Update docs for v2: rewrite README, add ARCHITECTURE_v2, archive v1#750
Open
excelle08 wants to merge 1 commit into
Open
Update docs for v2: rewrite README, add ARCHITECTURE_v2, archive v1#750excelle08 wants to merge 1 commit into
excelle08 wants to merge 1 commit into
Conversation
Summary: Refresh the FeedSim documentation to reflect the v2 recommended job (`feedsim_dlrm`) and 700 ms p95 SLA: - `README.md`: rewritten around `feedsim_dlrm` as the current recommended job — 700 ms SLA, one instance per host (no v1 autoscale-per-100-cores rule), a converged BGM result JSON with CPU model and hostname redacted, plus advanced usage sections for fixed-QPS sweeps, multi-instance mode (`num_instances` / `feedsim_autoscale_dlrm`), and both job-level and runner-level parameter tables. Legacy content moved out. - `README_v1.md`: legacy `feedsim_autoscale` (PageRank, 500 ms SLA) docs preserved verbatim with a "Migrating to v2" pointer. - `ARCHITECTURE_v2.md`: new public architecture doc covering the three processes (Driver, Leaf, mock_services), the four named thread pools inside `LeafNodeRank` (`ThriftSrv.IO`, `SREventBase`, `RANKER`, `GlobalCPUThread`), the request control flow with executor hops, the data flow (request/response CDFs + Silesia corpus + DLRM model), the wire protocol (Rocket over fizz+ALPN "rs", per-channel ZSTD), and a v1↔v2 comparison table. Explicitly notes that v2 today runs a single request-response per driver call and that multi-hop session flow is planned future work. - `docs/architecture_v2.svg`: component-focused SVG diagram (Driver | Leaf with 4 named pools + modules | mock_services), colored legend for request / fanout / response / hop arrows, shared-inputs strip at the bottom. No code changes. Differential Revision: D110821423
meta-cla
Bot
added
the
CLA Signed
|
@excelle08 has exported this pull request. If you are a Meta employee, you can view the originating Diff in D110821423. |
meta-codesync Bot
pushed a commit
that referenced
this pull request
Jul 7, 2026
…750) Summary: Pull Request resolved: #750 Refresh the FeedSim documentation to reflect the v2 recommended job (`feedsim_dlrm`) and 700 ms p95 SLA: - `README.md`: rewritten around `feedsim_dlrm` as the current recommended job — 700 ms SLA, one instance per host (no v1 autoscale-per-100-cores rule), a converged BGM result JSON with CPU model and hostname redacted, plus advanced usage sections for fixed-QPS sweeps, multi-instance mode (`num_instances` / `feedsim_autoscale_dlrm`), and both job-level and runner-level parameter tables. Legacy content moved out. - `README_v1.md`: legacy `feedsim_autoscale` (PageRank, 500 ms SLA) docs preserved verbatim with a "Migrating to v2" pointer. - `ARCHITECTURE_v2.md`: new public architecture doc covering the three processes (Driver, Leaf, mock_services), the four named thread pools inside `LeafNodeRank` (`ThriftSrv.IO`, `SREventBase`, `RANKER`, `GlobalCPUThread`), the request control flow with executor hops, the data flow (request/response CDFs + Silesia corpus + DLRM model), the wire protocol (Rocket over fizz+ALPN "rs", per-channel ZSTD), and a v1↔v2 comparison table. Explicitly notes that v2 today runs a single request-response per driver call and that multi-hop session flow is planned future work. - `docs/architecture_v2.svg`: component-focused SVG diagram (Driver | Leaf with 4 named pools + modules | mock_services), colored legend for request / fanout / response / hop arrows, shared-inputs strip at the bottom. No code changes. Reviewed By: charles-typ Differential Revision: D110821423 fbshipit-source-id: 4ec7515eca55274cf169df92b325a09de32fc1d2
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.
Summary:
Refresh the FeedSim documentation to reflect the v2 recommended job (
feedsim_dlrm) and 700 ms p95 SLA:README.md: rewritten aroundfeedsim_dlrmas the current recommended job — 700 ms SLA, one instance per host (no v1 autoscale-per-100-cores rule), a converged BGM result JSON with CPU model and hostname redacted, plus advanced usage sections for fixed-QPS sweeps, multi-instance mode (num_instances/feedsim_autoscale_dlrm), and both job-level and runner-level parameter tables. Legacy content moved out.README_v1.md: legacyfeedsim_autoscale(PageRank, 500 ms SLA) docs preserved verbatim with a "Migrating to v2" pointer.ARCHITECTURE_v2.md: new public architecture doc covering the three processes (Driver, Leaf, mock_services), the four named thread pools insideLeafNodeRank(ThriftSrv.IO,SREventBase,RANKER,GlobalCPUThread), the request control flow with executor hops, the data flow (request/response CDFs + Silesia corpus + DLRM model), the wire protocol (Rocket over fizz+ALPN "rs", per-channel ZSTD), and a v1↔v2 comparison table. Explicitly notes that v2 today runs a single request-response per driver call and that multi-hop session flow is planned future work.docs/architecture_v2.svg: component-focused SVG diagram (Driver | Leaf with 4 named pools + modules | mock_services), colored legend for request / fanout / response / hop arrows, shared-inputs strip at the bottom.No code changes.
Differential Revision: D110821423