Just out of interest: Would that mean my 1.3 GHz Celeron would be in the slowest 20%? Then there must be lots of racehorses in here! ;-) I mean, I know it's not that fast a machine but I thought it might be a bit better, say, somewhere in the middle... maybe slowest 40% or so, but 20?
Just out of interest: Would that mean my 1.3 GHz Celeron would be in the slowest 20%? Then there must be lots of racehorses in here! ;-) I mean, I know it's not that fast a machine but I thought it might be a bit better, say, somewhere in the middle... maybe slowest 40% or so, but 20?
My guess is that it could have been classed a slow computer initially, on its specification, but as you seem to run it 24/7 I think it might end up in the fast computer class.
The average host computer only gets just over 100 credits/day, if you are running Einstein and Seti 50/50 then with the new faster app your total/day should be >150 credits.
That's good to know, thx 4 info. And yes, atm my little comp is running about 24/7... not primarily because of BOINC but of course, it's not bad for my credit at all ;-)
Evesham is a 2.533GHz Foniks is a 3.2GHz, BUT, Evesham has 3 hours per day CPU quota to Einstein, whilst Foniks has only 1, (for the purists out there, roughly - I know it is not accurate).
Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.
The exact algorithm can be seen here. If the host needs a new data file, and the credit_per_cpu_sec < 0.0013, then the host will be given a data file which has short workunits.
A faster host may get either a 'long' or 'short' workunit data file.
My MBP goes back and forfth: some are large WU, some are small, and my PB does the same.
My MBP takes about 113 cpusec for 1 credit with small WU, and 122 cpusec for large WU.
The PB takes about 280 cpusec for 1 credit for both large and small WU.
Over on the SETI side, the MBP takes about 118 cpusec for 1 credit with large WU, but about 163 cpusec for 1 credit with small WU. This is using Alex's latest optimized app. The unoptimized app takes about 180 cpusec per credit for the large WU.
Bruce: thanks for putting that "switch" in. My old iMac 333 MHz takes 59K cpusec to crank out a small WU worth 20.32 credits; it would take most of a week for a large WU.
How did you get these numbers? I could use this for figuring out which projects my new box is most suitable for...
Go to your account, look at the results for the computer of interest. You'll see a column that has the cpu-seconds it took to run the WU, and another column shows you how many credits were earned. Divide cpu-sec by credits to get cpusec per credit. Or, divide credits by cpusec, and get the credit per cpusec that Bruce mentioned.
Thanks a lot. I'm really curious how this will turn out for my new box. As it is an AMD 64, I expect it to be quite successful in Einstein especially... at least, that's what most people say...
And by the way, my laptop is averaging at 0,0037 credits per cpu sec, so my getting only short WUs would be sheer coincidence.
Just out of interest: Would
)
Just out of interest: Would that mean my 1.3 GHz Celeron would be in the slowest 20%? Then there must be lots of racehorses in here! ;-) I mean, I know it's not that fast a machine but I thought it might be a bit better, say, somewhere in the middle... maybe slowest 40% or so, but 20?
RE: Just out of interest:
)
My guess is that it could have been classed a slow computer initially, on its specification, but as you seem to run it 24/7 I think it might end up in the fast computer class.
The average host computer only gets just over 100 credits/day, if you are running Einstein and Seti 50/50 then with the new faster app your total/day should be >150 credits.
Andy
That's good to know, thx 4
)
That's good to know, thx 4 info. And yes, atm my little comp is running about 24/7... not primarily because of BOINC but of course, it's not bad for my credit at all ;-)
My results show that my
)
My results show that my Evesham node gets only long wu's and Foniks gets only short wu's.
Evesham is a 2.533GHz Foniks is a 3.2GHz, BUT, Evesham has 3 hours per day CPU quota to Einstein, whilst Foniks has only 1, (for the purists out there, roughly - I know it is not accurate).
Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.
The exact algorithm can be
)
The exact algorithm can be seen here. If the host needs a new data file, and the credit_per_cpu_sec < 0.0013, then the host will be given a data file which has short workunits.
A faster host may get either a 'long' or 'short' workunit data file.
Cheers,
Bruce
Director, Einstein@Home
Credit per CPU second... now
)
Credit per CPU second... now that doesn't sound to me as if it has to do with how long you have BOINC run?
My MBP goes back and forfth:
)
My MBP goes back and forfth: some are large WU, some are small, and my PB does the same.
My MBP takes about 113 cpusec for 1 credit with small WU, and 122 cpusec for large WU.
The PB takes about 280 cpusec for 1 credit for both large and small WU.
Over on the SETI side, the MBP takes about 118 cpusec for 1 credit with large WU, but about 163 cpusec for 1 credit with small WU. This is using Alex's latest optimized app. The unoptimized app takes about 180 cpusec per credit for the large WU.
Bruce: thanks for putting that "switch" in. My old iMac 333 MHz takes 59K cpusec to crank out a small WU worth 20.32 credits; it would take most of a week for a large WU.
beadman
Join Team MacNN
How did you get these
)
How did you get these numbers? I could use this for figuring out which projects my new box is most suitable for...
RE: How did you get these
)
Go to your account, look at the results for the computer of interest. You'll see a column that has the cpu-seconds it took to run the WU, and another column shows you how many credits were earned. Divide cpu-sec by credits to get cpusec per credit. Or, divide credits by cpusec, and get the credit per cpusec that Bruce mentioned.
beadman
Join Team MacNN
Thanks a lot. I'm really
)
Thanks a lot. I'm really curious how this will turn out for my new box. As it is an AMD 64, I expect it to be quite successful in Einstein especially... at least, that's what most people say...
And by the way, my laptop is averaging at 0,0037 credits per cpu sec, so my getting only short WUs would be sheer coincidence.