benchmarks

Benchmark scripts

Training and inference scripts with Intel-optimized MKL

Prerequisites

The benchmarking scripts can be run on Linux and require the following dependencies to be installed:

• Docker
• Python 2.7 or later
• git
• wget for downloading pre-trained models

Use Cases

Use Case	Framework	Model	Mode	Benchmarking Instructions
Adversarial Networks	TensorFlow	DCGAN	Inference	FP32
Content Creation	TensorFlow	DRAW	Inference	FP32
Image Recognition	TensorFlow	Inception ResNet V2	Inference	FP32
Image Recognition	TensorFlow	Inception V3	Inference	Int8 FP32
Image Recognition	TensorFlow	MobileNet V1	Inference	FP32
Image Recognition	TensorFlow	ResNet 101	Inference	Int8 FP32
Image Recognition	TensorFlow	ResNet 50	Inference	Int8 FP32
Image Recognition	TensorFlow	SqueezeNet	Inference	FP32
Image Segmentation	TensorFlow	Mask R-CNN	Inference	FP32
Image Segmentation	TensorFlow	UNet	Inference	FP32
Language Translation	TensorFlow	Transformer Language	Inference	FP32
Object Detection	TensorFlow	Fast R-CNN	Inference	Int8 FP32
Object Detection	TensorFlow	R-FCN	Inference	FP32
Object Detection	TensorFlow	SSD-MobileNet	Inference	FP32
Recommendation	TensorFlow	NCF	Inference	FP32
Recommendation	TensorFlow	Wide & Deep Large Dataset	Inference	Int8 FP32
Recommendation	TensorFlow	Wide & Deep	Inference	FP32
Text-to-Speech	TensorFlow	WaveNet	Inference	FP32