src: add fast path to TextEncoder.encodeInto

[anonrig]

Original commit message:

[fastcall] Implement support for onebyte string arguments

Refs: v8/v8@bc831f8

Benchmark CI: https://ci.nodejs.org/view/Node.js%20benchmark/job/benchmark-node-micro-benchmarks/1259/

 confidence improvement accuracy (*) (**) (***)
util/text-encoder.js op='encodeInto' type='v8-one-byte-string' n=10000 len=0 1.99 % ±4.94% ±6.58% ±8.57%
util/text-encoder.js op='encodeInto' type='v8-one-byte-string' n=10000 len=1024 *** 13.53 % ±6.84% ±9.10% ±11.86%
util/text-encoder.js op='encodeInto' type='v8-one-byte-string' n=10000 len=256 5.36 % ±6.97% ±9.29% ±12.12%
util/text-encoder.js op='encodeInto' type='v8-one-byte-string' n=10000 len=32768 *** 15.07 % ±7.99% ±10.63% ±13.84%
util/text-encoder.js op='encodeInto' type='v8-two-byte-string' n=10000 len=0 -5.38 % ±6.23% ±8.31% ±10.85%
util/text-encoder.js op='encodeInto' type='v8-two-byte-string' n=10000 len=1024 1.45 % ±6.77% ±9.01% ±11.74%
util/text-encoder.js op='encodeInto' type='v8-two-byte-string' n=10000 len=256 -2.36 % ±7.55% ±10.05% ±13.08%
util/text-encoder.js op='encodeInto' type='v8-two-byte-string' n=10000 len=32768 -0.49 % ±2.34% ±3.11% ±4.06%
util/text-encoder.js op='encode' type='v8-one-byte-string' n=10000 len=0 1.62 % ±6.95% ±9.25% ±12.07%
util/text-encoder.js op='encode' type='v8-one-byte-string' n=10000 len=1024 -2.32 % ±5.28% ±7.03% ±9.17%
util/text-encoder.js op='encode' type='v8-one-byte-string' n=10000 len=256 -4.58 % ±6.07% ±8.08% ±10.54%
util/text-encoder.js op='encode' type='v8-one-byte-string' n=10000 len=32768 4.25 % ±11.64% ±15.49% ±20.16%
util/text-encoder.js op='encode' type='v8-two-byte-string' n=10000 len=0 -6.16 % ±6.84% ±9.11% ±11.86%
util/text-encoder.js op='encode' type='v8-two-byte-string' n=10000 len=1024 -6.52 % ±7.40% ±9.85% ±12.83%
util/text-encoder.js op='encode' type='v8-two-byte-string' n=10000 len=256 -6.35 % ±6.74% ±8.97% ±11.68%
util/text-encoder.js op='encode' type='v8-two-byte-string' n=10000 len=32768 -5.08 % ±7.21% ±9.60% ±12.50%
Be aware that when doing many comparisons the risk of a false-positive
result increases. In this case, there are 16 comparisons, you can thus
expect the following amount of false-positive results:
0.80 false positives, when considering a 5% risk acceptance (*, **, ***),
0.16 false positives, when considering a 1% risk acceptance (**, ***),
0.02 false positives, when considering a 0.1% risk acceptance (***)

[nodejs-github-bot]

Review requested:

• @nodejs/gyp
• @nodejs/v8-update

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/48264/

[nodejs-github-bot]

V8 CI: https://ci.nodejs.org/job/node-test-commit-v8-linux/nodes=benchmark-ubuntu1804-intel-64,v8test=v8test/5027/

[nodejs-github-bot]

V8 CI: https://ci.nodejs.org/job/node-test-commit-v8-linux/nodes=rhel8-s390x,v8test=v8test/5027/

[nodejs-github-bot]

V8 CI: https://ci.nodejs.org/job/node-test-commit-v8-linux/nodes=rhel8-ppc64le,v8test=v8test/5027/

[targos]

I would be more inclined to accept this if the new capabilities were used by a subsequent commit. Otherwise the change doesn't bring anything to Node.js.

[anonrig]

I would be more inclined to accept this if the new capabilities were used by a subsequent commit. Otherwise the change doesn't bring anything to Node.js.

@targos Should I include my changes depending on this to this particular pull request, and add commit-queue-rebase?

[targos]

That would be fine, yes.

[ShogunPanda]

These changes look fine to me. I'm approving, but I agree on what @targos said.

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/48278/

resolved

[tniessen]

Nit: we already have so many files dealing with encodings, does this really need to be a new one (instead of, e.g., staying in node_buffer or maybe string_bytes)? And if so, is this the encoding file?

[anonrig]

I don’t have any specific thoughts on this. I’ll try to move the encode utf8 to the node_encoding once this is implementation improves the existing benchmark.

[tniessen]

I'm still not sure why this is a new file when it only covers a tiny part of all encoding-related routines.

[nodejs-github-bot]

V8 CI: https://ci.nodejs.org/job/node-test-commit-v8-linux/nodes=rhel8-s390x,v8test=v8test/5040/

[nodejs-github-bot]

V8 CI: https://ci.nodejs.org/job/node-test-commit-v8-linux/nodes=rhel8-ppc64le,v8test=v8test/5040/

[anonrig]

@tniessen Can you review this, since in your last review, you requested changes?

[anonrig]

@targos If it's ok, I recommend creating a pull request for merging the v8 cherry-pick, and continue this performance experiment in this pull request. There are several new areas (wasm for example) that can leverage this cherry-pick.

[tniessen]

There is no guarantee that V8 would take the fast path.

Suggested change

	// For loop is required to trigger the fast path for encodeInto
	// Since v8 fast path is only triggered when v8 optimization starts.
	// Using a loop increases the chances of triggering the fast path for encodeInto
	// because V8 heuristically optimizes based on information gathered at runtime.

[tniessen]

I'm still not sure why this is a new file when it only covers a tiny part of all encoding-related routines.

[aduh95]

There are git conflicts, no Jenkins CI run on the last commit, and two collaborators blocking the PR, removing author readyPRs that have at least one approval, no pending requests for changes, and a CI started. .

[anonrig]

Thanks, @aduh95. Since other pull requests are merged, I’ll update this pull request only focusing on encodeInto.