On Fri, Jul 17, 2009 at 5:31 PM, Stéphane Ducasse <[email protected]
> wrote:
> Hi
>
> David shaffer reported to me in fuzzy way that pharo is slow on linux.
> Dear linux users can you report if you experienced the same and
> on which
> machine
> distributions
> vm
> image
>
> please compare the core and the dev (with browser usage)
>
Ok, let me explain MY experience.
Most of my Pharo work is in Linux (at home). In windows I do little things
because it is my work pc.
My data:
- Dual core 2 1.6 GZ, 2GB RAM
- Ubuntu 8.10
- VM: Exupery: pharo-vm-0.15.1b-linux.zip
- Images: latest
Now my comments:
1) Mac hardware is much better than normal pcs
2) Mac VM seems faster than others. I don't know if it is because 1) or any
other reason.
3) I did find Pharo still slow. I don't blame to it. I think they are doing
the good setps:
a) indentify the slow stuff
b) try to disable them from Pharo image. This point is REALLY important
for me because all of the nice work you are doing won't be see by others.
They open an image and if it takes 3 seconds to open a browser, they will
close it and never use it again. And what about our 5000 green test? MIT?
closure? To trash...
c) look at the slow stuff and try to improve it
4) I really don't see a big difference beetween Linux and Windows
performance in my case. Both are more or less the same situation.
5) I did see difference BIG difference with this two things:
a) compare core to dev
b) compare package browser with the simpler one
c) compare seaside one click to pharo. I think this is because of b)
Because of this, in my pharo images I evaluate this when I download a new
image:
ToolSet default: StandardToolSet.
SystemBrowser default: OBSystemBrowserAdaptor.
Preferences enable: #fastDragWindowForMorphic.
Preferences disable: #dynamicProtocolActivation.
With this, my image is usable enough. But not fast.
6) If you have the image in a SUN or over the network, that's VERY slow.
7) In windows, take care of antivirus like Nod32 because they won't let you
use Pharo (mostly when you are snapshooting).
My proposal: There were some benchmarks trough different threads. So, what
about recollecting all the benchmark we can (even create more if we can) and
put them together in a script. We can do like a Poll and send an email to
the mailing list. One we have several results, we can extract information
from there.
>From each user we must know which OS and version, which VM and version.
The benchmarks can also be run with different configuration (using for
example what I told above) and over different images (like core and pharo).
Perhaps having all this information we can extract something useful.
Best,
Mariano
>
> _______________________________________________
> Pharo-project mailing list
> [email protected]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>
_______________________________________________
Pharo-project mailing list
[email protected]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
