Incubator-mxnet: mxnet-cu100 vs mxnet-cu100mkl vs mxnet
I'm a bit confused as to which packages i need and the benefits of one vs the others.
Currently i have mxnet 1.4.1 installed and the project i have runs fine on a CPU,
I'm using a machine with a GPU that has NVIDIA 418.40.04 drivers and CUDA 10.1 installed.
So ideally id like to use that.
Do i need both mxnet in addition to mxnet-cu100/mkl installed ?
What are the differences mxnet-cu100 vs mxnet-cu100mkl ?
Hoping this answer clears up questions others might have as well.
Charles-Schleich
All 3 comments
Hey, this is the MXNet Label Bot.
Thank you for submitting the issue! I will try and suggest some labels so that the appropriate MXNet community members can help resolve it.
Here are my recommended labels: Installation
mxnet-label-bot
on 16 May 2019
It's a good question.
mxnet-cuXX with cuDNN acceleration, good performance on GPU
mxnet-cuXXmkl with both cuDNN and MKL-DNN library acceleration, good performance on both CPU and GPU.
mxnet-cu92mkl means the package is built with CUDA/cuDNN and MKL-DNN enabled and the CUDA version is 9.2.
pengzhao-intel
on 16 May 2019
Suppose the question is answered. Closing now and feel free to re-open.
pengzhao-intel
on 20 May 2019
Related issues
realbns2008
路
3Comments
zy-huang
路
3Comments
Fzz123
路
3Comments
sbodenstein
路
3Comments
xzqjack
路
3Comments
Most helpful comment
It's a good question.
mxnet-cuXXwith cuDNN acceleration, good performance on GPUmxnet-cuXXmklwith both cuDNN and MKL-DNN library acceleration, good performance on both CPU and GPU.http://mxnet.incubator.apache.org/versions/master/install/index.html?platform=Linux&language=Python&processor=CPU