![]() |
|
|
#1 |
|
1976 Toyota Corona years forever!
"Wayne"
Nov 2006
Saskatchewan, Canada
467610 Posts |
I have been monitoring the Top Producers reports regularly for the last year or so and formally tracking in since May 4, 2006.....because I love statistics.
Based on P90 CPU Hrs/day the top 3 are clearly: Account: P90 CPU Hrs/Day ======= ============ ahmerali: 332245.7 curtisc: 319294.0 Team_Prime_Rib: 220514.2 For this parameter only, since May "ahmerali" has always been slightly ahead of "curtisc" with "TPR" a respectable third. Everyone else is far behind. I must be misunderstanding something because one would expect that their progress in the last 8 or so months would be relative to this statistic. Alas, it is not: From May 4 to Jan 24 the progress on the LL P90* CPU yrs column has been: Account: LL P90* CPU yrs increase ======= =================== ahmerali: 12226.41 curtisc: 31686.60 ==> Almost 3 times ahmerali Team_Prime_Rib: 15541.10 ==> about 30 more than ahmerali. How can it be that ahmerali consistently has MORE hrs/day yet relatively and significantly LESS total yrs increase in the last 8 months? |
|
|
|
|
|
#2 |
|
Sep 2006
Brussels, Belgium
168610 Posts |
I had been thinking on the same lines, until I remembered that the CPU Hrs per day is computed over the whole life of an account : <total work done since the creation of the account> / <number of days the account exists>. This means that a "recent" account can seem to have a higher speed than an old one. If one compares the work done in the 265 days since 2006-05-04 to 2007-01-24 one can get the speed over that period :
Code:
ahmerali : 12 226.41 * 365 * 24 / 265 = 404 164 P90 CPU Hrs a day curtisc : 31 686.60 * 365 * 24 / 265 = 1 047 451 P90 CPU Hrs a day Team_Prime_Rib : 15 541.10 * 365 * 24 / 265 = 513 736 P90 CPU Hrs a day |
|
|
|
|
|
#3 |
|
1976 Toyota Corona years forever!
"Wayne"
Nov 2006
Saskatchewan, Canada
22·7·167 Posts |
That helps .... I had falsely assumed the CPU Hrs/Day was more instantaneous, for example the number of CPU hrs in the last day.
That also explains why my hrs/day number keeps growing even though I have been running with the same 4 computers for almost a year. I started in Sept 2002 with one 400 Mhz PC. I now have a total of about 10 Ghz so my average too will continue to grow. And it clears up why the number grows so consistently for the big 3. I was thinking: "Gee, are they upgrading one or more computer every day?" And finally, this same computation would suggest that the "ganser" account that shows a drop of 290 hrs per day and 2 placings (6th to 8th) in the same 8 months must have in fact dropped a LOT of computing power (or quit almost completely). In fact since September his P90 years has increased only 17 vs. at least 600 to 16,000 for everyone else on the top 10. He/she has 2 current assignments. Thanks again for enlightening me. |
|
|
|
|
|
#4 |
|
Nov 2003
22·5·373 Posts |
Who TF CARES????
The objective is to find Mersenne primes. What difference does it make how many CPU hours get contributed by each user?? Some users have more machines than others. This whole discussion seems to imply that because someone has more cpu resources that they are somehow more deserving of credit. I think this attitude is ridiculous. One should participate as a labor of love and not because one gets credit for "CPU Hours". |
|
|
|
|
|
#5 |
|
1976 Toyota Corona years forever!
"Wayne"
Nov 2006
Saskatchewan, Canada
111048 Posts |
You misinterpret my intentions.
I am purely a statistician at heart. I like to analyze statistics and in the process seek to understand them. When I can't make sense of them (usually due to my ignorance of the facts) it becomes a challenge for me to find the answer. As for a labour of love ... I have been fascinated by numbers, especially prime numbers for much of my adult life. It has consumed many hours of my life ... hour of fun for me. I have written programs to generate primes, spent months tuning them, looking for better algorithms, looking for trends, etc., etc. Granted I am FAR, FAR behind the GIMPS team in knowledge and ability. I now happily contribute to their cause. |
|
|
|
|
|
#6 | |
|
Mar 2005
17010 Posts |
Quote:
as a sad statto myself I applaud both your efforts, but of course the easy way to view these sort of stats is on the primerib website. There we recently saw the 'exciting' three-way tussle between ekugimps, philnelson and TempleU-CAS at places 9, 10 and 11. Richard |
|
|
|
|
|
|
#7 |
|
"Mike"
Aug 2002
25·257 Posts |
curtisc - Account created: 06 Jul 1998 15:01 UTC
Team_Prime_Rib - Account created: 30 Aug 1997 23:06 UTC We don't know ahmerali's creation date. |
|
|
|
|
|
#8 |
|
Aug 2002
223 Posts |
|
|
|
|
|
|
#9 |
|
"Mike"
Aug 2002
25×257 Posts |
We knew that would bring you out of the woodwork.
|
|
|
|
|
|
#10 |
|
Jul 2005
19 Posts |
My account was created June 03, 2005. Does that change the stats interpretation?
|
|
|
|
|
|
#11 |
|
1976 Toyota Corona years forever!
"Wayne"
Nov 2006
Saskatchewan, Canada
22·7·167 Posts |
I understand the way the stats are computed now ... HOWEVER ... I'm impressed ... you contributed a lot in only 1.5 years ... nearing 3rd place!!!! And I thought I was doing good making the top 1,000 in 4.5 years.
|
|
|
|
![]() |
Similar Threads
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| New GPU Compute System | airsquirrels | GPU Computing | 90 | 2017-12-08 00:13 |
| "fun" contributors thread | Dubslow | PrimeNet | 2 | 2017-02-06 15:12 |
| New Compute Box | Christenson | Hardware | 0 | 2011-01-15 04:44 |
| My throughput does not compute... | petrw1 | Hardware | 9 | 2007-08-13 14:38 |
| Photos of some of our contributors | cheesehead | Lounge | 0 | 2004-12-06 19:56 |