You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@mxnet.apache.org by "Kellen Sunderland (JIRA)" <ji...@apache.org> on 2018/11/25 05:26:00 UTC
[jira] [Commented] (MXNET-703) Integrate TensorRT Runtime
[ https://issues.apache.org/jira/browse/MXNET-703?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16698068#comment-16698068 ]
Kellen Sunderland commented on MXNET-703:
-----------------------------------------
No worries. What you're asking is clear. So the underlying library
(onnx-tensorrt) moved to this version and opset soon after migrating to
TRT5. Almost immediately we began seeing serious correctness regressions
with the old onnx version and opset, so updating them to a supported
version seems like the path of least resistance to enable TRT5 support.
After changing the opset and updating the library the output is exactly as
we'd expect (i.e. matching MXNets output). So basically we want to update
onnx-tensorrt to fix bugs and support new hardware, to do so we need to
serialize onnx in exactly this format.
On Nov 24, 2018 7:10 PM, "ASF GitHub Bot (JIRA)" <ji...@apache.org> wrote:
[
https://issues.apache.org/jira/browse/MXNET-703?focusedWorklogId=169187&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-169187
]
ASF GitHub Bot logged work on MXNET-703:
----------------------------------------
Author: ASF GitHub Bot
Created on: 25/Nov/18 03:09
Start Date: 25/Nov/18 03:09
Worklog Time Spent: 10m
Work Description: marcoabreu commented on a change in pull request
#13310: [MXNET-703] Update to TensorRT 5, ONNX IR 3. Fix inference bugs.
URL:
https://github.com/apache/incubator-mxnet/pull/13310#discussion_r236059363
##########
File path: src/operator/contrib/nnvm_to_onnx.cc
##########
@@ -78,8 +84,13 @@ op::TRTParam ConvertNnvmGraphToOnnx(
ModelProto model_proto;
// Need to determine IR versions and features to support
- model_proto.set_ir_version(static_cast<int64>(2));
+ auto opset_proto = model_proto.add_opset_import();
+ opset_proto->set_version(static_cast<int64>(8));
Review comment:
I see. To me it's just not clear why these specific versions have been
chosen and what will happen if either side gets updated.
I'm on my phone, so sorry for being so Brief. Let me know if you would
like be to elaborate my question
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Issue Time Tracking
-------------------
Worklog Id: (was: 169187)
Time Spent: 32.5h (was: 32h 20m)
compatible sub-graphs with library calls to TRT.
https://cwiki.apache.org/confluence/display/MXNET/Runtime+Integration+with+TensorRT
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
> Integrate TensorRT Runtime
> --------------------------
>
> Key: MXNET-703
> URL: https://issues.apache.org/jira/browse/MXNET-703
> Project: Apache MXNet
> Issue Type: Improvement
> Reporter: Kellen Sunderland
> Priority: Major
> Time Spent: 32.5h
> Remaining Estimate: 0h
>
> We would like to integrate TensorRT in MXNet to transparently substitute compatible sub-graphs with library calls to TRT.
>
> See design proposal for more information: https://cwiki.apache.org/confluence/display/MXNET/Runtime+Integration+with+TensorRT
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@mxnet.apache.org
For additional commands, e-mail: issues-help@mxnet.apache.org