mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Joerg Schad" <jo...@mesosphere.io>
Subject Re: Review Request 38158: Refactored Value::Ranges coalesce().
Date Sat, 26 Sep 2015 18:45:13 GMT


> On Sept. 23, 2015, 6:29 p.m., Ben Mahler wrote:
> > Where is the benchmark for this change? :(
> 
> Joerg Schad wrote:
>     We benchmarked the code with a proprietary benchmark. I added https://issues.apache.org/jira/browse/MESOS-3502
aiming at adding an open-source benchmark for this.
> 
> Alexander Rukletsov wrote:
>     Ben, I'm with you regarding benchmarking this change. But we thougt about this patch
as a hotfix improvement, which makes the behavior not worse than the status quo. Basically,
Jörg removed extra memory allocations and cleaned up deletions from a linear array.
>     
>     I think we still need a deeper refactoring here (similar to `state.json`), with caching
intermediate results or other ideas that can drastically speed up the performance. This will
be a bigger project, which will definitely require a thorough benchmark.

FYI The benchmark has been published via MESOS-3493.


- Joerg


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/38158/#review100261
-----------------------------------------------------------


On Sept. 23, 2015, 12:37 p.m., Joerg Schad wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/38158/
> -----------------------------------------------------------
> 
> (Updated Sept. 23, 2015, 12:37 p.m.)
> 
> 
> Review request for mesos, Bernd Mathiske, Joris Van Remoortere, and Till Toenshoff.
> 
> 
> Bugs: MESOS-3051
>     https://issues.apache.org/jira/browse/MESOS-3051
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> The goal of this refactoring was to reuse the Ranges objects as much as possible, as
prior there was substantial time spend in allocation/destruction (MESOS-3051).
> 
> 
> Diffs
> -----
> 
>   src/common/values.cpp 750264e603b4cde2011f07f4434a4b34fe3e512f 
>   src/tests/resources_tests.cpp 0318885336409f7cc9dbd4a3daa9b52db197bbd1 
>   src/tests/values_tests.cpp fc35d97894a2de6207b9337180e2160e6f2cb1f5 
> 
> Diff: https://reviews.apache.org/r/38158/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Joerg Schad
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message