Performance

测试环境

  • AWS EC2 eu-central-1 c4.xlarge (4x E5-2666 v3 @ 2.90GHz)
  • Ubuntu 16.04.1 LTS (HVM, SSD, 20161115, ami-82cf0aed)
  • Node.js v6.9.1

竞争者

  • jimp v0.2.27 - Image processing in pure JavaScript. Bilinear interpolation only.
  • lwip v0.0.9 - Wrapper around CImg. Compiles outdated, insecure dependencies from source.
  • mapnik v3.5.14 - Whilst primarily a map renderer, Mapnik contains bitmap image utilities.
  • imagemagick-native v1.9.3 - Wrapper around libmagick++, supports Buffers only.
  • imagemagick v0.1.3 - Supports filesystem only and "has been unmaintained for a long time".
  • gm v1.23.0 - Fully featured wrapper around GraphicsMagick's gm command line utility.
  • sharp v0.17.0 / libvips v8.4.2 - Caching within libvips disabled to ensure a fair comparison.

任务

Decompress a 2725x2225 JPEG image, resize to 720x588 using Lanczos 3 resampling (where available), then compress to JPEG at a "quality" setting of 80.

结果

模块 输入 输出 Ops/sec Speed-up
jimp (bilinear) buffer buffer 1.06 1.0
lwip buffer buffer 1.87 1.8
mapnik buffer buffer 2.91 2.7
imagemagick-native buffer buffer 4.03 3.8
imagemagick file file 7.10 6.7
gm buffer buffer 7.08 6.7
gm file file 7.10 6.7
sharp stream stream 27.61 26.0
sharp file file 28.41 26.8
sharp buffer file 28.71 27.1
sharp file buffer 28.60 27.0
sharp buffer buffer 29.08 27.4

Greater libvips performance can be expected with caching enabled (default) and using 8+ core machines, especially those with larger L1/L2 CPU caches.

The I/O limits of the relevant (de)compression library will generally determine maximum throughput.

基准测试先决条件

Requires both ImageMagick and GraphicsMagick:

brew install imagemagick
brew install graphicsmagick
sudo apt-get install imagemagick libmagick++-dev graphicsmagick
sudo yum install ImageMagick-devel ImageMagick-c++-devel GraphicsMagick

运行基准测试

git clone https://github.com/lovell/sharp.git
cd sharp
npm install
cd test/bench
npm install
npm test