What happened Dec 22? ( Big jump in assignments )
 

In reviewing my recent work assignments, I noticed that in the morning of Dec 22 I got some manual assignments of 95 542 xxx.
Then, later in the day, I got some starting 98 010 xxx.
New assignments have been steadily increasing from there.
I looked at some of the statistics on mersenne.ca and don't see anything abnormal.
What happened?

Several power users (Ben Delo, Ryan Propper) who joined the project this year have nearly tripled the daily throughput - ISTR it was averaging ~250 1st-time tests per day, currently we are getting 700-800 per day. So the exponents handed out by the server have been climbing much more rapidly than they used to - at the current rate I expect 1st-time exponents will pass 100M around end of January (quick math: ~50000 primes between 99M and 100M, figure ~20000 of those have survived TF/p-1, divide by 800/day).

Aaron/George can comment authoritatively, but I expect what happened to you on the day in question is that one of the above folks reserved a large batch of expos, perhaps enough to keep all their machines well-fed through the holidays.

Sadly we seen to have lost Ryan.
I hope he is just secretly running more massive workload in the background to appear shortly.

[QUOTE=petrw1;533792]Sadly we seen to have lost Ryan.
I hope he is just secretly running more massive workload in the background to appear shortly.[/QUOTE]

I'm not doing any GIMPS work at the moment, but may do some more double-checking in the new year. :smile:

[QUOTE=ryanp;533793]may do some more double-checking in the new year. :smile:[/QUOTE]

:thumbs-up:

[QUOTE=ewmayer;533788]Several power users (Ben Delo, Ryan Propper) who joined the project this year have nearly tripled the daily throughput - ISTR it was averaging ~250 1st-time tests per day, currently we are getting 700-800 per day. So the exponents handed out by the server have been climbing much more rapidly than they used to - at the current rate I expect 1st-time exponents will pass 100M around end of January (quick math: ~50000 primes between 99M and 100M, figure ~20000 of those have survived TF/p-1, divide by 800/day).

Aaron/George can comment authoritatively, but I expect what happened to you on the day in question is that one of the above folks reserved a large batch of expos, perhaps enough to keep all their machines well-fed through the holidays.[/QUOTE]

That makes sense, Ernst; thanks.

[QUOTE=ryanp;533793]I'm not doing any GIMPS work at the moment, but may do some more double-checking in the new year. :smile:[/QUOTE]There was a whole bunch of triple checks that you have assigned to you. Are you going to continue on with those?

[QUOTE=ewmayer;533788]...at the current rate I expect 1st-time exponents will pass 100M around end of January.[/QUOTE]

Eh, it might happen just a smidge sooner than that - just grabbed a fresh assignment for one of my cellphone-cluster nodes, it is 99959257. On the chance that maybe some user's work-fetch script has run rampant I check the next 2 smaller unfactored exponents, but both are assigned to different users.

[url]https://www.mersenne.org/report_milestones/[/url] says there are over 1200 exponents available for first time primality testing below 90M. I'm routinely getting new assignments in 90M.

It is Cat 3 that is approaching 100M while Cat 4 is approaching 103M:

[url]https://www.mersenne.org/thresholds/[/url]

If Ryan and Ben would set their accounts to do Cat 1 and 0, (and 20% DC's) that would help keep the first time LL compressed.