R package installation fails on mac osx 10.11

[leocnj]

Following the instruction provided by the mxnet document

install.packages("drat", repos="https://cran.rstudio.com")
drat:::addRepo("dmlc")
install.packages("mxnet")

I run the above commands in RStudio and met the following errors - mostly from the last step

> install.packages("mxnet")
curl: (22) The requested URL returned error: 404 Not Found
Warning in install.packages :
download had nonzero exit status
Warning in install.packages :
cannot open compressed file  /var/folders/df/ctxs2j3d14n8jp16xy_8n8g9nslbq4/T//Rtmpg1hpr7/file76f02c177987', probable reason 'No such file or directory'
curl: (22) The requested URL returned error: 404 Not Found
Warning in install.packages :
download had nonzero exit status
Warning in install.packages :
cannot open compressed file  '/var/folders/df/ctxs2j3d14n8jp16xy_8n8g9nslbq4/T//Rtmpg1hpr7/file76f02c177987', probable reason   'No such file or directory'
Error in install.packages : cannot open the connection

My R SessionInfo() is

R version 3.1.2 (2014-10-31)
Platform: x86_64-apple-darwin13.4.0 (64-bit)

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

loaded via a namespace (and not attached):
[1] tools_3.1.2

[leocnj]

I also tried the second approach (using Rscript) described in the document. Unfortunately, the last step "R CMD INSTALL mxnet_0.5.tar.gz" gave me the following errors

installing to /Library/Frameworks/R.framework/Versions/3.2/Resources/library/mxnet/libs
** R
** demo
** inst
** preparing package for lazy loading
** help
*** installing help indices
** building package indices
** installing vignettes
** testing if installed package can be loaded
Error : .onLoad failed in loadNamespace() for 'mxnet', details:
  call: dyn.load(file, DLLpath = DLLpath, ...)
  error: unable to load shared object '/Library/Frameworks/R.framework/Versions/3.2/Resources/library/mxnet/libs/libmxnet.so':
  dlopen(/Library/Frameworks/R.framework/Versions/3.2/Resources/library/mxnet/libs/libmxnet.so, 10): Library not loaded: @rpath/libcudart.7.5.dylib
  Referenced from: /Library/Frameworks/R.framework/Versions/3.2/Resources/library/mxnet/libs/libmxnet.so
  Reason: image not found
Error: loading failed
Execution halted
ERROR: loading failed

Any thought on how to successfully install R-package? Thanks.

[tqchen]

One comment on second approach, I noticed you build with CUDA, consider add path to libcudart.7.5.dylib to your library path. I guess it is LD_LIBRARY_PATH(it is in linux) but I am not 100 percent sure

[leocnj]

After updating R to the latest 3.2.2, the first approach WORKS!

[tqchen]

Glad it works for you, I would also recommend try the second approach, as it gives you R version with CUDA, while default OSX build from drat seems is CPU only

[thirdwing]

Just to be clear. In the drat repo, we only have CPU version for only R 3.2.*.

[leocnj]

@tqchen and @thirdwing thanks very much. Will try the second approach to enable GPU for the R-package.

[leocnj]

Tried the second approach - compiling the R pkg from source to be able to use GPU in mxnet-R

I have been trying installing the R package (w/ GPU support) on OS X EI Capitan. After doing

• disabling Integrity Protection as suggested by @changkun (see fail to install mxnet using GPU #575)
• set LD_LIBRARY_PATH to include "where the libcudart.7.0.5 is"
• set DYLD_FALLBACK_LIBRARY_PATH to include "where the libcudart.7.0.5 is"; note that this step may not be mandatory.

I finally can compile the R package and a testing running on both CPU and GPU works inside the R sessions initialized in a terminal.

However, the only left issue is that mxnet cannot be loaded inside RStudio (the latest version).

> require(mxnet)
Loading required package: mxnet
Error : .onLoad failed in loadNamespace() for 'mxnet', details:
  call: dyn.load(file, DLLpath = DLLpath, ...)
  error: unable to load shared object '/Library/Frameworks/R.framework/Versions/3.2/Resources/library/mxnet/libs/libmxnet.so':
  dlopen(/Library/Frameworks/R.framework/Versions/3.2/Resources/library/mxnet/libs/libmxnet.so, 10): Library not loaded: @rpath/libcudart.7.5.dylib
  Referenced from: /Library/Frameworks/R.framework/Versions/3.2/Resources/library/mxnet/libs/libmxnet.so
  Reason: image not found

[thirdwing]

This is a known issue I met this summer. I think it has been fixed and I can talk about it with Kevin from Rstudio later.

Can you try to run

Sys.getenv("LD_LIBRARY_PATH")

in RStudio and show me the results?

[leocnj]

Inside Rstudio

> Sys.getenv("LD_LIBRARY_PATH")
[1] ""

In a contrast, in terminal

 echo $LD_LIBRARY_PATH 
/usr/local/cuda/lib:/Users/lchen/cudnn3/cuda/lib:/Developer/NVIDIA/CUDA-7.5/lib:/Users/lchen/torch/install/lib:

[thirdwing]

Exactly the same problem I met during this summer. I will come back to you later.

[danmalter]

install.packages("drat", repos="https://cran.rstudio.com")
library(drat)
drat:::addRepo("dmlc")
install.packages("mxnet")
library(mxnet)

When trying to install from the above code on a Mac (OS X Yosemite Version 10.10.4), I get the following error. Any help would be appreciated. This is with R version 3.2.4

Error : .onLoad failed in loadNamespace() for 'mxnet', details:
call: dyn.load(file, DLLpath = DLLpath, ...)
error: unable to load shared object
'/Library/Frameworks/R.framework/Versions/3.2/Resources/library/mxnet/libs/libmxnet.so': dlopen(/Library/Frameworks/R.framework/Versions/3.2/Resources/library/mxnet/libs/libmxnet.so, 10):
Library not loaded: /usr/local/lib/libgomp.1.dylib
Referenced from:
/Library/Frameworks/R.framework/Versions/3.2/Resources/library/mxnet/libs/libmxnet.so
Reason: image not found
Error: package or namespace load failed for ‘mxnet’

[thirdwing]

Sorry, it is my fault. I built the R package with OpenMP. I will upload a new one soon.

[danmalter]

Thanks for the quick response.

[thirdwing]

I have just upload a new one. Can you try again?

On Mar 23, 2016, at 9:21 PM, Danny Malter notifications@github.com wrote:

Thanks for the quick response.

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub #771 (comment)