This repository has been archived by the owner on Nov 17, 2023. It is now read-only.
Fixed and re-enables TensorRT steps #14960
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
lebeg
reviewed
May 15, 2019
| @@ -18,7 +18,7 @@ | |||
| # | |||
| # Dockerfile to run MXNet on Ubuntu 16.04 for CPU | |||
|
|
|||
| FROM nvidia/cuda:10.0-cudnn7-devel | |||
| FROM nvidia/cuda:10.0-cudnn7-devel-ubuntu16.04 | |||
There was a problem hiding this comment.
Ubuntu 18 was intentional, I think
@KellenSunderland
There was a problem hiding this comment.
I'll remove that - and we can move it to a different PR if necessary
There was a problem hiding this comment.
Not a massive deal either way. I think both are fairly well supported.
perdasilva
force-pushed
the
tensorrt_fix
branch
2 times, most recently
from
098c744 to
a9549b5
Compare
lebeg
approved these changes
May 15, 2019
perdasilva
changed the title
perdasilva
force-pushed
the
tensorrt_fix
branch
2 times, most recently
from
b079042 to
165b798
Compare
perdasilva
changed the title
perdasilva
changed the title
|
@KellenSunderland Could you please review/merge this if it looks ok for you? |
roywei
approved these changes
May 15, 2019
|
@roywei thank you ^^ I'm not a comitter, so I can't T_T |
stu1130
approved these changes
May 15, 2019
haohuanw
pushed a commit
to haohuanw/incubator-mxnet
that referenced
this pull request
Jun 23, 2019
* Pins libnvinfer versions * Sets cudnn to version 7.5.0 in tensorrt environment * Re-enables TensorRT stages
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
TensorRT steps were blocking CI (see #14961). It seems there's been a change in the nvidia apt repository. Then
libnvinfer-devseems to have been updated and no longer was getting installed properly. This PR updates the TensorRT environment pinning these dependencies.After the initial commit, the compilation stage was passing, but the test stage failed. It complained that TensorRT had been compiled with cuDNN 7.5.0 but was linking against 7.5.1 (which comes stock in the nvidia/cuda:10.0-cudnn7-devel image used as the base image). So, in a subsequent commit, I've had to pin the cuDNN version to 7.5.0.
Checklist
Essentials
Please feel free to remove inapplicable items for your PR.
Changes
Pins TensorRT dependency versions and cudnn version