OneDNN_EX

Enviroments

• Windows 10 laptop
• CPU i7-11375H
• Opencv 3.4.5
• OneDNN 2.50 (2021.4)

OneDNN Simple Classification model (cpu, F32)

• resnet18 Model
• Function abstraction in an easy-to-use form
• Primitive function validation
• Weight loader
• Data loader
• Image preprocess
• Modeling
• Match all results with PyTorch
• Optimization
• Performace evaluation(Execution time of 100 iteration for one 224x224x3 image)

	Pytorch	OneDNN	OneDNN	OneDNN
Description	general	Specific data type	Any data type	Any data type + fused post ops
Avg Duration time [ms]	27.59 ms	67.65 ms	24.40 ms	22.81 ms
FPS [frame/sec]	36 fps	15 fps	41 fps	44 fps
File	Resnet18_py/inference.py	resnet18_v1.cpp	resnet18_v2.cpp	resnet18_v3.cpp

Reference

• oneDNN github : https://github.com/oneapi-src/oneDNN#installation
• onednn v2.50 documentation :https://oneapi-src.github.io/oneDNN/
• download : https://www.intel.com/content/www/us/en/developer/tools/oneapi/onednn.html#gs.i239i5