FazBrowse GitHub Viewer | Trending |
URL:
| Home
Tools: [Download Repo ZIP]   [Original HTTPS Page]

src: add Latin1 fast path in StringBytes::Encode utf8 · nodejs/node@5b3bb28 · GitHub

/ node Public

Commit 5b3bb28

Browse files
mertcanaltin
authored and
aduh95
committed
src: add Latin1 fast path in StringBytes::Encode utf8
Signed-off-by: Mert Can Altin <mertgold60@gmail.com> PR-URL: #63385 Reviewed-By: Anna Henningsen <anna@addaleax.net> Reviewed-By: Matteo Collina <matteo.collina@gmail.com> Reviewed-By: Gürgün Dayıoğlu <hey@gurgun.day>
1 parent 7433c3d commit 5b3bb28

2 files changed

Lines changed: 132 additions & 14 deletions

File tree

‎benchmark/buffers/buffer-tostring-utf8-latin1.js‎

Lines changed: 53 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,53 @@
1+
'use strict';
2+
3+
const common = require('../common.js');
4+
5+
const bench = common.createBenchmark(main, {
6+
size: [64, 1024, 16384, 262144, 4194304],
7+
content: ['ascii', 'latin1', 'utf8_mixed', 'latin1_then_cjk'],
8+
n: [1e4],
9+
});
10+
11+
function buildBuffer(kind, size) {
12+
if (kind === 'ascii') {
13+
return Buffer.alloc(size, 0x61);
14+
}
15+
if (kind === 'latin1') {
16+
const pair = Buffer.from([0xC3, 0xA9]);
17+
const buf = Buffer.alloc(size);
18+
for (let i = 0; i + 2 <= size; i += 2) pair.copy(buf, i);
19+
return buf;
20+
}
21+
if (kind === 'utf8_mixed') {
22+
const cjk = Buffer.from([0xE4, 0xB8, 0xAD]);
23+
const buf = Buffer.alloc(size);
24+
let i = 0;
25+
while (i + 4 <= size) {
26+
buf[i++] = 0x61;
27+
cjk.copy(buf, i);
28+
i += 3;
29+
}
30+
return buf;
31+
}
32+
if (kind === 'latin1_then_cjk') {
33+
const pair = Buffer.from([0xC3, 0xA9]);
34+
const cjk = Buffer.from([0xE4, 0xB8, 0xAD]);
35+
const buf = Buffer.alloc(size);
36+
const mid = (size >> 1) & ~1;
37+
for (let i = 0; i + 2 <= mid; i += 2) pair.copy(buf, i);
38+
cjk.copy(buf, mid);
39+
for (let i = mid + 3; i + 2 <= size; i += 2) pair.copy(buf, i);
40+
return buf;
41+
}
42+
throw new Error('unknown content: ' + kind);
43+
}
44+
45+
function main({ n, size, content }) {
46+
const buf = buildBuffer(content, size);
47+
48+
bench.start();
49+
for (let i = 0; i < n; i++) {
50+
buf.toString('utf8');
51+
}
52+
bench.end(n);
53+
}

‎src/string_bytes.cc‎

Lines changed: 79 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -582,21 +582,86 @@ MaybeLocal<Value> StringBytes::Encode(Isolate* isolate,
582582
return ExternOneByteString::NewFromCopy(isolate, buf, buflen);
583583
}
584584

