Skip to content

Latest commit

 

History

History

benchmarks

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
README.md
README.md
 
 
batchnormalization3d_benchmark.ts
batchnormalization3d_benchmark.ts
 
 
benchmark.ts
benchmark.ts
 
 
benchmark_util.ts
benchmark_util.ts
 
 
benchmarks.ts
benchmarks.ts
 
 
chartjs.d.ts
chartjs.d.ts
 
 
conv_benchmarks.ts
conv_benchmarks.ts
 
 
demo-header.html
demo-header.html
 
 
index.html
index.html
 
 
karma.conf.js
karma.conf.js
 
 
karma_test.ts
karma_test.ts
 
 
math-benchmark-run-groups.ts
math-benchmark-run-groups.ts
 
 
math-benchmark.html
math-benchmark.html
 
 
matmul_benchmarks.ts
matmul_benchmarks.ts
 
 
package.json
package.json
 
 
polymer-spec.ts
polymer-spec.ts
 
 
pool_benchmarks.ts
pool_benchmarks.ts
 
 
reduction_ops_benchmark.ts
reduction_ops_benchmark.ts
 
 
tsconfig.json
tsconfig.json
 
 
unary_ops_benchmark.ts
unary_ops_benchmark.ts
 
 
yarn.lock
yarn.lock
 
 

README.md

TensorFlow.js Benchmark Tools

This is a micro benchmark to measure the performance of TensorFlow.js kernel ops.

Usage

First install all dependencies. Benchmark tools refers local TensorFlow.js code instead of release versions.

$ yarn

Launch the server to host benchmark application.

$ yarn server

http://localhost:8080 show the benchmark tool for various kind of kernel ops.

  • Batch Normalization 3D
  • Matrix multiplication
  • Convolutional Ops
  • Pooling Ops
  • Unary Ops
  • Reduction Ops

Each benchmark suite runs kernel ops with specific size of input in target backend implementation. This benchmark tools support following backends for now.

  • CPU
  • WebGL