By the way, Michael, just what do you have crunching there??? I'd love to know what kind of hotrod you're running with...
OK, Warhawk,
ASUS A7N8X deluxe mobo
Athlon XP-M 2600 Barton-core CPU (M is for Mobile)
2x256MB Corsair XMS DDR400 memory
ATI-brand 9600Pro vidcard
Enermax 350W PSU (weak link, I think)
2x80GB WD IDE hdd
250GB Seagate SATA hdd
Thermaltake Volcano 12 HSF
Sorry, I'll have to fill in the settings after the pain pill kicks in, back in 30-40 minutes.
Michael
I've got several AMDs running in my farm... Comparing apples to apples (as best as one can in this case) they seem to kick P4 butt... Am I right or is this a case of not being able to match up work unit results properly??
It's easier to beg for forgiveness that it is to ask for permission...
By the way, Michael, just what do you have crunching there??? I'd love to know what kind of hotrod you're running with...
OK, Warhawk,
WinXP sp2
ASUS A7N8X deluxe mobo
Athlon XP-M 2600 Barton-core CPU (M is for Mobile)
2x256MB Corsair XMS DDR400 memory
ATI-brand 9600Pro vidcard
Enermax 350W PSU (weak link, I think)
2x80GB WD IDE hdd
250GB Seagate SATA hdd
Thermaltake Volcano 12 HSF
Sorry, I'll have to fill in the settings after the pain pill kicks in, back in 30-40 minutes.
edit:
CPU running @ 2.544GHz (according to CPU-Z), 203FSB x 12.5
memory timings 2-3-2-6, 2.5T
Vcore 1.675, Vmemory 2.7
intake fans: front 12cm, side door 8cm
exhaust fans: rear panel 2x12cm, top "blowhole" 8cm
CPU temp (according to ASUS Probe, while crunching) 43C = 109F All WinXP eyecandy and unnecessary services disabled.
Comparing WU times for alberts is useless - their sizes are all over the map.
Comparing Einsteins, however, is valid, though you'll have to go to the last page of my result list to find the most recent. Range of Einstein WU times over the past several months was from 5hrs to 5 hrs 5 mins (18000-18300 seconds). In Novenber, before we ran the furnace, the room temp was ~ 55F, and I was able to raise FSB to 206, for a clock of 2.585, and got off a 2-day string of under-5hr times, some in the 17700sec range, AMD FX-57 territory. :-)
My RAC dropped quite a bit during and after my recent (Dec 22 - Jan 5) hospital stay, because 7 days into it, there was a power outage here. Since my return, powering back up to find albert-only download availability, I got stuck with some low-claiming and slow-reporting co-crunchers, and RAC did not pick up to the 350 area I'd enjoyed pre-hospital. Two days ago, I decided to temporarily give Seti a shot, at 50% share, and the rig is crunching Seti at about 1hr 7min for a regular full WU, so it's receiving 20+ credits/crunch hour there, if I extrapolate the trend of completed quorum results through the remainder.
Not to be an AMD fanboy, but Intel architecture, especially P4, is at a disadvantage, at least on this project, partly due to no SSE (of any variety) coding in the app. I haven't seen any single-proc Intels crunch Einsteins in the 5 hour range.
Michael
microcraft
"The arc of history is long, but it bends toward justice" - MLK
I've got several AMDs running in my farm... Comparing apples to apples (as best as one can in this case) they seem to kick P4 butt... Am I right or is this a case of not being able to match up work unit results properly??
The AMD "number" of relative performance seems to be pretty reliable. The difficulty you get into is comparing apples to apples. If you match a P4 3.2 GHz to a AMD 3200 it will appear that the AMD is always beating the P4, and it is on processing time. But, things are more nearly equal when you consider that the P4 is doing 2 things at the same time. And so, THROUGHPUT wise, the P4 will be equal or better.
That being said, with HT going by the way-side, we should be able to more easily compare apples to apples in the future when we are using "pure" single threaded cores to single threaded cores.
At the moment, my guess would be that the AMD's are a little ahead of the P4 in these cases ... at least for the moment. Speed/power product I would also guess to be in AMDs favor. Cost too ...
My single threaded AMD 3500 is proportionally faster than the 2.8 GHz P4 that I have that is single threaded, but not over what I would expect.
Einstein@Home member since 20 Dec 2005
Total credit 125,695.46
Recent average credit 5,000.55 <=======
At least it is for me!!!
Remarkable!!! Did you build any of those hosts? Interested?
