticconstexprunsignedkMaxTiers = 2;

Can we not include any references to tiers in this PR? We could add support for tracking latency per class, per pool, right?


cachelib/allocator/CacheAllocator-inl.h line 2231 at r1 (raw file):

template <typename CacheTrait>
double CacheAllocator<CacheTrait>::slabsApproxFreePercentage(TierId tid) const {

I think this should not part of the latency tracking patch.


cachelib/allocator/CacheAllocator-inl.h line 2236 at r1 (raw file):

template <typename CacheTrait>
AllocationClassBaseStat CacheAllocator<CacheTrait>::getAllocationClassStats(

Same here, you should probably just store allocLatenencyNs.

You can rebase onto #42 perhaps and just use getACStats which I've added if you think it will be easier.

@guptask

Copy link
Copy Markdown
Author

cachelib/allocator/CacheAllocator-inl.h line 2231 at r1 (raw file):

Previously, igchor (Igor Chorążewicz) wrote…

I think this should not part of the latency tracking patch.

Done.

@guptask

Copy link
Copy Markdown
Author

cachelib/allocator/CacheAllocator-inl.h line 2236 at r1 (raw file):

Previously, igchor (Igor Chorążewicz) wrote…

Same here, you should probably just store allocLatenencyNs.

You can rebase onto #42 perhaps and just use getACStats which I've added if you think it will be easier.

Done.

@guptask

Copy link
Copy Markdown
Author

cachelib/allocator/Cache.h line 89 at r1 (raw file):

Previously, igchor (Igor Chorążewicz) wrote…

Can we not include any references to tiers in this PR? We could add support for tracking latency per class, per pool, right?

Done.

@guptask

guptask commented Dec 28, 2022

Copy link
Copy Markdown
Author

I'm closing this PR and raising a new one (#47) that is rebased against Igor's AC Stats upstream PR (#42)

@guptaskguptask closed this Dec 28, 2022
Sign up for freeto join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants

@guptask@igchor