You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@mxnet.apache.org by zh...@apache.org on 2017/12/19 06:41:18 UTC
[incubator-mxnet] branch v1.0.0 updated: fix multi worker
dataloader deadlock (#9126)
This is an automated email from the ASF dual-hosted git repository.
zhasheng pushed a commit to branch v1.0.0
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git
The following commit(s) were added to refs/heads/v1.0.0 by this push:
new 4ef905d fix multi worker dataloader deadlock (#9126)
4ef905d is described below
commit 4ef905d8367fd838696b30b677cc03ed28de8692
Author: Eric Junyuan Xie <pi...@users.noreply.github.com>
AuthorDate: Mon Dec 18 21:36:03 2017 -0800
fix multi worker dataloader deadlock (#9126)
---
python/mxnet/gluon/data/dataloader.py | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/python/mxnet/gluon/data/dataloader.py b/python/mxnet/gluon/data/dataloader.py
index beb228e..48ff4ef 100644
--- a/python/mxnet/gluon/data/dataloader.py
+++ b/python/mxnet/gluon/data/dataloader.py
@@ -220,10 +220,11 @@ class DataLoader(object):
for idx, batch in enumerate(self._batch_sampler):
key_queue.put((idx, batch))
+ num_batches = idx + 1
data_buffer = {}
curr_idx = 0
- for _ in range(len(self._batch_sampler)):
+ for _ in range(num_batches):
idx, batch = data_queue.get()
data_buffer[idx] = batch
while curr_idx in data_buffer:
--
To stop receiving notification emails like this one, please contact
['"commits@mxnet.apache.org" <co...@mxnet.apache.org>'].