585-
if (buflen >= 32 && simdutf::validate_utf8(buf, buflen)) {
586-
// We know that we are non-ASCII (and are unlikely Latin1), use 2-byte
587-
// In the most likely case of valid UTF-8, we can use this fast impl
588-
// For very short input, it is slower, so we limit min size
589-
size_t u16size = simdutf::utf16_length_from_utf8(buf, buflen);
590-
if (u16size > static_cast<size_t>(v8::String::kMaxLength)) {
591-
isolate->ThrowException(ERR_STRING_TOO_LONG(isolate));
592-
return MaybeLocal<Value>();
585+
// Latin1-fits fast path: one-byte V8 string, half the heap of UTF-16.
586+
// Capped at 1 MiB (above that the prescan cost erases the win).
587+
constexpr size_t kLatin1Max = 1u << 20;
588+
if (buflen >= 256 && buflen <= kLatin1Max) {
589+
// Skip the allocation when any byte >= 0xC4 (UTF-8 lead for a
590+
// codepoint > U+FF). Inner loop has no early exit so clang
591+
// vectorizes it.
592+
constexpr size_t kChunk = 64;
593+
bool maybe_latin1 = true;
594+
size_t i = 0;
595+
for (; i + kChunk <= buflen; i += kChunk) {
596+
uint8_t acc = 0;
597+
for (size_t j = 0; j < kChunk; j++) {
598+
acc |= static_cast<uint8_t>(buf[i + j]) >= 0xC4 ? 1 : 0;
599+
}
600+
if (acc) {
601+
maybe_latin1 = false;
602+
break;
603+
}
604+
}
605+
if (maybe_latin1) {
606+
for (; i < buflen; i++) {
607+
if (static_cast<uint8_t>(buf[i]) >= 0xC4) {
608+
maybe_latin1 = false;
609+
break;
610+
}
611+
}
612+
}
613+
if (maybe_latin1) {
614+
MaybeStackBuffer<char, 4096> latin1;
615+
latin1.AllocateSufficientStorage(buflen);
616+
simdutf::result l1 = simdutf::convert_utf8_to_latin1_with_errors(
617+
buf, buflen, latin1.out());
618+
if (l1.error == simdutf::error_code::SUCCESS) {
619+
return ExternOneByteString::NewFromCopy(
620+
isolate, latin1.out(), l1.count);
621+
}
622+
}
623+
}
624+
625+
if (buflen >= 32) {
626+
// Single-pass UTF-16: over-allocate (1 char16_t per byte), then
627+
// shrink. Above 1 MiB the exact-size 3-pass below is cheaper.
628+
constexpr size_t kSinglePassMax = 1u << 20;
629+
if (buflen <= kSinglePassMax) {
630+
MaybeStackBuffer<uint16_t, 256> u16;
631+
u16.AllocateSufficientStorage(buflen);
632+
simdutf::result r = simdutf::convert_utf8_to_utf16_with_errors(
633+
buf, buflen, reinterpret_cast<char16_t*>(u16.out()));
634+
if (r.error == simdutf::error_code::SUCCESS) {
635+
if (r.count > static_cast<size_t>(v8::String::kMaxLength)) {
636+
isolate->ThrowException(ERR_STRING_TOO_LONG(isolate));
637+
return MaybeLocal<Value>();
638+
}
639+
if (u16.IsAllocated()) {
640+
uint16_t* data = u16.out();
641+
u16.Release();
642+
uint16_t* shrunk = static_cast<uint16_t*>(
643+
realloc(data, r.count * sizeof(uint16_t)));
644+
if (shrunk == nullptr) shrunk = data;
645+
return ExternTwoByteString::New(isolate, shrunk, r.count);
646+
}
647+
return String::NewFromTwoByte(isolate,
648+
u16.out(),
649+
v8::NewStringType::kNormal,
650+
static_cast<int>(r.count));
651+
}
652+
} else if (simdutf::validate_utf8(buf, buflen)) {
653+
size_t u16size = simdutf::utf16_length_from_utf8(buf, buflen);
654+
if (u16size > static_cast<size_t>(v8::String::kMaxLength)) {
655+
isolate->ThrowException(ERR_STRING_TOO_LONG(isolate));
656+
return MaybeLocal<Value>();
657+
}
658+
return EncodeTwoByteString(
659+
isolate, u16size, [buf, buflen, u16size](uint16_t* dst) {
660+
size_t written = simdutf::convert_valid_utf8_to_utf16(
661+
buf, buflen, reinterpret_cast<char16_t*>(dst));
662+
CHECK_EQ(written, u16size);
663+
});
593664
}
594-
return EncodeTwoByteString(
595-
isolate, u16size, [buf, buflen, u16size](uint16_t* dst) {
596-
size_t written = simdutf::convert_valid_utf8_to_utf16(
597-
buf, buflen, reinterpret_cast<char16_t*>(dst));
598-
CHECK_EQ(written, u16size);
599-
});
600665
}
601666

602667
val =

0 commit comments

Comments
 (0)
Back | FazBrowse Home | New Git URL