[arch-commits] Commit in python-pytorch/trunk (PKGBUILD)

[Sven-Hendrik Haase]

    Date: Tuesday, October 16, 2018 @ 21:27:02
  Author: svenstaro
Revision: 394861

upgpkg: python-pytorch 1.0rc1-3

Fix CUDNN option

Modified:
  python-pytorch/trunk/PKGBUILD

----------+
 PKGBUILD |    7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

Modified: PKGBUILD
===================================================================
--- PKGBUILD	2018-10-16 21:23:31 UTC (rev 394860)
+++ PKGBUILD	2018-10-16 21:27:02 UTC (rev 394861)
@@ -5,7 +5,7 @@
 pkgname=("python-pytorch" "python-pytorch-cuda")
 _pkgname="pytorch"
 pkgver=1.0rc1
-pkgrel=2
+pkgrel=3
 pkgdesc="Tensors and Dynamic neural networks in Python with strong GPU acceleration"
 arch=('x86_64')
 url="https://pytorch.org"
@@ -113,7 +113,7 @@
   # Uncomment and modify the following line to enable Intel MKL and magma support
   # export CMAKE_PREFIX_PATH=/opt/intel/mkl/include:/opt/intel/mkl/lib/intel64:/opt/magma \
 
-  msg2 "Building without cuda"
+  echo "Building without cuda"
   export NO_CUDA=1
   export WITH_CUDNN=0
   export USE_OPENCV=1
@@ -122,10 +122,11 @@
   cd "$srcdir/${_pkgname}-${pkgver}"
   python setup.py build
 
-  msg2 "Building with cuda"
+  echo "Building with cuda"
   export CC=gcc-7
   export CXX=g++-7
   export NO_CUDA=0
+  export WITH_CUDNN=1
   export CUDAHOSTCXX=g++-7
   export CUDA_HOME=/opt/cuda
   export CUDNN_LIB_DIR=/opt/cuda/lib64