Skip to content
quick and dirty inference time benchmark for TFLite gles delegate
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
.idea
app
gradle/wrapper Initial commit Jan 27, 2019
.gitignore Initial commit Jan 27, 2019
LICENSE Create LICENSE Jan 28, 2019
README.md Update README.md Mar 15, 2019
build.gradle update dependencies Mar 31, 2019
gradle.properties Initial commit Jan 27, 2019
gradlew Initial commit Jan 27, 2019
gradlew.bat Initial commit Jan 27, 2019
settings.gradle

README.md

glDelegateBench

quick and dirty inference time benchmark for TFLite gles delegate

The TensorFlow team announced TFLite GPU delegate and published related docs [2][3] in Jan 2019. But except Mobilenet V1 classifier, there is no publicly available app to evaluate it, so I wrote a quick and dirty app to evaluate other models.

For the 4 public models mentioned in [1], I got the following numbers on Pixel 2.

model name CPU 1 thread (ms) CPU 4 threads (ms) GPU (ms)
Mobilenet 150 75 21
PoseNet 183 96 40
DeepLab V3 219 131 91
Mobilenet SSD V2 COCO 264 158 49

On Xiaomi Mi 9, I got

model name CPU 1 thread (ms) CPU 4 threads (ms) GPU (ms)
Mobilenet 39 35 15
PoseNet 48 47 19
DeepLab V3 61 64 65
Mobilenet SSD V2 COCO 69 75 36

Check https://github.com/freedomtan/glDelegateBenchmark/ for iOS code

[1] https://medium.com/tensorflow/tensorflow-lite-now-faster-with-mobile-gpus-developer-preview-e15797e6dee7

[2] https://www.tensorflow.org/lite/performance/gpu

[3] https://www.tensorflow.org/lite/performance/gpu_advanced

You can’t perform that action at this time.