[Vm-dev] SpurVM performance "regression"?

[Nicolas Cellier]

2016-04-05 19:40 GMT+02:00 Holger Freyther <holger at freyther.de>:

>
>
> > On 03 Apr 2016, at 14:33, Nicolas Cellier <
> nicolas.cellier.aka.nice at gmail.com> wrote:
> >
> > +25% is certainly more than noise and a concern
>
> I agree. but if it is 25% more quick but wrong, I might end up preferring
> the fixed VM. Do you know if there is any other public build between 3427
> and 3602?
>
>
Nope, I don't know.
This is once again really a good job for a bot.
For example, some kind soul did this:
http://build.squeak.org/job/SqueakTrunkPerformance/performance/

But I don't know which VM flavour, target architecture, image version, nor
what condition triggers the performance test...
There's also a lot uncertainty related to load condition of the Jenkins
server...
Ideally those kind of tests should be isolated on a dedicated machine.


> kind regards
>         holger
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.squeakfoundation.org/pipermail/vm-dev/attachments/20160405/fcabfcdb/attachment.htm