mersenneforum.org  

Go Back   mersenneforum.org > New To GIMPS? Start Here! > Information & Answers

Reply
 
Thread Tools
Old 2009-03-07, 07:45   #1
lidocorc
 
lidocorc's Avatar
 
Nov 2008
Rosenheim, Germany

23·3 Posts
Default Reliability and confidence level

In GIMPS's assignment rules two terms are used: "confidence level" and "reliability". How are they defined and measured? Can I find the values of my CPUs somewhere?
lidocorc is offline   Reply With Quote
Old 2009-03-07, 13:27   #2
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

3·2,281 Posts
Default

Quote:
Originally Posted by lidocorc View Post
In GIMPS's assignment rules two terms are used: "confidence level" and "reliability". How are they defined and measured?
When you submit an LL result the reliability goes up if the run had no errors, it goes down if the LL run had errors. The confidence level is simply the number of LL tests you've reported (i.e. how confident the server is in the generated reliability score)

Last fiddled with by garo on 2009-08-18 at 12:14 Reason: confidence->reliability
Prime95 is offline   Reply With Quote
Old 2009-03-07, 14:04   #3
lidocorc
 
lidocorc's Avatar
 
Nov 2008
Rosenheim, Germany

1816 Posts
Default

Thank you for responding.

Your explanation of "confidence level" is according to what I see in the statistics Prime Net is recording. But what is it about "reliability"? Many of our machines got 1.0. There are others which got 0.98. I didn't notice any error reports by these machines (in prime.log or results.txt). Where does the difference come from?
lidocorc is offline   Reply With Quote
Old 2009-03-07, 17:47   #4
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

3·2,281 Posts
Default

Actually, reliability is a rolling average. You get 1.0 for a successful double-check. You get a 0.0 for a proven bad result (should only happen if someone verified the exponent before you submitted your result). You get 0.5 (I think) for an error plagued unverified test. You get 0.98 for a unverified LL test.
Prime95 is offline   Reply With Quote
Old 2009-03-08, 06:51   #5
xorbe
 
xorbe's Avatar
 
Feb 2009

7310 Posts
Default

Quote:
Originally Posted by Prime95 View Post
Actually, reliability is a rolling average. You get 1.0 for a successful double-check. You get a 0.0 for a proven bad result (should only happen if someone verified the exponent before you submitted your result). You get 0.5 (I think) for an error plagued unverified test. You get 0.98 for a unverified LL test.
What happens if an LL-D later disagrees with the output of a previous LL? Are both machines placed into question until a second LL-D can determine who was wrong and who was right?
xorbe is offline   Reply With Quote
Old 2009-03-08, 13:00   #6
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

3·2,281 Posts
Default

Quote:
Originally Posted by xorbe View Post
What happens if an LL-D later disagrees with the output of a previous LL? Are both machines placed into question until a second LL-D can determine who was wrong and who was right?
No changes are made to either machine. I only look at the error code returned with an LL test.
Prime95 is offline   Reply With Quote
Old 2009-08-11, 04:04   #7
petrw1
1976 Toyota Corona years forever!
 
petrw1's Avatar
 
"Wayne"
Nov 2006
Saskatchewan, Canada

4,289 Posts
Default

Quote:
Originally Posted by Prime95 View Post
Actually, reliability is a rolling average. You get 1.0 for a successful double-check. You get a 0.0 for a proven bad result (should only happen if someone verified the exponent before you submitted your result). You get 0.5 (I think) for an error plagued unverified test. You get 0.98 for a unverified LL test.
My one CPU has a Reliabilty of 0.81 and it has LL results of:
3 under "Verified LL Results"
- 1 was a Verified DC
- 1 had 2 rounding errors that it recovered from and ended up being a Verified LL
- the third had no errors and was verified

1 under "Unverified LL Results" --- error code is blank

0 under "Bad test results"

So I think there should be a rolling average of:
0.98 for the unverified LL
1.00 for the successful DC
.........
and if I get nothing for an unverified LL
and 0.45 (rather than 0.5) for the verified LL with 2 rounding errors
===============
Then it comes out to the 0.81 I have.

It this possibly correct?

Thanks
petrw1 is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Reliability and Confidence of CPUs? AlexJohn73 Information & Answers 4 2015-11-22 09:26
what are Reliability and Confidence? dragonbud20 Information & Answers 10 2015-10-21 03:26
Reliability, Confidence and exponent assignments tha PrimeNet 36 2013-12-06 09:23
Overclocking and reliability lidocorc Hardware 8 2009-03-24 12:38
NewPGen reliability Cruelty Riesel Prime Search 3 2006-02-15 05:15

All times are UTC. The time now is 08:13.

Thu Jun 4 08:13:13 UTC 2020 up 71 days, 5:46, 0 users, load averages: 1.33, 1.47, 1.41

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, 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.