mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > PrimeNet

Reply
 
Thread Tools
Old 2015-09-01, 18:21   #1
petrw1
1976 Toyota Corona years forever!
 
petrw1's Avatar
 
"Wayne"
Nov 2006
Saskatchewan, Canada

22·7·167 Posts
Default Interesting but not surprising point-in-time stats

From the last "Recent Results" report:
3000 results in just over 3 hours

Code:
Type    ##      %%      GDays   %%
C	68	2.27%	8537.1	69.02%
F	41	1.37%	58.0	0.47%
F-ECM	1	0.03%	0.7	0.01%
NF	2799	93.30%	3580.7	28.95%
NF-ECM	63	2.10%	108.4	0.88%
NF-PM1	28	0.93%	84.0	0.68%
So even with all the GPUs doing TF work-type C (LL and DC) still produce 2.5X the work.
Of the NF: 44% of the results and 94% of the GDays are from GPUs (surprise!!! NOT)
Of the C: 35% (almost 3,000 GDays) belong to curtisc (no surprise here either)
.....................Equates to 24% of the total output
.................... Equates to over 83% of the thruput of all the GPUs combined.
.....................Pretty amazing considering GPUs produce up to 100X more GDays than CPUs.
petrw1 is offline   Reply With Quote
Old 2015-09-01, 19:09   #2
manfred4
 
manfred4's Avatar
 
Mar 2014
Germany

23·3·5 Posts
Default

For a real breakdown you would need a whole week of results I would imagine, because most GPU workers check in their results only once per day / once per week and the CPU workers report instantly when they finish.
I think the week-stats would have more GPU work than this.
manfred4 is offline   Reply With Quote
Old 2015-09-01, 19:25   #3
petrw1
1976 Toyota Corona years forever!
 
petrw1's Avatar
 
"Wayne"
Nov 2006
Saskatchewan, Canada

124416 Posts
Default

Quote:
Originally Posted by manfred4 View Post
For a real breakdown you would need a whole week of results I would imagine, because most GPU workers check in their results only once per day / once per week and the CPU workers report instantly when they finish.
I think the week-stats would have more GPU work than this.
Agreed...
petrw1 is offline   Reply With Quote
Old 2015-09-01, 19:35   #4
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

260316 Posts
Default

Quote:
Originally Posted by manfred4 View Post
For a real breakdown you would need a whole week of results I would imagine, because most GPU workers check in their results only once per day / once per week...
Not entirely true.

Oliver checks in his work once a week (usually on a Sunday), as did Mike. But most participants of GPU72 check in their results at least once a day thanks to MISFIT and mfloop.

Quote:
Originally Posted by manfred4 View Post
I think the week-stats would have more GPU work than this.
Possibly. But, without question, the larger the temporal domain (within reason) the better the accuracy of the analysis. I tend to go with monthly stats for trending.
chalsall is online now   Reply With Quote
Reply



Similar Threads
Thread Thread Starter Forum Replies Last Post
A bug or... a surprising result???... lycorn PrimeNet 20 2011-06-26 19:52
A surprising result on QS/NFS matrix size jasonp Factoring 3 2008-01-07 20:23
Surprising news headlines dsouza123 Soap Box 2 2007-06-26 21:41
Time Zone of Stats? edorajh NFSNET Discussion 3 2005-09-17 09:34
calculation time not added to account-stats ulikilian Software 1 2004-10-06 23:10

All times are UTC. The time now is 00:08.


Sat Jul 17 00:08:04 UTC 2021 up 49 days, 21:55, 1 user, load averages: 1.50, 1.56, 1.50

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.