Can I use the same calib.table while building different batch size of .eninge file?

marcoslucianops / DeepStream-Yolo

NVIDIA DeepStream SDK 7.0 / 6.4 / 6.3 / 6.2 / 6.1.1 / 6.1 / 6.0.1 / 6.0 / 5.1 implementation for YOLO models

MIT License

1.39k stars 344 forks source link

Can I use the same calib.table while building different batch size of .eninge file? #414

Open shao77622 opened 11 months ago

shao77622 commented 11 months ago

Firstly, I build the model_b1_gpu0_int8.engine file with calibration, and I got the calib.table file. Then I need to build other batch size of .engine file for different batch inference. Do I need to make calibration every time? thank you!

marcoslucianops commented 11 months ago

You don't need to recalibrate. Just keep the calib.table file and it will use the generated calibration.