Crunch Time S5 Workunits

Pepperammi
Pepperammi
Joined: 20 Feb 05
Posts: 131
Credit: 437,943
RAC: 0

RE: Hm, are Pentium 4's

Message 38145 in response to message 38143

Quote:
Hm, are Pentium 4's always so bad at Einstein or is yours only totaly messed up? The long WUs take 9:00 - 9:15 hours here with an Athlon 64 3500+ (2200 mhz).
(or do you have HT enabled so it crunches 2 WUs at a time? then think about disabling HT)

The HT one does have HT enabled. Usually means can do little more work but right now its only working on that one einstien unit as its all its got. Other is proper dual core. Can do 64bit but i'm only using standard winXP. Neither have any problem ever in the past and their having no trouble with any other projects either. Seems i have some extremely long WUs.

Even Rafael said his P D 930 at 4.2ghz is taking very long and you can't really get much faster cpu than that.

Maybe your right and theres something wrong with this new app and really doesn't like pentium 4's?

Pepperammi
Pepperammi
Joined: 20 Feb 05
Posts: 131
Credit: 437,943
RAC: 0

RE: As other's have

Message 38146 in response to message 38144

Quote:


As other's have indicated, I would suggest your computer needs sorting out. This unit resultid=34188053 has been processed by a P3 @ 924 MHz in under 24hrs.

Andy

I understand your thinking. Its was my first thought thats somethings wrong but it usually (and still is right now as i write) crunching away happily and faster than some comparable machines on other projects.

[EDIT]
just looked at at the units in my results page and it seems i'm on par with others that have returned results already. Its the units not my machine.

Rafael Cesare Valente "CrackBoy"
Rafael Cesare V...
Joined: 22 Apr 06
Posts: 22
Credit: 122,573
RAC: 0

RE: RE: Hm, are Pentium

Message 38147 in response to message 38145

Quote:
Quote:
Hm, are Pentium 4's always so bad at Einstein or is yours only totaly messed up? The long WUs take 9:00 - 9:15 hours here with an Athlon 64 3500+ (2200 mhz).
(or do you have HT enabled so it crunches 2 WUs at a time? then think about disabling HT)

The HT one does have HT enabled. Usually means can do little more work but right now its only working on that one einstien unit as its all its got. Other is proper dual core. Can do 64bit but i'm only using standard winXP. Neither have any problem ever in the past and their having no trouble with any other projects either. Seems i have some extremely long WUs.

Even Rafael said his P D 930 at 4.2ghz is taking very long and you can't really get much faster cpu than that.

Maybe your right and theres something wrong with this new app and really doesn't like pentium 4's?

Pepperammi, I could also reach 4.5GHz on the DualCore, but it generates alot of heat and the gain on time for S5 WU is about 15 minutes faster than @ 4.2GHz, so, I prefer save my CPU than burn it out right? The cost benefits didn't value... ( sorry for my english ).

I think that the AMDs CPU does better because it can do more instruction per CPU cycle compared to the Intel CPUs.

Again... let's pray for akos fishish his work on the standard app and give us later a optimized one, if 2x faster will be good for everyone.

Best Regards,

Winterknight
Winterknight
Joined: 4 Jun 05
Posts: 482
Credit: 190,207,734
RAC: 113,232

Pepperammi, One question, are

Pepperammi,
One question, are you doing other projects?

I ask because that p3 is a dual and my youngest also has a P4 3.0 and we try to do two projects at same time rather that two units from same project at once. On original Einsteins and Alberts it didn't make much difference but it improves Seti time by about 15%. Now I wonder wether memory bandwidth is having an effect.

Andy

Pepperammi
Pepperammi
Joined: 20 Feb 05
Posts: 131
Credit: 437,943
RAC: 0

RE: Pepperammi, One

Message 38149 in response to message 38148

Quote:

Pepperammi,
One question, are you doing other projects?

I ask because that p3 is a dual and my youngest also has a P4 3.0 and we try to do two projects at same time rather that two units from same project at once. On original Einsteins and Alberts it didn't make much difference but it improves Seti time by about 15%. Now I wonder wether memory bandwidth is having an effect.

Andy

I've heard lot of people say they've seen that. On the older HT system i use to see that sometimes but when i switched to optimized apps on seti and einstien it mostly went away. Next upgrade i did to that system was change to dual channel memory. Did'nt decrease the run times but never saw much increase/decrease in run times when running two units from difrent projects.

This newer system is dual core and has dual channel ddr2 memory. I don't notice any diference in runtimes when running diferent projects at same time. Memory bandwidth shouldn't be a problem. Thinks its more to do with L1 cache and HT trying to use same parts of the cpu at same time when its doing two units from same project.

As i say the HT syatem is currently only running that one einstien unit so should in theory be getting an extra boost from the virtual cpu too.

Have to grin an bare it untill hopefully the situation improves. Akosf (or anyone elses?) more cpu specific optimizations will be much loved.

Winterknight
Winterknight
Joined: 4 Jun 05
Posts: 482
Credit: 190,207,734
RAC: 113,232

RE: RE: Pepperammi, One

Message 38150 in response to message 38149

Quote:
Quote:

Pepperammi,
One question, are you doing other projects?

I ask because that p3 is a dual and my youngest also has a P4 3.0 and we try to do two projects at same time rather that two units from same project at once. On original Einsteins and Alberts it didn't make much difference but it improves Seti time by about 15%. Now I wonder wether memory bandwidth is having an effect.

Andy

I've heard lot of people say they've seen that. On the older HT system i use to see that sometimes but when i switched to optimized apps on seti and einstien it mostly went away. Next upgrade i did to that system was change to dual channel memory. Did'nt decrease the run times but never saw much increase/decrease in run times when running two units from difrent projects.

This newer system is dual core and has dual channel ddr2 memory. I don't notice any diference in runtimes when running diferent projects at same time. Memory bandwidth shouldn't be a problem. Thinks its more to do with L1 cache and HT trying to use same parts of the cpu at same time when its doing two units from same project.

As i say the HT syatem is currently only running that one einstien unit so should in theory be getting an extra boost from the virtual cpu too.

Have to grin an bare it untill hopefully the situation improves. Akosf (or anyone elses?) more cpu specific optimizations will be much loved.

Seems a bit ruff that a cpu with clock over 3 * my P3 is less than twice as fast.
Still waiting for an S5 unit on my Pent M.

Pepperammi
Pepperammi
Joined: 20 Feb 05
Posts: 131
Credit: 437,943
RAC: 0

RE: Seems a bit ruff that

Message 38151 in response to message 38150

Quote:

Seems a bit ruff that a cpu with clock over 3 * my P3 is less than twice as fast.
Still waiting for an S5 unit on my Pent M.

Yes >:-[

That was a different unit tho. Maybe that unit was a nice easy one :) Can't tell till other machines come back on that one.

Rafael Cesare Valente "CrackBoy"
Rafael Cesare V...
Joined: 22 Apr 06
Posts: 22
Credit: 122,573
RAC: 0

RE: RE: RE: Pepperammi,

Message 38152 in response to message 38150

Quote:
Quote:
Quote:

Pepperammi,
One question, are you doing other projects?

I ask because that p3 is a dual and my youngest also has a P4 3.0 and we try to do two projects at same time rather that two units from same project at once. On original Einsteins and Alberts it didn't make much difference but it improves Seti time by about 15%. Now I wonder wether memory bandwidth is having an effect.

Andy

I've heard lot of people say they've seen that. On the older HT system i use to see that sometimes but when i switched to optimized apps on seti and einstien it mostly went away. Next upgrade i did to that system was change to dual channel memory. Did'nt decrease the run times but never saw much increase/decrease in run times when running two units from difrent projects.

This newer system is dual core and has dual channel ddr2 memory. I don't notice any diference in runtimes when running diferent projects at same time. Memory bandwidth shouldn't be a problem. Thinks its more to do with L1 cache and HT trying to use same parts of the cpu at same time when its doing two units from same project.

As i say the HT syatem is currently only running that one einstien unit so should in theory be getting an extra boost from the virtual cpu too.

Have to grin an bare it untill hopefully the situation improves. Akosf (or anyone elses?) more cpu specific optimizations will be much loved.

Seems a bit ruff that a cpu with clock over 3 * my P3 is less than twice as fast.
Still waiting for an S5 unit on my Pent M.

FYI, my Centrino Pentium M 1.6 GHz ( Dell Latitude D600 ) is doing 1WU in 13.55 Hours.

Best Regards,

pindakoe
pindakoe
Joined: 5 Mar 05
Posts: 2
Credit: 58,056
RAC: 0

Just over 50% through my

Just over 50% through my first S5 WU, but getting there took well over 20 hours... This is on a P3-866 MHz, so dated hardware. However, Akosf optimising app took 3-5 hours, so I am looking at a maybe 8 times longer execution times for a 'long' WU (unless the second half processes much faster).

somebodley
somebodley
Joined: 3 Mar 06
Posts: 119
Credit: 180,339
RAC: 0

My G5 Mac 2.0Gig crunches a

My G5 Mac 2.0Gig crunches a long WU in 10 hours. As it is a 'Dual', it therefore turns out a WU every 5 hours. (It can be argued!)
So far, so good

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.