Yes, pretty soon you will not be able to buy a single core anymore. ;)
But what I tried to explain is that the projejt needs additional power and therefor the WUs have to be longer.
BTW, it would be interesting to have a statistic which shows which CPU's does how many work.
Thanks for help
Hans-Peter
Have a look here. This list is ranked by average credit per CPU type.
Notice the Average Credit per CPU (the far right column), Intel's Core architecture leave older computers in the dust.
This growing performance delta is what is causing all of this headache.
There are 10^11 stars in the galaxy. That used to be a huge number. But it's only a hundred billion. It's less than the national deficit! We used to call them astronomical numbers. Now we should call them economical numbers. - Richard Feynman
It appears AMD processors on Vista are hit by this performance issue too. On my X2 4800+ (at 2.4GHZ), it took one core around 90,000 seconds (approx 25 hours) to finish one of these work units. However, an Opteron 175 on Linux, which has a clock difference of being 200mhz slower than my X2, finishes the same WU in 60,000 seconds. That's 30,000 seconds difference!
Here are some recent results with my OWN calulated RAC figures. Most machines are running 24/7 on Win XP.
P940 3.2 dual 22.5H for 284, RAC 604
P4 3.2 +HT 69H for 529, RAC 368
P4 3.0 +HT 24.21H for 174, RAC 345
P4 3.0 +HT 60.5H for 429, RAC 340
P4 1.4 70H for 412, RAC 141
P111 1000 73.4H for 330, RAC 108
AMD Athlon 850 37.9H for 162, RAC 103
The total RAC adds up to ~2000 which is half what it was under Seti, so I agree with the view that optimising would about double the output. For interest the proportions of credit achieved between the CPU's is almost identical to Seti. For all practical purposes it doesn't seem worth running anything less than a P4 3.0Ghz or equivalent on this S5R2 run.
It appears AMD processors on Vista are hit by this performance issue too. On my X2 4800+ (at 2.4GHZ), it took one core around 90,000 seconds (approx 25 hours) to finish one of these work units. However, an Opteron 175 on Linux, which has a clock difference of being 200mhz slower than my X2, finishes the same WU in 60,000 seconds. That's 30,000 seconds difference!
It appears AMD processors on Vista are hit by this performance issue too. On my X2 4800+ (at 2.4GHZ), it took one core around 90,000 seconds (approx 25 hours) to finish one of these work units. However, an Opteron 175 on Linux, which has a clock difference of being 200mhz slower than my X2, finishes the same WU in 60,000 seconds. That's 30,000 seconds difference!
I don't think I want to run a project that has a performance hit this big for Windows.
So, all of the calculated results count for the project. You can find some significant differences between different systems, but don't leave or hate this project because of them. Differences are changing by the time, but never will vanish.
Hm, but the differences are very very big and it shouldn't be like that... If it goes on like this you just feel like wasting very much CPU Time, because the Windows program seems to have bugs or whatever...
I agree there will always be small differences, but this huge difference can't be normal...
Hm, but the differences are very very big and it shouldn't be like that... If it goes on like this you just feel like wasting very much CPU Time, because the Windows program seems to have bugs or whatever...
I agree there will always be small differences, but this huge difference can't be normal...
Hm, but the differences are very very big and it shouldn't be like that... If it goes on like this you just feel like wasting very much CPU Time, because the Windows program seems to have bugs or whatever...
You don't waste the CPU time because each results are important.
So, this is not the reason of your feeling.
Quote:
I agree there will always be small differences, but this huge difference can't be normal...
I think you are only a bit envious.
Would be better to prohibit the linux machines? ;-)
Yep, really strange. Intel seems to be affected, too, in some cases. Akos, would you mind having a look at the "Information about S5" thread? We have some experiments running there about CPU performance, especially on Intels.
Yes, pretty soon you will not
)
Yes, pretty soon you will not be able to buy a single core anymore. ;)
But what I tried to explain is that the projejt needs additional power and therefor the WUs have to be longer.
cu,
Michael
RE: BTW, it would be
)
Have a look here. This list is ranked by average credit per CPU type.
Notice the Average Credit per CPU (the far right column), Intel's Core architecture leave older computers in the dust.
This growing performance delta is what is causing all of this headache.
There are 10^11 stars in the galaxy. That used to be a huge number. But it's only a hundred billion. It's less than the national deficit! We used to call them astronomical numbers. Now we should call them economical numbers. - Richard Feynman
It appears AMD processors on
)
It appears AMD processors on Vista are hit by this performance issue too. On my X2 4800+ (at 2.4GHZ), it took one core around 90,000 seconds (approx 25 hours) to finish one of these work units. However, an Opteron 175 on Linux, which has a clock difference of being 200mhz slower than my X2, finishes the same WU in 60,000 seconds. That's 30,000 seconds difference!
Here's the WU: http://einsteinathome.org/workunit/33706159
I don't think I want to run a project that has a performance hit this big for Windows.
Here are some recent results
)
Here are some recent results with my OWN calulated RAC figures. Most machines are running 24/7 on Win XP.
P940 3.2 dual 22.5H for 284, RAC 604
P4 3.2 +HT 69H for 529, RAC 368
P4 3.0 +HT 24.21H for 174, RAC 345
P4 3.0 +HT 60.5H for 429, RAC 340
P4 1.4 70H for 412, RAC 141
P111 1000 73.4H for 330, RAC 108
AMD Athlon 850 37.9H for 162, RAC 103
The total RAC adds up to ~2000 which is half what it was under Seti, so I agree with the view that optimising would about double the output. For interest the proportions of credit achieved between the CPU's is almost identical to Seti. For all practical purposes it doesn't seem worth running anything less than a P4 3.0Ghz or equivalent on this S5R2 run.
Mrs Miggins - A lady of uncommon refinement
RE: It appears AMD
)
I'm runnig Vista too on an Athlon 64 3500+ at 2,5 GHz and its the same here...
I really hope they will fix this soon, but it doesn't seem so yet ;(
RE: It appears AMD
)
So, all of the calculated results count for the project. You can find some significant differences between different systems, but don't leave or hate this project because of them. Differences are changing by the time, but never will vanish.
Hm, but the differences are
)
Hm, but the differences are very very big and it shouldn't be like that... If it goes on like this you just feel like wasting very much CPU Time, because the Windows program seems to have bugs or whatever...
I agree there will always be small differences, but this huge difference can't be normal...
RE: Hm, but the differences
)
Exactly the point I was making.
RE: Hm, but the differences
)
You don't waste the CPU time because each results are important.
So, this is not the reason of your feeling.
I think you are only a bit envious.
Would be better to prohibit the linux machines? ;-)
edit: Tell me boy... I also dislike them. ;-)
Yep, really strange. Intel
)
Yep, really strange. Intel seems to be affected, too, in some cases. Akos, would you mind having a look at the "Information about S5" thread? We have some experiments running there about CPU performance, especially on Intels.