0

Mxnet is supposed to build and run, on CPU as well as on GPU, for multiple OSs including Windows.

I'm trying to build mxnet from source on Windows Server 2016 that has NVIDIA K80 GPU on it.

I followed all the instructions in https://mxnet.incubator.apache.org/get_started/windows_setup.html but not able to move past the point of building mxnet in Visual Studio 2013.

The error I'm seeing is

'mshadow::cuda::AddTakeGrad' : ambiguous call to overloaded function indexing_op.h

If I fix this generic call to AddTakeGrad to make it a specific call to mshadow::cuda::, then some other polymorphic function ends up with the same error and so on ...

I tried searching a lot to find if anyone was successful in building mxnet for windows (on both cpu mode and gpu mode) but couldn't find any.

Question: Has anyone been able to successfully build mxnet on Windows? If so, could you help with this error as well as any specific instructions to get it to build for both cpu mode as well as gpu mode?

Arun K. Buduri
  • 5
  • 2
  • Is there any reason for not using the prebuilt lib? – Qiang Kou Sep 01 '17 at 14:33
  • Thanks @QiangKou ... i'm exploring it currently though it is not clear if it supports both cpu and gpu mode. I' trying to build and run Deformable Convnets - https://github.com/msracver/Deformable-ConvNets ... for Ubuntu, OS X and Windows. So far I have it running on Ubuntu but having trouble building on Windows .... I don't intend on changing any mxnet source code, so if prebuilt works on both cpu and gpu and on all platforms, that will be great. However, i'm also curious to find out why i'm having these build errors. – Arun K. Buduri Sep 01 '17 at 16:37
  • I am sorry that I am not a Windows expert for your problem. However, we provide the daily Windows build for CPU and GPU. You can find them from https://github.com/yajiedesign/mxnet/releases – Qiang Kou Sep 01 '17 at 18:48
  • @QiangKou Thank You!! Looks like the official Releases link doesn't have prebuilt binaries after 20160531. Is yours the official mxnet or just personal? Just curious as to why prebuilt binaries are not available in "mxnet/releases"? We are taking your latest binaries and replacing it in the folders from 20160531. I'll let you know if compiling Deformable Convnets worked with it. – Arun K. Buduri Sep 01 '17 at 19:34
  • This is official prebuilt binaries and it is not compiled by me. This used to be in the "mxnet/releases" and are removed temporarily. The links will be added back after we formalize the release process. – Qiang Kou Sep 01 '17 at 19:41
  • @QiangKou Great, thanks. For one of the recent issues opened "full vs 2015 solution files!", you mentioned you generate solution file using cmake. Can you share any other specific steps to build mxnet for windows? For ex., Latest rc2 release notes mentions to use the source code tarball and not to clone from github. – Arun K. Buduri Sep 01 '17 at 19:57
  • @QiangKou We were able to get Deformable Convnet demos running on a gpu-based machine using the mxnet prebuilt binaries (link you provided). Now we are trying to run the demos in cpu-only mode and running into this error ... does this mean mxnet cpu mode isnt inplemented? ... [02:05:01] d:\program files (x86)\jenkins\workspace\mxnet\mxnet\src\operator\contrib\./nn/deformable_im2col.h:113: only implemented in GPU – Arun K. Buduri Sep 07 '17 at 09:48
  • Some operators might only have GPU implementation. I suggest you open an issue on github. – Qiang Kou Sep 07 '17 at 14:20
  • @QiangKou I just realized that Qiang. Could you help with any info on how to implement the cpu mode for deformable operators? Can we take the code from the similar gpu-mode function and copy it in cpu-mode function? – Arun K. Buduri Sep 07 '17 at 21:53

1 Answers1

0

These days it should be possible to just pip install.

Simon Corston-Oliver
  • 226
  • 1
  • 6