[incubator-mxnet] branch master updated: Fix tvm dependency for docker (#12479)

[ha...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

haibin pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git


The following commit(s) were added to refs/heads/master by this push:
     new 2bcfc08  Fix tvm dependency for docker (#12479)
2bcfc08 is described below

commit 2bcfc08019d82d725d9b3aa0dc7794c0d63f8694
Author: Tianqi Chen <tq...@users.noreply.github.com>
AuthorDate: Fri Sep 7 13:17:34 2018 -0800

    Fix tvm dependency for docker (#12479)
---
 ci/docker/install/ubuntu_tvm.sh     | 14 +++++++-------
 tests/python/gpu/test_tvm_bridge.py |  1 -
 2 files changed, 7 insertions(+), 8 deletions(-)

diff --git a/ci/docker/install/ubuntu_tvm.sh b/ci/docker/install/ubuntu_tvm.sh
index 4f5cb42..2ee4e53 100755
--- a/ci/docker/install/ubuntu_tvm.sh
+++ b/ci/docker/install/ubuntu_tvm.sh
@@ -25,14 +25,14 @@ cd tvm
 # This is a stable tag that support MXNet TVM bridge.
 # We use this since support for mxnet bridge just checked
 # into master and there is yet a version tag
-git checkout 30eaf463e34d7c301357c31a010945d11df16537
+git checkout v0.4
+
+cp cmake/config.cmake .
+echo set\(USE_CUDA /usr/local/cuda\) >> config.cmake
+echo set\(USE_LLVM llvm-config-5.0\) >> config.cmake
+echo set\(USE_RPC ON\) >> config.cmake
+echo set\(USE_GRAPH_RUNTIME ON\) >> config.cmake
 
-cp make/config.mk
-echo USE_CUDA=1 >> config.mk
-echo LLVM_CONFIG=llvm-config-5.0 >> config.mk
-echo USE_RPC=1 >> config.mk
-echo USE_GRAPH_RUNTIME=1 >> config.mk
-echo CUDA_PATH=/usr/local/cuda >> config.mk
 make -j$(nproc)
 
 cd python
diff --git a/tests/python/gpu/test_tvm_bridge.py b/tests/python/gpu/test_tvm_bridge.py
index c3cf2dd..5c87536 100644
--- a/tests/python/gpu/test_tvm_bridge.py
+++ b/tests/python/gpu/test_tvm_bridge.py
@@ -21,7 +21,6 @@ import mxnet as mx
 import numpy as np
 import unittest
 
-@unittest.skip("Flaky test https://github.com/apache/incubator-mxnet/issues/12473")
 def test_tvm_bridge():
     # only enable test if TVM is available
     try: