[DO NOT MERGE] docs/eval: 10k simnet reports (single-topic + 5-topic)#90
Draft
srene wants to merge 9 commits into
Draft
[DO NOT MERGE] docs/eval: 10k simnet reports (single-topic + 5-topic)#90srene wants to merge 9 commits into
srene wants to merge 9 commits into
Conversation
srene
marked this pull request as draft
July 14, 2026 07:35
srene
changed the title
srene
changed the title
…fig runs) with real overhead
…; annotate missing regs/ads received
…rored); overhead figures as per-time rates
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Two eval reports characterizing per-node message/byte overhead as a function of ID-space distance to the topic, at 10k nodes with the reg-3/search-20 config.
docs/eval/report-overhead-single-topic-10k.mddocs/eval/report-overhead-5-topics-10k.mddocs/eval/overhead-10k-1top-vs-5top.pngMeasured on the integration build (
topdisc+ #71 #81 #83 #84 + simnet hooks). Key finding: single-topic concentrates load on a few near-topic registrars (~45× byte skew); 5 topics spread it (~9× skew, higher baseline).