David Shaffer <cdshaffer at acm.org> wrote:
> goran at krampe.se wrote:
> 
> >I am mainly doing maintenance but I have done some profiling/tuning of
> >the code and there are a few places that can be improved quite a bit (I
> >have a few patches lined up) - but it already performs well enough for
> >most people. And I have done some stressing of Kom using "ab" (Apache
> >benchmark) and the numbers are really good.
> 
> I suspect that it is platform dependent.  Under Linux performance "out
> of the box" really stinks.  This is particularly true for file serving

I actually didn't do file serving. :)

> (Squeak's VM-blocking file I/O really gets in the way here) but response
> times are sub-standard even for non-file based code.  I started to work

Well, I thought they looked pretty ok - not sure what sub-standard
means. And I did use Linux as the platform.

> on improving things: http://minnow.cc.gatech.edu/squeak/539 but I have

Yes, I noticed a while back. Whenever you feel you have any changes for
the Kom codebase, let me know.

> since discovered easier ways to do the async I/O (using OSProcess) under
> Linux.  I also feel that Linux could stand to have a "port" of the
> threaded I/O model used in the Windows VM.  There are many significant
> problems with the current AIO framework performance-wise under heavy loads.
> 
> David

regards, Göran