Hi all, I want know two informations.
- How to find out how many cores that an engine is being executed
- Is it possible to create multiple tensorrt engines
- Is it possible to create multiple execution contexts from the same engine
If it is possible then please let me know how this can be done. Please clarify these doubts.
Thanks and Regards
Nagaraj Trivedi
Hi,
1. No such info but you can monitor GPU utilization with tegrastats.
2. Yes.
3. Yes, you can create multiple contexts from the same serialized engine model.
Examples can be found in trtexec:
https://github.com/NVIDIA/TensorRT/blob/release/8.5/samples/trtexec/README.md#example-6-tune-throughput-with-multi-streaming
$ /usr/src/tensorrt/bin/trtexec --help
...
--streams=N Instantiate N engines to use concurrently (default = 1)
...
Thanks.
system
Closed
5
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.