[quote]Who do you mean? Obviously not TPR[/quote]
When we joined TPR most of our work was merged into the team's account, so we consider TPR's listing to be fairly accurate.

TPR itself is almost 7 years old.

[QUOTE=jinydu;122519]Who do you mean? Obviously not TPR[/QUOTE]

[url]http://www.teamprimerib.com/rr1/bin/member.php[/url]

I see... One of the perks of being a member of Team Prime Rib is having those 7-day, 30-day and 90-day statistics, even if one is not in Primenet's Top 500.

The webmaster did such a good job, it's a shame he can no longer be contacted.

[QUOTE=jinydu;122547]I see... One of the perks of being a member of Team Prime Rib is having those 7-day, 30-day and 90-day statistics, even if one is not in Primenet's Top 500.[/QUOTE]

There's more to it. If you're a member of a team, you will no longer have an entry in PrimeNet! GIMPS doesn't have the full "Team > Member > Computer" hierarchy -- it has only "Account > Computer" hierarchy (you can flag an account as a team, but that's it) (PS:- this is true for V4. I think V5 addresses this).

So, in order to track TPR member's individual stats, they have a mapping of computer names to member name -- this is done entirely outside PrimeNet/GIMPS.

The latest round of test completions has brought my CPU hrs/day above 4000 for the first time ever!

[CODE]Account ID LL P90* Exponents Fact.P90 Exponents P90 CPU
CPU yrs LL Tested CPU yrs* w/ Factor hrs/day
-------------- ------- --------- -------- --------- -------
jinydu 693.062 274 5.809 4 4041.00[/CODE]

My new comp just finished its first two numbers (dual-core, so it finished first two on same day...no primes, as you might have guessed), bringing my P90 years to 57 and position to 6286. :smile:
The new assignment for my second core is <10,000 from a FFT cutoff, and Prime95 automatically detected that it was within rounding error limits, so I get to use the next-smallest FFT size. :smile:
My first core got assigned a number close to 287,000 from the cutoff, so it'll probably need to use the next bigger size (not sure yet because it's running some P-1 before the LL). :sad:

[quote=Mini-Geek;122735]The new assignment for my second core is <10,000 from a FFT cutoff, and Prime95 automatically detected that it was within rounding error limits, so I get to use the next-smallest FFT size. :smile:
My first core got assigned a number close to 287,000 from the cutoff, so it'll probably need to use the next bigger size (not sure yet because it's running some P-1 before the LL). :sad:[/quote]Can someone explain this for a noob? :question:

[quote=tallguy;122820]Can someone explain this for a noob? :question:[/quote]
How long Prime95 takes for different numbers is mainly related to the CPU running it and how big the number is. Keep the CPU constant, and the only real factor is how big the number is. It's not really just, directly, how big the number is, though. It's how big of an FFT is needed to accurately do the calculations with the number. The size a number is before it requires a larger FFT for accuracy is called (or at least I'm calling it) the cutoff. [URL]http://www.mersenne.org/bench.htm[/URL] lists the cutoff from a 2048K FFT to 2560K FFT at 39.50M (that is, anything above 2^39500000-1). I was assigned a number barely above this (2^39509597-1). Prime95 ran a test and saw that, for my computer and this number, it could still run at 2048K, meaning I'll finish the number much faster.
The other one is apparently too far away from the 2048K-2560K cutoff to run at the smaller one, since my computer automatically is using the larger one.
I hope that explains it. If not, here's the short short version: my computer's running one number faster than I was expecting, and another that barely missed the point of being able to run faster.

[quote=Mini-Geek;122825]How long Prime95 takes for different numbers is mainly related to the CPU running it and how big the number is. Keep the CPU constant, and the only real factor is how big the number is. It's not really just, directly, how big the number is, though. It's how big of an FFT is needed to accurately do the calculations with the number. The size a number is before it requires a larger FFT for accuracy is called (or at least I'm calling it) the cutoff. [URL]http://www.mersenne.org/bench.htm[/URL] lists the cutoff from a 2048K FFT to 2560K FFT at 39.50M (that is, anything above 2^39500000-1). I was assigned a number barely above this (2^39509597-1). Prime95 ran a test and saw that, for my computer and this number, it could still run at 2048K, meaning I'll finish the number much faster.
The other one is apparently too far away from the 2048K-2560K cutoff to run at the smaller one, since my computer automatically is using the larger one.[/quote]How can I see the results of the self-test? I'm not seeing it in results.txt
[quote=Mini-Geek;122825]I hope that explains it. If not, here's the short short version: my computer's running one number faster than I was expecting, and another that barely missed the point of being able to run faster.[/quote]The first explanation worked just fine! :cool:

[quote=tallguy;122832]How can I see the results of the self-test? I'm not seeing it in results.txt[/quote]
It's not the standard self-test that makes sure your computer is stable. It's a special self-test because the number was just outside of the range. It's not common, in fact it's the first I've had to do that in 11 tests (and, statistically speaking, should be far rarer than 1 in 11). Therefore, it's unlikely that you would have it in your results.txt file. Here it is from my results.txt: [code]Trying 1000 iterations for exponent 39509597 using 2048K FFT.
If average roundoff error is above 0.2425, then a larger FFT will be used.
Final average roundoff error is 0.24118, using 2048K FFT for exponent 39509597.[/code]

[quote=Mini-Geek;122835]It's not the standard self-test that makes sure your computer is stable. It's a special self-test because the number was just outside of the range. It's not common, in fact it's the first I've had to do that in 11 tests (and, statistically speaking, should be far rarer than 1 in 11). Therefore, it's unlikely that you would have it in your results.txt file. Here it is from my results.txt: [code]Trying 1000 iterations for exponent 39509597 using 2048K FFT.
If average roundoff error is above 0.2425, then a larger FFT will be used.
Final average roundoff error is 0.24118, using 2048K FFT for exponent 39509597.[/code][/quote]Makes sense... I also saw your other thread on this topic, so I think I've got a decent handle on it now.

Looks like the bottom line is that if one doesn't like which side of the line they fall, the only recourse is to unreserve the exponent and try for something more to your liking! :rolleyes: