Upgrading from tensorrt 7.1.0 to 7.1.3

AI & Data Science Deep Learning (Training & Inference) TensorRT
,
1

Description

I developed a plugin in tensorrt version 7.1.0 and it was working great without any problem the engine was refittable and almost every minute i refite it. now I upgraded my tensorrt to 7.1.3 and the same code without any change taking inference witout any problem but after reffing the engine when taking inference it shows this error and abort.

terminate called without an active exception
Aborted (core dumped)

I must say the boolean that refitter passes is true and it seems there is no problem with refitting but after refitting it dosent take inference any more. and I created engine with kfp16 and kSTRICT_TYPES flags

Environment

TensorRT Version : 7.1.3
GPU Type : jetson tx2
jetpack Version : 4.4
Python Version (if applicable) : 3.6.9

2

TensorRT Release 7.1.3 (Linux and Windows users).

Could you please try TensorRT Release 7.2.0 (Linux, JetPack, and PowerPC users)?

Thanks

3

Tancks for replying. i saw somewhere tensorrt on jetson tx2 only can be installed by jetpack.
i check jetPack right now and it is with tensorrt 7.1.3 still.
if there is another way for upgrading it I would be happy to now.

4

https://docs.nvidia.com/jetson/jetpack/release-notes/
Sorry, latest Jetpack support TRT 7.1.3 only. Request you to use this Jetpack to upgrade to TRT 7.1.3