You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@mxnet.apache.org by GitBox <gi...@apache.org> on 2020/05/23 21:28:07 UTC
[GitHub] [incubator-mxnet] JONGGON opened a new issue #18396: Mxnet build error / c++ , python diffrent result in mxnet 1.6.0 / and so on
JONGGON opened a new issue #18396:
URL: https://github.com/apache/incubator-mxnet/issues/18396
1. I think mxnet1.6.0 is unstable.
As Mentioned https://github.com/apache/incubator-mxnet/issues/17874
Mxnet1.6.0 build is still incomleted. . .
2. When I loaded 'json, params' trained in mxnet python in c++. .
Different results are shown
I'm currently testing it with a Yolov3 model.
In mxnet 1.5.1 version, it works
3. The Operation of amp module is different for each version
4. To use mxnet cpp
When using a cmake or Ninja build, it looks like it is being built,
Gpu is not recognized at runtime
My environment
Ubuntu16.04, Ubuntu18.04
Cuda 10.0 ~ 10.2.
These are the problem I have found so far.
Thank you
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [incubator-mxnet] ChaiBapchya commented on issue #18396: Mxnet build error / c++ , python diffrent result in mxnet 1.6.0 / and so on
Posted by GitBox <gi...@apache.org>.
ChaiBapchya commented on issue #18396:
URL: https://github.com/apache/incubator-mxnet/issues/18396#issuecomment-633193494
Hey @JONGGON
> 1. I think mxnet1.6.0 is unstable.
> As Mentioned #17874
In that linked issue, it was found that the steps weren't followed from the correct page.
Please follow steps for Ubuntu + Cuda setup : https://mxnet.apache.org/get_started/ubuntu_setup
> 2\. The Operation of amp module is different for each version
Can you substantiate what do you mean by this? Any reproducible steps & logs for the issue would be great too
> 3\. To use mxnet cpp
> When using a cmake or Ninja build, it looks like it is being built,
> Gpu is not recognized at runtime
Does this mean you get error on `mx.gpu()`? Did you verify if cuda is installed properly along with other nvidia drivers [`nvidia-smi` and `nvcc --version` return correct values?]
What's the value of `mx.runtime.Features()`?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org