[QUOTE=stebbo;443567]Not sure if this is the right area for this, but one of my new machines has just returned a result of C - Mismatch. (exponent was 42576029). I'm assuming this means that I've returned a different residue than the previous LL assignment.

I'm curious as to what happens now.
[/QUOTE]

It needs to be checked again by somebody else to find out which result is correct. I've assigned it. You may look at your result page in a few days.

[QUOTE=fivemack;443568]May I get someone to triple-check 38849663?[/QUOTE]

Got it.

[QUOTE=rudi_m;443571]It needs to be checked again by somebody else to find out which result is correct. I've assigned it. You may look at your result page in a few days.



Got it.[/QUOTE]

Madpoo just ran a query for me. We're down to ~ 170 ThzDays of unassigned triple checks. If we keep the pedal to the floor for the next ~60 days to get caught up on the backlog we should be able to maintain same-week turn around service on all mismatches going forward. Just another 10+ years and we can catchup and offer same-month double checks :bangheadonwall:

[QUOTE=airsquirrels;443602]Just another 10+ years and we can catchup and offer same-month double checks :bangheadonwall:[/QUOTE]

Honestly I think the change that all new contributors get DCs by default has shifted the balance of power enough that eventually the DC wave will catch up.

[QUOTE=airsquirrels;443602] Just another 10+ years and we can catchup and offer same-month double checks[/QUOTE]

Maybe same-month double checks are not ideal. Wouldn't it be better do double check with different software version on different hardware?

[QUOTE=rudi_m;443617]Maybe same-month double checks are not ideal. [/QUOTE]

Why?

[QUOTE=rudi_m;443617]Wouldn't it be better do double check with different software version on different hardware?[/QUOTE]

Possibly, but not necessarily.

The existing software has proven itself time and time again to be sane, and to sometimes prove that the hardware supporting the software isn't.

[QUOTE=chalsall;443618]Why?



Possibly, but not necessarily.

The existing software has proven itself time and time again to be sane, and to sometimes prove that the hardware supporting the software isn't.[/QUOTE]

Of course, but in theory it's possible that mprime only fails for one particular exponent. The risk to slip such bug would be a bit lower if we double check years later using a different version or even better a completely different implementation.

[QUOTE=rudi_m;443617]Maybe same-month double checks are not ideal. Wouldn't it be better do double check with different software version on different hardware?[/QUOTE]
[QUOTE=chalsall;443618]Why?[/QUOTE]Allowing for about a year or even 2 between the front ends of LL and DC is, IMHO a fine thing. It allows for faster turnarounds on the slower (relative) machines that do the DC. It allows silicon based, electrically powered space heaters to contribute to the project in a useful manner. This is especially true now that GPU's are crushing the TF realm.

[QUOTE=rudi_m;443617]Maybe same-month double checks are not ideal.[/QUOTE]
We agree with this (sorry Chris :razz:)
Year-range DCs are very good in keeping old computers in business. We should NOT struggle to do LL and DC in the same day/week/month. Let some breadcrumbs to the little birds too, otherwise they are out of business. Keep the milestone advancing, i.e. [URL="https://en.wikipedia.org/wiki/Sandworm_(Dune)#Riding"]hammer the tail of Shai Hulud[/URL], but don't pull its tail over its head. That would be very bad for you...

I was checking on self-verified work (a few have built up).

The user who has some machines that spit out a lot of bad results has, for some reason, done a few self-verified things... maybe to make sure his results matched, but why he did exponents in the 100M size (not 100M digits, but I mean like this one) to test, I have no idea.

Anyway, I'm glad I took the 3 100M exponents he's self verified and did the P-1 testing that he frankly should have done... found a factor on one of them. Saves me time when I eventually do the independent test. :smile:

[URL="http://www.mersenne.org/report_exponent/?exp_lo=102641599&full=1"]M102641599[/URL]

Can 44750009 please be triple checked. My machine returned a mismatch on that exponent.

Thanks,
Chris.

[QUOTE=stebbo;448809]Can 44750009 please be triple checked. [/QUOTE]

Enqueued