未验证 提交 158d5674 编写于 作者: T Tao Luo 提交者: GitHub

Merge pull request #8755 from QiJune/disable_nccl

disable nccl in memopt fit a line unittest
......@@ -31,7 +31,7 @@ use_nccl = False
place = fluid.CPUPlace()
if fluid.core.is_compiled_with_cuda():
device_type = 'CUDA'
use_nccl = True
use_nccl = False
place = fluid.CUDAPlace(0)
places = fluid.layers.get_places(device_count=0, device_type=device_type)
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册