Request for new work insertion

[hhh]

Unless we want to go back to DoubleCheck(TM) again for a while, I think it would be a good time to insert work up to 17M.

H.

[GP500]

I woudl say do some double checking.

Server's Assignment Queues	0 minutes old	Last 24 Hours
Queue Name	# Tests	Min n	Max n	# Done	n Increase

first-pass	2250	15853724	15999785	62	+7090

second-pass	35899	  11526569	13999781	6	+624
Overall	        38149	11526569	15999785	200	+624

more then 4M between Low-n for second and Low-n of first-pass.

4 Miljoen numbers that's just huge.
The change that there is a needed check in there is not small.

Double checks haven been proven to be needed.

[jasong]

4 Miljoen numbers that's just huge.
The change that there is a needed check in there is not small.

Double checks haven been proven to be needed.

Assuming first-pass was run dry AND completed, what's a good level to shoot for in second pass?

Anybody want to tackle this for me?

[vjs]

Based on past experience we should run secondpass out to the last prime found.

19249•2^13018586+1

perhaps somewhere around 13.2M or the level of the eleet accound of the past? Might be optimal


--------------

However this time around I'm going to agree with hhh.

When this que runs out let it do doublechecks to around 12M.
Reinsert firstpass to 17M then back to double till 13.5M

I don't think we will find another prime until around 18-18.5M firstpass. No to say that we havn't missed one... I just think firstpass will need to get to that level before we find another.

[vjs]

BTW my guess this time around...


55459 That will be very cool prime.

[Zuzu]

BTW my guess this time around...


55459 That will be very cool prime.

This is the most likely, but I would prefer low-weights 67607 or 22699. It would be coolest if 55459 were the last...
BTW the estimations I made months ago lead to a 50% chance for the next prime before n = 24M (the 22M I estimated then were based on a 14.3M limit on the firstpass tests at that time, neglecting secondpass; now we're nearing 16M). That said, once again, we can be lucky (a prime before 20M, for 67607 or 22699) or unlucky (next prime after 35-40M, for 55459). Insh'Allah!

[engracio]

Ooooooops.. fail over to second pass is not happening.

Gonna be a long wait for more work units.

The well is dry,,,

[DOSGuy]

Yup, the server isn't automatically sending out second pass tests now that first pass is out of work. One of my clients finished a test and couldn't grab any new work, so I just added QQQsecondpass to my username and the unemployed client grabbed an 11.5M right away. That works fine as a temporary fix until the server starts assigning work to first pass usernames again. Unfortunately, I'm forced to wonder how many hundreds of computers will run idle and go unnoticed if the assignment queue isn't fixed quickly.

[Sceptic]

I sent Alien88 an e-mail about the situation, so hopefully the f-p queue is repopulated again soon

Sceptic

[Alien88]

I sent Alien88 an e-mail about the situation, so hopefully the f-p queue is repopulated again soon

Sceptic

Thanks for the e-mail about it.. I'm insanely busy so I forgot to keep an eye on the queue. First pass didn't quite fail over like I expected it to, but I changed it to hand out second-pass for the time being.

Repopulating the queue will take some bit of time for various reasons, but it should happen soon. In the mean time, we'll be crunching away at second-pass and alternatively I might repopulation the error-fix queue for tests that still don't have enough matching residues.

[vjs]

This would be a very good idea.

I might repopulation the error-fix queue for tests that still don't have enough matching residues.

There shouldn't be too many...


If you can analyze the results for particular IP adress that wouild be even cooler. If an IP adress is generating alot of errors early doublechecks for those would be prudent.