Github user rxin commented on the issue:
https://github.com/apache/spark/pull/14762
hm since github switches over to the new code review functionality, I've
been getting only parts of the comment notifications and don't really see all
the notifications anymore. This is weird and I mis
Github user srowen commented on the issue:
https://github.com/apache/spark/pull/14762
Ah, I see. What do you think, in any event? this could be tweaked if you
had a concern.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well
Github user srowen commented on the issue:
https://github.com/apache/spark/pull/14762
Merged to master
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or i
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14762
**[Test build #3292 has
finished](https://amplab.cs.berkeley.edu/jenkins/job/NewSparkPullRequestBuilder/3292/consoleFull)**
for PR 14762 at commit
[`ea44a77`](https://github.com/apache/spark/commit
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14762
**[Test build #3292 has
started](https://amplab.cs.berkeley.edu/jenkins/job/NewSparkPullRequestBuilder/3292/consoleFull)**
for PR 14762 at commit
[`ea44a77`](https://github.com/apache/spark/commit/
Github user srowen commented on the issue:
https://github.com/apache/spark/pull/14762
@rxin if you're OK with this I think we could proceed. I like the more
unified code path, but this shows it probably doesn't matter either way. I
personally would also be OK with only following the n
Github user srowen commented on the issue:
https://github.com/apache/spark/pull/14762
OK, it seems like a toss-up for x86. I'd expect that the difference here is
very small in any event, and this at least doesn't contradict that. I'm OK with
this as a result. @rxin your thoughts?
--
Github user jlhitt commented on the issue:
https://github.com/apache/spark/pull/14762
@srowen and @rxin, Sorry for the delay in getting this data to you.. Let me
know if you have any questions.
To see if there were any regressions we ran tests on 2-chip Broadwell E5 v4
10core
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14762
**[Test build #3283 has
finished](https://amplab.cs.berkeley.edu/jenkins/job/NewSparkPullRequestBuilder/3283/consoleFull)**
for PR 14762 at commit
[`2edba67`](https://github.com/apache/spark/commit
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14762
**[Test build #3283 has
started](https://amplab.cs.berkeley.edu/jenkins/job/NewSparkPullRequestBuilder/3283/consoleFull)**
for PR 14762 at commit
[`2edba67`](https://github.com/apache/spark/commit/
Github user srowen commented on the issue:
https://github.com/apache/spark/pull/14762
@jlhitt mentioned looking into the actual performance difference. If it's
worse, it's easy to modify this new path to only affect aligned architectures.
My guess, and this is just based on r
Github user rxin commented on the issue:
https://github.com/apache/spark/pull/14762
"The argument about speed isn't about dead code but that aligned access
ought to be faster even on x86." -- was this validated at all?
---
If your project is set up for it, you can reply to this emai
Github user jlhitt commented on the issue:
https://github.com/apache/spark/pull/14762
BTW, any speed on SPARC is an improvement over not working at all... The
only regression to be worried about is x86.
---
If your project is set up for it, you can reply to this email and have your
Github user jlhitt commented on the issue:
https://github.com/apache/spark/pull/14762
We will test actual queries that that use these areas on both SPARC and
x86, and compare the x86 results to the master branch to verify no regression.
Will get back to you with results ASAP.
---
Github user rxin commented on the issue:
https://github.com/apache/spark/pull/14762
Hey I know you are saying it might not affect performance -- but please
give some evidence. E.g. if you are relying on JIT dead code elimination, show
some evidence that the code gets elided at runtime
Github user sumansomasundar commented on the issue:
https://github.com/apache/spark/pull/14762
@srowen I ran dev/lint-java, removed few additional white spaces, and
shortened few lines longer than 100 characters, then rebased it.
---
If your project is set up for it, you can reply t
Github user srowen commented on the issue:
https://github.com/apache/spark/pull/14762
Ah of course, good catch @jlhitt . Also @sumansomasundar could you run
`dev/lint-java` on your branch now? I think it may fail style checks due to the
minor issue of spacing around a "+" sign. It's n
Github user sumansomasundar commented on the issue:
https://github.com/apache/spark/pull/14762
Made the changes @srowen
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled a
Github user sumansomasundar commented on the issue:
https://github.com/apache/spark/pull/14762
made the changes @srowen
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled a
Github user sumansomasundar commented on the issue:
https://github.com/apache/spark/pull/14762
Made the changes and rebased it. @srowen
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this f
Github user srowen commented on the issue:
https://github.com/apache/spark/pull/14762
This needs a rebase too.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes
Github user sumansomasundar commented on the issue:
https://github.com/apache/spark/pull/14762
I have made the changes you suggested. Please take a look.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project doe
Github user sumansomasundar commented on the issue:
https://github.com/apache/spark/pull/14762
We are testing the performance on x86 systems to see that it does not have
any performance impact.
Do you suggest any specific benchmark that we can test?
---
If your project is set up
Github user srowen commented on the issue:
https://github.com/apache/spark/pull/14762
Ping @sumansomasundar
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so
Github user rxin commented on the issue:
https://github.com/apache/spark/pull/14762
Does your change have any performance impact?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14762
Can one of the admins verify this patch?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feat
26 matches
Mail list logo