stream: use the ring buffer for WHATWG stream request queues - #64431
Merged
nodejs-github-bot merged 1 commit intoJul 13, 2026
Merged
stream: use the ring buffer for WHATWG stream request queues#64431nodejs-github-bot merged 1 commit into
nodejs-github-bot merged 1 commit into
Conversation
Copy link
Copy Markdown
Member
Sorry, something went wrong.
mcollina
force-pushed
the
webstream-perf-round8
branch
from
e8dd939 to
e203888
Compare
July 11, 2026 20:48
Copy link
Copy Markdown
Member
|
Need a signature for first commit message. |
Sorry, something went wrong.
Copy link
Copy Markdown
Codecov Report✅ All modified and coverable lines are covered by tests. @@ Coverage Diff @@
## main #64431 +/- ##
==========================================
- Coverage 90.24% 90.23% -0.01%
==========================================
Files 741 741
Lines 241384 241391 +7
Branches 45480 45487 +7
==========================================
- Hits 217844 217827 -17
- Misses 15097 15129 +32
+ Partials 8443 8435 -8
... and 40 files with indirect coverage changes 🚀 New features to boost your workflow:
|
Sorry, something went wrong.
The [[queue]] backing the controllers became a ring buffer, but the read and write request queues (readRequests, readIntoRequests, writeRequests) were left as plain arrays consumed with ArrayPrototypeShift, which is O(n) and, even for the single pending request of the await-each regime, far slower than an indexed head advance. Back them with the same Queue, materialized lazily from the shared empty queue so acquiring a reader or constructing a writer allocates no request storage until a read or write actually parks. pipe-to: +4.7% to +9.4% (all 16 configs, ***) readable-read type=byob: +2.3% (**) parked read loop: +15%, write loop: +11% (local harness) Follow-up to nodejs#64312. Signed-off-by: Matteo Collina <hello@matteocollina.com>
mcollina
force-pushed
the
webstream-perf-round8
branch
from
e203888 to
e65f9be
Compare
July 12, 2026 10:48
jasnell
approved these changes
Jul 12, 2026
Copy link
Copy Markdown
Collaborator
Sorry, something went wrong.
Copy link
Copy Markdown
Contributor
|
Benchmark GHA (webstreams): https://github.com/nodejs/node/actions/runs/29233460863 Results
Benchmark results: confidence improvement accuracy (*) (**) (***) webstreams/creation.js kind='ReadableStream.tee' n=50000 0.64 % ±6.98% ±9.19% ±11.80% webstreams/creation.js kind='ReadableStream' n=50000 0.75 % ±7.98% ±10.51% ±13.49% webstreams/creation.js kind='ReadableStreamBYOBReader' n=50000 3.30 % ±6.13% ±8.08% ±10.36% webstreams/creation.js kind='ReadableStreamDefaultReader' n=50000 1.63 % ±7.61% ±10.03% ±12.86% webstreams/creation.js kind='TransformStream' n=50000 -0.51 % ±7.00% ±9.23% ±11.84% webstreams/creation.js kind='WritableStream' n=50000 -0.63 % ±7.84% ±10.33% ±13.26% webstreams/js_transfer.js n=10000 payload='ReadableStream' -0.43 % ±9.64% ±12.71% ±16.30% webstreams/js_transfer.js n=10000 payload='TransformStream' 0.77 % ±9.28% ±12.23% ±15.70% webstreams/js_transfer.js n=10000 payload='WritableStream' 0.83 % ±9.45% ±12.46% ±15.98% webstreams/pipe-to.js highWaterMarkW=1024 highWaterMarkR=1024 n=500000 * 7.38 % ±7.34% ±9.68% ±12.42% webstreams/pipe-to.js highWaterMarkW=1024 highWaterMarkR=2048 n=500000 * 8.21 % ±7.21% ±9.51% ±12.20% webstreams/pipe-to.js highWaterMarkW=1024 highWaterMarkR=4096 n=500000 * 7.71 % ±7.29% ±9.61% ±12.33% webstreams/pipe-to.js highWaterMarkW=1024 highWaterMarkR=512 n=500000 7.03 % ±7.25% ±9.56% ±12.27% webstreams/pipe-to.js highWaterMarkW=2048 highWaterMarkR=1024 n=500000 * 7.39 % ±7.26% ±9.57% ±12.28% webstreams/pipe-to.js highWaterMarkW=2048 highWaterMarkR=2048 n=500000 * 8.57 % ±7.26% ±9.57% ±12.28% webstreams/pipe-to.js highWaterMarkW=2048 highWaterMarkR=4096 n=500000 * 7.49 % ±7.15% ±9.43% ±12.10% webstreams/pipe-to.js highWaterMarkW=2048 highWaterMarkR=512 n=500000 * 7.74 % ±7.29% ±9.61% ±12.33% webstreams/pipe-to.js highWaterMarkW=4096 highWaterMarkR=1024 n=500000 * 7.58 % ±7.30% ±9.62% ±12.35% webstreams/pipe-to.js highWaterMarkW=4096 highWaterMarkR=2048 n=500000 * 9.04 % ±7.43% ±9.80% ±12.57% webstreams/pipe-to.js highWaterMarkW=4096 highWaterMarkR=4096 n=500000 * 9.27 % ±7.45% ±9.82% ±12.61% webstreams/pipe-to.js highWaterMarkW=4096 highWaterMarkR=512 n=500000 * 8.51 % ±7.42% ±9.78% ±12.55% webstreams/pipe-to.js highWaterMarkW=512 highWaterMarkR=1024 n=500000 * 7.52 % ±7.21% ±9.51% ±12.20% webstreams/pipe-to.js highWaterMarkW=512 highWaterMarkR=2048 n=500000 * 8.04 % ±7.35% ±9.68% ±12.42% webstreams/pipe-to.js highWaterMarkW=512 highWaterMarkR=4096 n=500000 * 8.19 % ±7.32% ±9.64% ±12.38% webstreams/pipe-to.js highWaterMarkW=512 highWaterMarkR=512 n=500000 6.34 % ±7.18% ±9.46% ±12.14% webstreams/readable-async-iterator.js type='bytes' n=100000 4.92 % ±9.89% ±13.04% ±16.73% webstreams/readable-async-iterator.js type='normal' n=100000 0.64 % ±7.19% ±9.47% ±12.15% webstreams/readable-read-buffered.js bufferSize=1 n=100000 -0.84 % ±6.44% ±8.48% ±10.89% webstreams/readable-read-buffered.js bufferSize=10 n=100000 -0.49 % ±5.97% ±7.87% ±10.10% webstreams/readable-read-buffered.js bufferSize=100 n=100000 -1.56 % ±6.02% ±7.94% ±10.18% webstreams/readable-read-buffered.js bufferSize=1000 n=100000 -0.08 % ±6.24% ±8.23% ±10.56% webstreams/readable-read.js type='byob' n=100000 0.48 % ±7.52% ±9.91% ±12.72% webstreams/readable-read.js type='normal' n=100000 ** 7.82 % ±5.83% ±7.69% ±9.87% Be aware that when doing many comparisons the risk of a false-positive result increases. In this case, there are 33 comparisons, you can thus expect the following amount of false-positive results: 1.65 false positives, when considering a 5% risk acceptance (*, **, ***), 0.33 false positives, when considering a 1% risk acceptance (**, ***), 0.03 false positives, when considering a 0.1% risk acceptance (***)
|
Sorry, something went wrong.
aduh95
approved these changes
Jul 13, 2026
aduh95
pushed a commit
that referenced
this pull request
Jul 21, 2026
The [[queue]] backing the controllers became a ring buffer, but the read and write request queues (readRequests, readIntoRequests, writeRequests) were left as plain arrays consumed with ArrayPrototypeShift, which is O(n) and, even for the single pending request of the await-each regime, far slower than an indexed head advance. Back them with the same Queue, materialized lazily from the shared empty queue so acquiring a reader or constructing a writer allocates no request storage until a read or write actually parks. pipe-to: +4.7% to +9.4% (all 16 configs, ***) readable-read type=byob: +2.3% (**) parked read loop: +15%, write loop: +11% (local harness) Follow-up to #64312. Signed-off-by: Matteo Collina <hello@matteocollina.com> PR-URL: #64431 Reviewed-By: James M Snell <jasnell@gmail.com> Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com>
aduh95
pushed a commit
that referenced
this pull request
Jul 21, 2026
The [[queue]] backing the controllers became a ring buffer, but the read and write request queues (readRequests, readIntoRequests, writeRequests) were left as plain arrays consumed with ArrayPrototypeShift, which is O(n) and, even for the single pending request of the await-each regime, far slower than an indexed head advance. Back them with the same Queue, materialized lazily from the shared empty queue so acquiring a reader or constructing a writer allocates no request storage until a read or write actually parks. pipe-to: +4.7% to +9.4% (all 16 configs, ***) readable-read type=byob: +2.3% (**) parked read loop: +15%, write loop: +11% (local harness) Follow-up to #64312. Signed-off-by: Matteo Collina <hello@matteocollina.com> PR-URL: #64431 Reviewed-By: James M Snell <jasnell@gmail.com> Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com>
RafaelGSS
pushed a commit
that referenced
this pull request
Jul 29, 2026
The [[queue]] backing the controllers became a ring buffer, but the read and write request queues (readRequests, readIntoRequests, writeRequests) were left as plain arrays consumed with ArrayPrototypeShift, which is O(n) and, even for the single pending request of the await-each regime, far slower than an indexed head advance. Back them with the same Queue, materialized lazily from the shared empty queue so acquiring a reader or constructing a writer allocates no request storage until a read or write actually parks. pipe-to: +4.7% to +9.4% (all 16 configs, ***) readable-read type=byob: +2.3% (**) parked read loop: +15%, write loop: +11% (local harness) Follow-up to #64312. Signed-off-by: Matteo Collina <hello@matteocollina.com> PR-URL: #64431 Reviewed-By: James M Snell <jasnell@gmail.com> Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com>
aduh95
pushed a commit
that referenced
this pull request
Aug 6, 2026
The [[queue]] backing the controllers became a ring buffer, but the read and write request queues (readRequests, readIntoRequests, writeRequests) were left as plain arrays consumed with ArrayPrototypeShift, which is O(n) and, even for the single pending request of the await-each regime, far slower than an indexed head advance. Back them with the same Queue, materialized lazily from the shared empty queue so acquiring a reader or constructing a writer allocates no request storage until a read or write actually parks. pipe-to: +4.7% to +9.4% (all 16 configs, ***) readable-read type=byob: +2.3% (**) parked read loop: +15%, write loop: +11% (local harness) Follow-up to #64312. Signed-off-by: Matteo Collina <hello@matteocollina.com> PR-URL: #64431 Reviewed-By: James M Snell <jasnell@gmail.com> Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com>
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.
The [[queue]] backing the controllers is now a ring buffer, but the read/write request queues (readRequests, readIntoRequests, writeRequests) were left as plain arrays consumed with ArrayPrototypeShift. Back them with the same Queue, materialized lazily from the shared empty queue so acquiring a reader or constructing a writer allocates no request storage until a read/write parks.
Follow-up to #64312.