src: implement FastByteLengthUtf8 with simdutf::utf8_length_from_latin1

PR-URL: #50840
Reviewed-By: Yagiz Nizipli <[email protected]>
Reviewed-By: Joyee Cheung <[email protected]>
Reviewed-By: James M Snell <[email protected]>
Reviewed-By: Vinícius Lourenço Claro Cardoso <[email protected]>

Author: lemire

benchmark/buffers/buffer-bytelength-string.js

@@ -2,7 +2,8 @@
 const common = require('../common');
 
 const bench = common.createBenchmark(main, {
-  type: ['one_byte', 'two_bytes', 'three_bytes', 'four_bytes'],
+  type: ['one_byte', 'two_bytes', 'three_bytes',
+         'four_bytes', 'latin1'],
   encoding: ['utf8', 'base64'],
   repeat: [1, 2, 16, 256], // x16
   n: [4e6],
@@ -14,6 +15,8 @@ const chars = {
   two_bytes: 'ΰαβγδεζηθικλμνξο',
   three_bytes: '挰挱挲挳挴挵挶挷挸挹挺挻挼挽挾挿',
   four_bytes: '𠜎𠜱𠝹𠱓𠱸𠲖𠳏𠳕𠴕𠵼𠵿𠸎𠸏𠹷𠺝𠺢',
+  latin1: 'Un homme sage est supérieur à toutes ' +
+    'les insultes qui peuvent lui être adressées, et la meilleure réponse est la patience et la modération.',
 };
 
 function getInput(type, repeat, encoding) {

src/node_buffer.cc

@@ -743,13 +743,17 @@ void SlowByteLengthUtf8(const FunctionCallbackInfo<Value>& args) {
 
 uint32_t FastByteLengthUtf8(Local<Value> receiver,
                             const v8::FastOneByteString& source) {
-  uint32_t result = 0;
+  // For short inputs, the function call overhead to simdutf is maybe
+  // not worth it, reserve simdutf for long strings.
+  if (source.length > 128) {
+    return simdutf::utf8_length_from_latin1(source.data, source.length);
+  }
   uint32_t length = source.length;
+  uint32_t result = length;
   const uint8_t* data = reinterpret_cast<const uint8_t*>(source.data);
   for (uint32_t i = 0; i < length; ++i) {
     result += (data[i] >> 7);
   }
-  result += length;
   return result;
 }