[1.x] Enable CUDA Graphs for TRT

[Kh4L]

Description

This PR enables CUDA Graphs for TRT FP32 and FP16.
CUDA Graph for INT8 requires more complex work and will be added in a future PR.

[mxnet-bot]

Hey @Kh4L , Thanks for submitting the PR
All tests are already queued to run once. If tests fail, you can trigger one or more tests again with the following commands:

• To trigger all jobs: @mxnet-bot run ci [all]
• To trigger specific jobs: @mxnet-bot run ci [job1, job2]

---

CI supported jobs: [clang, edge, unix-gpu, windows-gpu, sanity, website, windows-cpu, centos-gpu, miscellaneous, centos-cpu, unix-cpu]

---

Note:
Only following 3 categories can trigger CI :PR Author, MXNet Committer, Jenkins Admin.
All CI tests must pass before the PR can be merged.

[samskalicky]

LGTM

[xxxpsyduck]

Sorry if commenting here is inappropriate. But I'm looking for a document about how to use mxnet (1.7) with tensorrt but I can't find it anywhere. Can you give me a quick guide?

[samskalicky]

Sorry if commenting here is inappropriate. But I'm looking for a document about how to use mxnet (1.7) with tensorrt but I can't find it anywhere. Can you give me a quick guide?

Hi @xxxpsyduck heres the docs page for TensorRT: https://mxnet.apache.org/versions/1.7/api/python/docs/tutorials/performance/backend/tensorrt/tensorrt.html If you have any more questions, please file a separate GitHub issue with the questions and we'll discuss more there. Good luck!

[xxxpsyduck]

thanks a lot!