[qosorch]: Optimizations of qosorch #488
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
…_SCHEDULER_GROUP just once
|
I'll update the PR with more changes |
pavel-shirshov
changed the title
lguohan
approved these changes
Apr 27, 2018
EdenGri
pushed a commit
to EdenGri/sonic-swss
that referenced
this pull request
Feb 28, 2022
oleksandrivantsiv
pushed a commit
to oleksandrivantsiv/sonic-swss
that referenced
this pull request
Mar 1, 2023
…ters() (sonic-net#488) Each time a new counter/attribute type is added to flex counter infrastructure, collectCounters() is tailed with the logic of collecting the stats for the new type. As more and more possible id lists are being added to the flex counter, the collectCounters() member function is becoming bulky to check against all possible id lists, both existing and newly added ones. Current list is already significantly long---port counters, queue counters, queue attributes, pg counters, pg attributes, and buffer pool counters. And most of them are irrelevant to a particular flex counter instance thread. E.g., For buffer pool watermark thread, only buffer pool counters matter, not the rest, port counters, queue counters, queue attributes, pg counters, pg attributes. Improvement: Separate each counter/attribute type to be a member function of FlexCounter with the same function prototype. Have a data member (unordered_map) to maintain pointers to member functions for each flex counter instance, and install only the essential counter collection member functions at the counter/attribute list set operation. Choice of data member structure: Originally considered unordered_set with element T to be a pointer to member function. However, the element type does not support a certain operations, and thus can not be taken as a key, but only as a value. In doing so, we can avoid checking against irrelevant counter/attribute id lists, and keep a fixed size for collectCounters function. Signed-off-by: Wenda Ni <[email protected]>
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.
What I did
Optimized qosorch to make it run faster. I'm caching intermediate results, to avoid re-request the information which was already requested from ASIC.
Why I did it
It reduces qosorch working time significantly. For example on the target platform it was reduced by 4 secs.
Here the statistics of SAI requests before and after the change:
Before:
After:
How I verified it
Build and check sairedis.rec. You should see only one get of this attribute from the switch.
Details if related