Repository navigation
Add cweb frameworkfeat(frameworks): add cweb (C / io_uring) - #1682
Merged
Merged
Conversation
saad1737
requested review from
Kaliumhexacyanoferrat and
MDA2AV
as code owners
October 30, 2026 13:51
Kaliumhexacyanoferrat
force-pushed
the
add-cweb-framework
branch
from
October 30, 2026 13:57
519c742 to
ac5892e
Compare
MDA2AV
force-pushed
the
add-cweb-framework
branch
from
November 1, 2026 00:24
ac5892e to
51ee47e
Compare
Owner
|
/benchmark --save |
Contributor
|
👋 Benchmark request received. A collaborator will review and approve the run. |
Contributor
Benchmark ResultsFramework:
Composite score, main against this run, in the board's default view:
The results on main were measured in another session, so a few points up or down can be the machine and not the change. Full log |
Contributor
Author
|
Hi @MDA2AV — thank you for running the benchmark! Looking at the latency-500k-8cpu log (rate_ratio = 0.9015, p99 = 6.48s), I realized cweb was using sysconf(_SC_NPROCESSORS_ONLN) instead of sched_getaffinity(), which caused it to fork 128 worker processes onto the 8-core cpuset (and 128 workers onto the 64-core cpuset). I just pushed an update that: Uses sched_getaffinity() (CPU_COUNT) so cweb spawns the exact number of workers allowed by the container's --cpuset-cpus mask. Reduces MAX_BUFFER_SIZE to 16 KB to cut container RSS memory. Batches pipelined responses in READ / WS_READ and adds the json-comp profile. Could we please trigger one more benchmark run before merging? Thank you! |
MDA2AV
force-pushed
the
add-cweb-framework
branch
from
November 1, 2026 22:26
e51385f to
df76c34
Compare
Owner
|
/benchmark --save |
Contributor
|
👋 Benchmark request received. A collaborator will review and approve the run. |
Contributor
Benchmark ResultsFramework:
Composite score, main against this run, in the board's default view:
The results on main were measured in another session, so a few points up or down can be the machine and not the change. Full log |
Owner
|
Please set the "type" to experimental |
Contributor
Author
|
The type in meta.json has been updated to experimental |
…pelined responses, add json-comp
MDA2AV
force-pushed
the
add-cweb-framework
branch
from
November 3, 2026 15:04
c794e72 to
770e779
Compare
Owner
|
/benchmark --save |
Contributor
|
👋 Benchmark request received. A collaborator will review and approve the run. |
Contributor
Benchmark ResultsFramework:
Composite score, main against this run, in the board's default view:
The results on main were measured in another session, so a few points up or down can be the machine and not the change. Full log |
MDA2AV
approved these changes
Nov 3, 2026
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
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
Adds cweb (
type: emerging, Language:C), a zero-copy asynchronous HTTP/1.1 and WebSocket framework built from scratch in C99 on top of Linuxio_uring, pre-forkedSO_REUSEPORTworkers, and non-blockinglibpq.SO_REUSEPORTworker model with independent per-coreio_uringrings (IORING_SETUP_SINGLE_ISSUER | IORING_SETUP_COOP_TASKRUN).calloccontext pool).Transfer-Encoding: chunkeddecoding, RFC 6455 WebSocket frame coalescing, and non-blocking PostgreSQL (libpq+io_uringPOLLIN) state machine.Subscribed Profiles (10)
baseline,pipelined,limited-connzrk):latency-10k,latency-500k-8cpu,latency-1mecho-ws,echo-ws-pipeline,echo-ws-limitedasync-db(PostgreSQLlibpqasync state machine)Local Validation & Benchmark Summary (20-thread host)
./scripts/validate.sh cweb: 33 passed, 0 failed (including all 1,047 TCP fragmentation split offsets, chunked encoding, anti-cheat randomized checks,async-dbschema/empty-range checks, andvalidate-ws.py).benchmark.shhighlights (20 logical cores):baseline(4096c): 1.59M req/spipelined(4096c): 7.01M req/slimited-conn(4096c): 1.11M req/s (553k reconnects / 5s, 0 errors)async-db(1024c): 40.5K req/secho-ws(512c / 4096c / 16384c): 1.98M / 1.86M / 1.26M frames/secho-ws-pipeline(512c / 4096c / 16384c): 10.07M / 9.50M / 7.11M frames/secho-ws-limited(512c / 4096c): 1.12M / 1.13M frames/slatency-10k: p50 27µs | p99 141µs (cpu/req: 16.62µs,rate_ratio: 0.9987)latency-500k-8cpu: p50 63µs | p99 298µs (cpu/req: 6.00µs,rate_ratio: 0.9981)latency-1m: p50 365µs | p99 6.35ms (cpu/req: 6.05µs,rate_ratio: 0.9980)