Benchmarks

This page presents performance benchmark data for the DeGirum Orca AI accelerator, listing frames per second (FPS) for various models under a batch size of 1.

This page provides performance benchmarks for the DeGirum® Orca accelerator across a variety of models. The frames per second (FPS) numbers were generated by running the single_model_performance_test.ipynb notebook on an Orca accelerator (ORCA1). You can reproduce these results by running the same notebook locally or in the DeGirum AI Hub. All FPS numbers assume batch_size=1. We update this page periodically as our compiler and software evolve, adding more models and improving performance.

These benchmarks were last updated on Oct 23, 2023.

Model Name	FPS
efficientnet_es_imagenet--224x224_quant	187
mobiledet_coco--320x320_quant	128
mobilenet_v1_imagenet--224x224_quant	407
mobilenet_v2_imagenet--224x224_quant	360
resnet50_imagenet--224x224_pruned_quant	250
yolo_v5s_face_det--512x512_quant	126