Subject: Re: Performance of different processors
From: "m" <no_nomine@hotmail.com>
Date: 12/09/2003, 12:36
Newsgroups: alt.sci.seti

From my SetiQueue stats:

for the 'average' WU in the AR of about 0.45, I get around 5h 25m for a PIII
1.2GHz l/t with 640Megs of pc133. I could get around 8h 33m from a PIII
750MHz l/t with 512Megs of pc100. A (non-hyperthreading 'cos it's 133/533)
2.8GHz P4 with 1Gig of pc2100 DDR weighs in at around 3h 08m. An Athlon
@1.4GHz is clocking in at just over 5h. All with NT client 3.03, running
under SetiDriver.

I'm close to 10,000 units now, although my PIII 750 m/c blew up a  couple of
weeks  back.

HTH

M

"Erez Volach" <ivrit@netvision.net.il> wrote in message
news:3f616223$1@news.012.net.il...
Actually, it makes quite a lost fo sense. It depends on CPU architecture
and
efficiency most of the time, and also MOBO technology (northbridge and
memory).
Your P3 at 800 would supposedly do half time with double speed, at 1600
but,
as there's not P3 running so fast, we look up (or down, in reality) P4,
that
at 2400 (3X times "faster") not even managing twice the performance. a P4
Celery, at 2000 is just somewhat faster, although more than twice the
"speed". OTOH, the Athlon XP (being somewhat faster than an Athlon, which
is
somewhat faster than P3) is managing at 1533 (true "speed" of 1800+) to be
almost linear with the performance of the "old" P3. At that speed (1533)
it
is so much better than celeron (2000) and on par with a 2400 P4. As for
the
discrepancy between the P4 2.4 results - they could be due to motherboard,
memory amount, speed and / or latency settings, OS, other system factors
or
the version used.

"Alex Smariga" <ZAlex07@Smariga.com> wrote in message
news:NdncLKSOjoHH-pn2-665dlUEGz2ic@localhost...
I'm running Seti (3.03 and 3.08) on a range of processors and
operating systems.  I'm trying to understand why there is such a
variation between processors and versions.  I would like feedback as
to others experiences and maybe some recommendations on improving seti
performance.  Most of these machines are dedicated to seti most of the
time, so conflict with other running processes is minimal.

Processor          Seti version       Average time to process
P4 2.8C                        3.03                          3h17m
p4 2.66                         3.08                            3h34m
p3 .80                            3.03
8h30m
Cel 2.0                           3.08                           7h45m
Athlon 1.8+                   3.03                           4h30m
P4 2.4                             3.08
3h10m
P4 2.4                             3.03
4h30m
P4 2.53                           3.03                           4h0m

These machines are all over the place and I'd like some suggestions to
make them more finely tuned, please.

Alex



--