lib,src: improve writev() performance for Buffers #13187

mscdex · 2017-05-24T05:39:06Z

It's pretty common to write only Buffers to a socket, so we can optimize for that case in a couple of ways whenever we know we have all Buffers queued up for writev():

Reuse the existing array of chunks
Try to write the Buffers (or at least as much as possible) immediately without having to create a WriteWrap instance first. This optimization is one that is already being done when write()ing a single Buffer (or a single string for that matter) to a socket.

Benchmark results:

                                                improvement confidence      p.value
 net/net-c2s-cork.js dur=5 type="buf" len=1024     32.36 %        *** 3.042522e-23
 net/net-c2s-cork.js dur=5 type="buf" len=128      27.74 %        *** 5.759989e-59
 net/net-c2s-cork.js dur=5 type="buf" len=16       30.72 %        *** 1.594866e-57
 net/net-c2s-cork.js dur=5 type="buf" len=32       31.18 %        *** 8.568983e-62
 net/net-c2s-cork.js dur=5 type="buf" len=4        24.99 %        *** 2.220787e-64
 net/net-c2s-cork.js dur=5 type="buf" len=512      29.04 %        *** 3.058048e-32
 net/net-c2s-cork.js dur=5 type="buf" len=64       29.34 %        *** 5.167685e-50
 net/net-c2s-cork.js dur=5 type="buf" len=8        11.35 %        *** 4.258937e-43

CI: https://ci.nodejs.org/job/node-test-pull-request/8271/
CI for FIPS (since it failed initially): https://ci.nodejs.org/job/node-test-commit-linux-fips/8624/

Checklist

make -j4 test (UNIX), or vcbuild test (Windows) passes
commit message follows commit guidelines

Affected core subsystem(s)

net
stream

benjamingr · 2017-05-24T10:11:01Z

lib/_stream_writable.js

    while (entry) {
      buffer[count] = entry;
+      if (!entry.isBuf)


I'd probably do allBuffers = allBuffers && entry.isBuf

I tried something similar (continuing the rest of the loop without any conditional) and did not see any difference in performance.

addaleax · 2017-05-24T10:53:41Z

src/stream_base.cc

@@ -100,94 +100,118 @@ int StreamBase::Writev(const FunctionCallbackInfo<Value>& args) {

  Local<Object> req_wrap_obj = args[0].As<Object>();
  Local<Array> chunks = args[1].As<Array>();
+  bool all_buffers = args[2]->BooleanValue();


Can you use the overload that returns a Maybe<bool>, or alternatively use IsTrue() instead?

addaleax · 2017-05-24T10:54:36Z

src/stream_base.cc

+  if (all_buffers)
+    count = chunks->Length();
+  else
+    count = chunks->Length() >> 1;


nit: I’d really just write / 2, it’s clearer and the resulting code will be the same

That's copied from before these changes. It's also similar to what was already being used in js.

Yeah, but in JS that has a different meaning. ¯\_(ツ)_/¯ If you want to keep it it’s fine.

addaleax · 2017-05-24T10:58:32Z

src/stream_base.cc

+  wrap = GetAsyncWrap();
+  // NOTE: All tests show that GetAsyncWrap() never returns nullptr here. If it
+  // can then replace the CHECK_NE() with if (wrap != nullptr).
+  CHECK_NE(wrap, nullptr);


Heads up, this is going to conflict with #13174 which removes the comment (because GetAsyncWrap() does actually never return a null pointer), so feel free to remove it here as well

I'll just wait until that lands first and I'll rebase.

@mscdex Landed, you can rebase now

addaleax · 2017-05-24T10:59:33Z

src/stream_base.cc

-  size_t offset = 0;
-  for (size_t i = 0; i < count; i++) {
-    Local<Value> chunk = chunks->Get(i * 2);
+      if (Buffer::HasInstance(chunk))


nit: chunk->IsUint8Array()

This was also copied from before these changes. I think changing it to check for a more general type is outside the scope of this PR.

I think changing it to check for a more general type is outside the scope of this PR.

It’s literally the same thing, just inlined. :)

Actually HasInstance() in master and v8.x use IsArrayBufferView() while v7.x and v6.x use IsUint8Array(). Ideally this could be backported more easily if we kept HasInstance()?

Oh, right, we made that switch. Idc, you can also keep Buffer::HasInstance.

PR-URL: nodejs#13187 Reviewed-By: Benjamin Gruenbaum <benjamingr@gmail.com> Reviewed-By: Anna Henningsen <anna@addaleax.net>

PR-URL: #13187 Reviewed-By: Benjamin Gruenbaum <benjamingr@gmail.com> Reviewed-By: Anna Henningsen <anna@addaleax.net>

MylesBorins · 2017-07-17T16:54:11Z

Should we consider this for LTS? if so it needs to bake for a bit. Please change labels as appropriate

mscdex added c++ Issues and PRs that require attention from people who are familiar with C++. net Issues and PRs related to the net subsystem. performance Issues and PRs related to the performance of Node.js. stream Issues and PRs related to the stream subsystem. labels May 24, 2017

nodejs-github-bot added c++ Issues and PRs that require attention from people who are familiar with C++. net Issues and PRs related to the net subsystem. stream Issues and PRs related to the stream subsystem. labels May 24, 2017

benjamingr approved these changes May 24, 2017

View reviewed changes

addaleax approved these changes May 24, 2017

View reviewed changes

mscdex force-pushed the net-writev-perf branch 2 times, most recently from c5814a0 to fd84f30 Compare May 25, 2017 18:33

lib,src: improve writev() performance for Buffers

01a1022

PR-URL: nodejs#13187 Reviewed-By: Benjamin Gruenbaum <benjamingr@gmail.com> Reviewed-By: Anna Henningsen <anna@addaleax.net>

mscdex force-pushed the net-writev-perf branch from fd84f30 to 01a1022 Compare May 26, 2017 08:30

mscdex merged commit 01a1022 into nodejs:master May 26, 2017

mscdex deleted the net-writev-perf branch May 26, 2017 08:33

jasnell pushed a commit that referenced this pull request May 28, 2017

lib,src: improve writev() performance for Buffers

3becb02

PR-URL: #13187 Reviewed-By: Benjamin Gruenbaum <benjamingr@gmail.com> Reviewed-By: Anna Henningsen <anna@addaleax.net>

jasnell mentioned this pull request May 28, 2017

8.0.0 Release Proposal #12220

Closed

gibfahn mentioned this pull request Jun 15, 2017

Auditing for 6.11.1 nodejs/Release#230

Closed

3 tasks

MylesBorins added baking-for-lts PRs that need to wait before landing in a LTS release. lts-watch-v6.x labels Jul 17, 2017

MylesBorins removed the baking-for-lts PRs that need to wait before landing in a LTS release. label Aug 17, 2018

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

lib,src: improve writev() performance for Buffers #13187

lib,src: improve writev() performance for Buffers #13187

mscdex commented May 24, 2017 •

edited

Loading

benjamingr May 24, 2017

mscdex May 24, 2017

addaleax May 24, 2017

mscdex May 24, 2017

addaleax May 24, 2017

mscdex May 24, 2017

addaleax May 24, 2017

addaleax May 24, 2017

mscdex May 24, 2017

addaleax May 25, 2017

mscdex May 25, 2017

addaleax May 24, 2017

mscdex May 24, 2017

addaleax May 24, 2017

mscdex May 24, 2017 •

edited

Loading

addaleax May 24, 2017

MylesBorins commented Jul 17, 2017

lib,src: improve writev() performance for Buffers #13187

lib,src: improve writev() performance for Buffers #13187

Conversation

mscdex commented May 24, 2017 • edited Loading

Checklist

Affected core subsystem(s)

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mscdex May 24, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

MylesBorins commented Jul 17, 2017

mscdex commented May 24, 2017 •

edited

Loading

mscdex May 24, 2017 •

edited

Loading