Resunet tensorrt8.2 speed and video memory record table on Jetson Xavier NX (continuously supplemented later)

ResUnet-TensorRT8.2 Speed and video memory record table

precision	Pattern	Image size	Number of categories	batch	Number of threads	Reasoning time	Complete processing time	memory
FP32	20W 2CORE	512*640	2	1	1	168ms	179ms	2.2G
FP32	20W 2CORE	512*640	2	1	2	172ms	184ms	3.0G
FP16	20W 2CORE	512*640	2	1	1	58ms	68ms	1.6G
FP16	20W 2CORE	512*640	2	1	2	58ms	68ms	1.9G
FP32	20W 2CORE	512*612	6	1	1	167ms	209ms	2.2G
FP32	20W 2CORE	512*612	6	1	2	170ms	234ms	3.0G
FP16	20W 2CORE	512*612	6	1	1	57ms	97ms	1.6G
FP16	20W 2CORE	512*612	6	1	2	58ms	106ms	1.9G

explain ：
1. Pattern means Jetson Power consumption mode of the device , For my Jetson XAVIER NX Come on , All in all 8 Patterns , If you want to achieve maximum reasoning speed , choice 20W 2CORE Pattern . There is a... In the upper right corner of the main interface MODE The choice of , choice 20W 2CORE Mode is enough .
（ My choice 20W 6CORE Test and follow 20W 2CORE almost , Only soon 1ms, So choose 20W 2CORE that will do ）

2. Reasoning time refers to the average time of each graph doInference( The perform cudaMemcpyAsync） The reasoning time required .
Complete processing time reasoning time plus pre-processing and post-processing time .

3. about Jetson For equipment ,CPU and GPU share , So video memory is memory . about Jetson XAVIER NX Total memory 8G.
The viewing method cannot be used directly nvidia-smi The command line , Must install jetson-stats.
For specific operation methods, please refer to the following blog .
Jetson View video memory on device （ Memory ）——jetson-stats

4. Why do I have 8 A model , And under this system TensorRT yes 8.2.1.8 edition , No 7 edition , Guess the reason is that the image used in the initial burning system is relatively new .
And compared to TensorRT7 edition , The speed is fast 20ms, See my previous blog for details .
Jetson XAVIER NX On ResUnet-TensorRT7 Speed and video memory record table ( Continuous follow-up )