For the torch links, all torch supports have been removed in this PR:
https://github.com/apache/incubator-mxnet/pull/9072/files
It might because the master is not rebuilt.
2018-01-16 10:53 GMT-08:00 santhosh karuhatty :
> Hello All,
>
> I need help for these 2 broken links.
[
https://issues.apache.org/jira/browse/MXNET-4?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Chris Olivier updated MXNET-4:
--
Labels: mkl performance (was: performance)
> Performance: Refactor Random and ParallelRandom resources to
[
https://issues.apache.org/jira/browse/MXNET-4?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Chris Olivier updated MXNET-4:
--
Labels: performance (was: )
> Performance: Refactor Random and ParallelRandom resources to use MKL for
Hello All,
I need help for these 2 broken links.
1. Need the right link locations for the torch_function and torch_module
below.
URL - https://mxnet.incubator.apache.org/versions/master/how_to/torch.html
Broken Links
─
[
https://issues.apache.org/jira/browse/MXNET-4?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16327562#comment-16327562
]
Chris Olivier commented on MXNET-4:
---
Sounds good.
> Performance: Refactor Random and ParallelRandom
[
https://issues.apache.org/jira/browse/MXNET-4?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Chris Olivier updated MXNET-4:
--
Description:
Refactor Random and ParallelRandom resources to use MKL for MKL builds
Things such as
[
https://issues.apache.org/jira/browse/MXNET-4?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Chris Olivier updated MXNET-4:
--
Summary: Performance: Refactor Random and ParallelRandom resources to use
MKL for MKL builds (was:
[
https://issues.apache.org/jira/browse/MXNET-4?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Chris Olivier updated MXNET-4:
--
Summary: [MXNET-4] Performance: Refactor Random and ParallelRandom
resources to use MKL for MKL builds
Chris Olivier created MXNET-4:
-
Summary: Performance: Refactor Random and ParallelRandom resources
to use MKL for MKL builds
Key: MXNET-4
URL: https://issues.apache.org/jira/browse/MXNET-4
Project:
I agree, you don’t want to run tests until you feel the code is ready, but
stopping everything to fix warnings that aren’t likely to cause any actual
problems just slows down the process. I think that sanity checking makes
perfect sense, but the usual way to do that is to first run some “smoke”
@Christopher: I see your point, but the counter argument would be: "Why
should the project run fairly expensive tests (~20 minutes on few GPU
instances) for code that will require you to amend your commit anyway?" In
normal circumstances I'd completely agree with you and let the full tests
run
While you can debate the "broken windows" theory
(https://en.wikipedia.org/wiki/Broken_windows_theory) I think it has relevance
to code warnings: the more warnings you tolerate, the more likely you are to
end up with other undesirable things in your code. My preference has always
been to treat
I understand. What prevents you from disabling the -Werror flag on
your build configuration? I don't see where's the big issue. We have
already tens of flags to configure the build anyway.
I have been fixing every warning that I came across so far in the main
platforms that I have access. I
So you're proposing to have a stage AFTER test execution which would report
warnings as errors? While this is a good idea, I'm afraid that a fail-fast
would also have its benefits - especially considering that compilation only
takes a few minutes and consumes few resources while test execution
Personally, I don’t like treating warnings as errors because it prevents
compilation from completing and causes you to lose any ability to test the code
and get any other information. Killing the build because of a failed warning
for something that might not matter means that you may not find
I'd vote for having warnings as errors only for CI but not in general
builds which are getting executed by users on their local machine. Just in
case CI misses a warning due to a different version, this could block a
developer from compiling MXNet locally even though it might just be a
warning
Hi Chris
I get the rationale of the point you raise, but In my opinion, and
considering the complexity of C++ and the potential for difficult and
expensive to track bugs, I think this should be enabled by default for
both release and debug. A developer is free to disable warnings in his
own
+1
2018-01-16 9:08 GMT+01:00 Isabel Drost-Fromm :
> Hi,
>
>
>
> Am 16. Januar 2018 01:40:48 MEZ schrieb Steffen Rochel <
> steffenroc...@gmail.com>:
> ># Anybody can add themselves or a team as additional contributors
> ># to get notified about changes in a specific package.
>
Hi,
Am 16. Januar 2018 01:40:48 MEZ schrieb Steffen Rochel
:
># Anybody can add themselves or a team as additional contributors
># to get notified about changes in a specific package.
># See https://help.github.com/articles/about-teams how to setup teams.
>
>Hope we
19 matches
Mail list logo