fast-crc32c is a CRC-32C algorithm implementation for node.js, which uses hardware acceleration (via voxer/sse4_crc32 by Anand Suresh), and fallback to software implementation when hardware acceleration fails.
CRC-32C is faster then CRC-32, which could take advantage of full CPU operand register width (64bit) instead of CRC-32's 8bit.
When using hardware acceleration, CRC-32C is about 7x ~ 9x faster than software implemented CRC-32C.
The 3 tested implementations are:
- sse4_crc32c Hardware accelerated CRC-32C from sse4_crc32
- js_crc32c Javascript implemented CRC-32C
- js_crc32 Javascript implemented CRC-32 from buffer-crc32
$ yarn run benchmark
yarn run v1.17.3
$ node benchmark
sse4_crc32c_hw for inputs 1024B x 3,210,508 ops/sec ±2.09% (87 runs sampled)
sse4_crc32c_hw for inputs 16836819B, avg 2055B x 252 ops/sec ±2.90% (79 runs sampled)
sse4_crc32c_sw for inputs 1024B x 1,358,720 ops/sec ±1.31% (89 runs sampled)
sse4_crc32c_sw for inputs 16836819B, avg 2055B x 96.41 ops/sec ±1.42% (71 runs sampled)
js_crc32c for inputs 1024B x 409,615 ops/sec ±1.46% (89 runs sampled)
js_crc32c for inputs 16836819B, avg 2055B x 26.00 ops/sec ±0.72% (47 runs sampled)
js_crc32 for inputs 1024B x 403,262 ops/sec ±1.55% (90 runs sampled)
js_crc32 for inputs 16836819B, avg 2055B x 24.82 ops/sec ±1.70% (45 runs sampled)
1024B | 16836819B, avg 2055B | |
---|---|---|
sse4_crc32c_hw | 3,210,508 ops/sec | 252 ops/sec |
sse4_crc32c_sw | 1,358,720 ops/sec | 96.41 ops/sec |
js_crc32c | 409,615 ops/sec | 26.00 ops/sec |
js_crc32 | 403,262 ops/sec | 24.82 ops/sec |
npm install fast-crc32c@^2.0.0 --save
# For node version before 6
# npm install fast-crc32c@^1.0.0 --save
var crc32 = require('fast-crc32c');
var result = crc32.calculate(data, initial);
- data required, String|Buffer, data to digest
- initial optional, Number, initial CRC-32C digest