We can observe CPU and GPU utility from htop/nvtop. As far as I know, GPU utility is measured from average use of GPU resources (arithmetic, logic, memory) over a period of time. Frequently being used does not ensure high GPU utility. I am wondering if the way CPU utility is calculated is the same as GPU utility. Does htop consider average use of different RESOURCES, or just count if it was used at sampled time instants?
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Monitoring GPU Power Consumption | 1 | 5057 | December 1, 2010 | |
| GPU MHZ Utilization | 0 | 550 | January 24, 2018 | |
| overall time consumption computation how to compute how much time my GPU code is consuming ? | 0 | 1117 | May 18, 2009 | |
| GPU real-time monitoring in Windows using CUDA | 2 | 2849 | June 26, 2014 | |
| How can I estimate the overall execution time of CPU and GPU separately? | 0 | 591 | December 21, 2021 | |
| Monitoring GPU Utilization "Top" like utility for GPU | 8 | 6374 | July 28, 2010 | |
| How is GPU utilisation calculated (as returned by nvmlDeviceGetUtilizationRates(...))? | 0 | 624 | March 4, 2020 | |
| GPU Computing | 1 | 773 | April 17, 2013 | |
| Confused about GPU vs CPU speed in multiplication | 8 | 6562 | February 19, 2009 | |
| Relation between Utilization of CPU and GPU in HPC and Potential Reasons for Not Achieving Stable GPU Utilization | 1 | 434 | October 7, 2022 |