Yeah... I built both Xeon dual processor servers, five AMDs and a couple of the P4s that are currently running... Most of the boxes are factory P4s, however...
It's easier to beg for forgiveness that it is to ask for permission...
RE: RE: By the way,
)
I've got several AMDs running in my farm... Comparing apples to apples (as best as one can in this case) they seem to kick P4 butt... Am I right or is this a case of not being able to match up work unit results properly??
It's easier to beg for forgiveness that it is to ask for permission...
-AFFTC
RE: RE: By the way,
)
edit:
CPU running @ 2.544GHz (according to CPU-Z), 203FSB x 12.5
memory timings 2-3-2-6, 2.5T
Vcore 1.675, Vmemory 2.7
intake fans: front 12cm, side door 8cm
exhaust fans: rear panel 2x12cm, top "blowhole" 8cm
CPU temp (according to ASUS Probe, while crunching) 43C = 109F
All WinXP eyecandy and unnecessary services disabled.
Comparing WU times for alberts is useless - their sizes are all over the map.
Comparing Einsteins, however, is valid, though you'll have to go to the last page of my result list to find the most recent. Range of Einstein WU times over the past several months was from 5hrs to 5 hrs 5 mins (18000-18300 seconds). In Novenber, before we ran the furnace, the room temp was ~ 55F, and I was able to raise FSB to 206, for a clock of 2.585, and got off a 2-day string of under-5hr times, some in the 17700sec range, AMD FX-57 territory. :-)
My RAC dropped quite a bit during and after my recent (Dec 22 - Jan 5) hospital stay, because 7 days into it, there was a power outage here. Since my return, powering back up to find albert-only download availability, I got stuck with some low-claiming and slow-reporting co-crunchers, and RAC did not pick up to the 350 area I'd enjoyed pre-hospital. Two days ago, I decided to temporarily give Seti a shot, at 50% share, and the rig is crunching Seti at about 1hr 7min for a regular full WU, so it's receiving 20+ credits/crunch hour there, if I extrapolate the trend of completed quorum results through the remainder.
Not to be an AMD fanboy, but Intel architecture, especially P4, is at a disadvantage, at least on this project, partly due to no SSE (of any variety) coding in the app. I haven't seen any single-proc Intels crunch Einsteins in the 5 hour range.
Michael
microcraft
"The arc of history is long, but it bends toward justice" - MLK
RE: I've got several AMDs
)
The AMD "number" of relative performance seems to be pretty reliable. The difficulty you get into is comparing apples to apples. If you match a P4 3.2 GHz to a AMD 3200 it will appear that the AMD is always beating the P4, and it is on processing time. But, things are more nearly equal when you consider that the P4 is doing 2 things at the same time. And so, THROUGHPUT wise, the P4 will be equal or better.
That being said, with HT going by the way-side, we should be able to more easily compare apples to apples in the future when we are using "pure" single threaded cores to single threaded cores.
At the moment, my guess would be that the AMD's are a little ahead of the P4 in these cases ... at least for the moment. Speed/power product I would also guess to be in AMDs favor. Cost too ...
My single threaded AMD 3500 is proportionally faster than the 2.8 GHz P4 that I have that is single threaded, but not over what I would expect.
OK... this must be some kind
)
OK... this must be some kind of milestone...
Einstein@Home member since 20 Dec 2005
Total credit 125,695.46
Recent average credit 5,000.55 <=======
At least it is for me!!!
It's easier to beg for forgiveness that it is to ask for permission...
-AFFTC
RE: OK... this must be some
)
Remarkable!!! Did you build any of those hosts? Interested?
microcraft
"The arc of history is long, but it bends toward justice" - MLK
RE: RE: OK... this must
)
Yeah... I built both Xeon dual processor servers, five AMDs and a couple of the P4s that are currently running... Most of the boxes are factory P4s, however...
It's easier to beg for forgiveness that it is to ask for permission...
-AFFTC
7K in E@H!!! 4th machine
)
7K in E@H!!!
4th machine added to my E@H fleet (with 2 CPUs)!
RE: 7K in E@H!!! 4th
)
Way to go, Edo!! That new doolie will be moving you up in the standings FAST!
It's easier to beg for forgiveness that it is to ask for permission...
-AFFTC
100K BOINC :) Can't keep
)
100K BOINC :)
Can't keep up with Warhawk....
Took me since SETI BOINC went live.
Michael
Team Linux Users Everywhere
RE: 100K BOINC
)
Congratulations!
Who can?