* [parisc-linux] performance?
@ 2002-03-26 11:43 Peter Lavender
2002-03-26 15:44 ` Matthew Wilcox
0 siblings, 1 reply; 6+ messages in thread
From: Peter Lavender @ 2002-03-26 11:43 UTC (permalink / raw)
To: parisc-linux
Hi Everyone,
I know I have asked this sort of thing before, but I'm doing c++ this
semester and as such I have been using the hppa box (715/64) wiht 128
meg ram as my "dev" box.
Compiling and running of a program that finds perfect numbers upto
10000 is somewhat slow.. but that's ok..
It wasn't until I compiled it on my p90 with 48 meg ram running a a
zillion services and daemons that I noticed just how slow the 715 is.
Running the program proved this even further.
I thought the 715/64 was about similar in performance.. but this
simple observation shows this not to be the case.
As a webserver the 715 is fine... but I'm now thinking that when these
exercises and assignments start getting bigger that the 715 will be
just a little too slow for my liking. This isn't a bad thing.. just
an observation...
Regards,
Pete.
--
Due to a lack of imagination,
this signature will remain
under construction indefinately.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [parisc-linux] performance?
2002-03-26 11:43 [parisc-linux] performance? Peter Lavender
@ 2002-03-26 15:44 ` Matthew Wilcox
2002-03-27 7:06 ` Peter Lavender
0 siblings, 1 reply; 6+ messages in thread
From: Matthew Wilcox @ 2002-03-26 15:44 UTC (permalink / raw)
To: Peter Lavender; +Cc: parisc-linux
On Tue, Mar 26, 2002 at 10:43:58PM +1100, Peter Lavender wrote:
> It wasn't until I compiled it on my p90 with 48 meg ram running a a
> zillion services and daemons that I noticed just how slow the 715 is.
> Running the program proved this even further.
Would you mind posting some numbers?
Also, let's try to figure out what other variables are a factor. Can you
compare gcc-3.0 on your x86 to the current compiler?
--
Revolutions do not require corporate support.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [parisc-linux] performance?
2002-03-26 15:44 ` Matthew Wilcox
@ 2002-03-27 7:06 ` Peter Lavender
2002-03-27 12:00 ` Rafael E. Herrera
2002-03-28 5:06 ` Carlos O'Donell Jr.
0 siblings, 2 replies; 6+ messages in thread
From: Peter Lavender @ 2002-03-27 7:06 UTC (permalink / raw)
To: Matthew Wilcox; +Cc: Peter Lavender, parisc-linux
* Matthew Wilcox (willy@debian.org) wrote:
> > It wasn't until I compiled it on my p90 with 48 meg ram running a a
> > zillion services and daemons that I noticed just how slow the 715 is.
> > Running the program proved this even further.
>
> Would you mind posting some numbers?
>
> Also, let's try to figure out what other variables are a factor. Can you
> compare gcc-3.0 on your x86 to the current compiler?
The p90 system is an old stock install of redhat 6.0 with very little
done to the kernel.. it was recompiled with certain drivers built into
the kernel but that's about it.
[pete@piglet brad]$ g++ --version
egcs-2.91.66
The program is fairly simple in what it does.
I really don't want to install gcc-3.0 on this system as it really is
a stable workhorse and I'd rather not break.
But even running time is somewhat of a difference... lets see:
This is a very simple shell script that outputs date at start and
finish of the program running.. I'm no guru really.. :)
Run on HPPA 715/64
Wed Mar 27 18:02:20 EST 2002
Wed Mar 27 18:02:50 EST 2002
Run on P90
Wed Mar 27 18:00:32 EST 2002
Wed Mar 27 18:00:39 EST 2002
Just quickly you can see that it took 7 secs to do the same thing on
the p90 and 30 seconds on the hppa box.
The p90 runs alot of daemons, substantially more than the hppa box..
but this is a surprise.. I have some thoughts about the compiling, but
none what so ever about the general difference between the arches...
I just thought it was of interest as I thought the hppa box was of
similar processing capability to the p90.
Regards,
Pete.
--
Due to a lack of imagination,
this signature will remain
under construction indefinately.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [parisc-linux] performance?
2002-03-27 7:06 ` Peter Lavender
@ 2002-03-27 12:00 ` Rafael E. Herrera
2002-03-28 5:06 ` Carlos O'Donell Jr.
1 sibling, 0 replies; 6+ messages in thread
From: Rafael E. Herrera @ 2002-03-27 12:00 UTC (permalink / raw)
To: parisc-linux
FYI, on a 715/75 I run SETI@home and was able to process 4 units at an
average rate of ~99 hours/unit. Currently, I'm using a 715/80, but it
doesn't look to be doing much better.
--
Rafael
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [parisc-linux] performance?
2002-03-27 7:06 ` Peter Lavender
2002-03-27 12:00 ` Rafael E. Herrera
@ 2002-03-28 5:06 ` Carlos O'Donell Jr.
2002-03-28 6:10 ` Peter Lavender
1 sibling, 1 reply; 6+ messages in thread
From: Carlos O'Donell Jr. @ 2002-03-28 5:06 UTC (permalink / raw)
To: Peter Lavender; +Cc: parisc-linux
>
> The p90 system is an old stock install of redhat 6.0 with very little
> done to the kernel.. it was recompiled with certain drivers built into
> the kernel but that's about it.
>
> [pete@piglet brad]$ g++ --version
> egcs-2.91.66
>
> The program is fairly simple in what it does.
>
This is the crux of the problem. IA32 glibc and kernel are sprinkled
with copious amounts of hand tuned and optimized assembly. We are
nowhere near that level yet :)
What is it that your program does?
Like the beowulf.org people say:
"Your application is always the best benchmark."
Cheers,
Carlos.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [parisc-linux] performance?
2002-03-28 5:06 ` Carlos O'Donell Jr.
@ 2002-03-28 6:10 ` Peter Lavender
0 siblings, 0 replies; 6+ messages in thread
From: Peter Lavender @ 2002-03-28 6:10 UTC (permalink / raw)
To: Carlos O'Donell Jr., Peter Lavender, parisc-linux
* Carlos O'Donell Jr. (carlos@baldric.uwo.ca) wrote:
> > The p90 system is an old stock install of redhat 6.0 with very little
> > done to the kernel.. it was recompiled with certain drivers built into
> > the kernel but that's about it.
> >
> > [pete@piglet brad]$ g++ --version
> > egcs-2.91.66
> >
> > The program is fairly simple in what it does.
> >
>
> This is the crux of the problem. IA32 glibc and kernel are sprinkled
> with copious amounts of hand tuned and optimized assembly. We are
> nowhere near that level yet :)
I thought that the i386 may hvae some performance tweaks in it, since
this is by far the most used platform for development.
> What is it that your program does?
Nothing special, it's just an exercise in programming. It searches
for perfect numbers and then prints out the prime number in the list
of factors.
It's not very big, so if you want to take a look, I'll email it to
you.. :)
Regards,
Pete.
--
Due to a lack of imagination,
this signature will remain
under construction indefinately.
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2002-03-28 6:11 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2002-03-26 11:43 [parisc-linux] performance? Peter Lavender
2002-03-26 15:44 ` Matthew Wilcox
2002-03-27 7:06 ` Peter Lavender
2002-03-27 12:00 ` Rafael E. Herrera
2002-03-28 5:06 ` Carlos O'Donell Jr.
2002-03-28 6:10 ` Peter Lavender
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox