When i use nvprof, i can get the unified memory profiling result at the end of the program. And when i use nvvp, i can get the realtime information about the unified memory traffic information.
But How can i get the unified memory traffic information with API? I want to know how much data has transfer to Device already?
======== Unified Memory profiling result:
Device “Tesla V100-PCIE-32GB (0)”
Count Avg Size Min Size Max Size Total Size Total Time Name
138 59.362KB 4.0000KB 980.00KB 8.000000MB 988.4800us Host To Device
24 170.67KB 4.0000KB 0.9961MB 4.000000MB 347.1680us Device To Host
9 - - - - 2.670272ms Gpu page fault groups
Total CPU Page faults: 36
- Enable unified memory profiling - If selected for the GPU that supports Unified Memory, the Unified Memory related memory traffic to and from each GPU is collected on your system and displayed on timeline.
- see: https://devblogs.nvidia.com/transitioning-nsight-systems-nvidia-visual-profiler-nvprof/