Oops - New Prime! (M49 related)

[Gordon]

Originally Posted by petrw1

Seems YES: My first Pentium 4 from about 22 years ago took 1 month for a LL test in the current Leading Edge 33M range.

Project wasn't running in 1994, seems like you have dodgy memory!

[Madpoo]

Originally Posted by TObject

What kind of notification was expected? Sound?
Was it the NT Service flavor of Prime95? Services cannot make sounds.

The system is on the Primenet server side and is supposed to send an email to the discoverer as well as George, Scott, et al.

So that's separate from the setting to have the system running Prime95 make a noise, which I'm sure Curtis disabled on those systems (or as you mentioned, they're running the service version, which makes more sense).

[Xyzzy]

We were going through some old files today and we found this old screen capture.

[LaurV]

Originally Posted by Xyzzy

We were going through some old files today and we found this old screen capture.

I give you a sticker if you remember what you had in the "Mike" folder shown on that screen capture!

[davar55]

Originally Posted by Prime95

I know I was the first human to know about M42603801 for the same reason as Aaron. When email notification works, there is no way to know which human is the first to open up the email to become the first human aware of a new prime.
Correct attribution all depends on one's definition of "discoverer".
BTW, it is a pretty cool feeling to be the only one on the planet to know about a new Mersenne prime.

It's super-cool to have been one of the first.
This project and its members are doing something
of great value to the world at large, even if most
of them know practically nothing about it.
Congratulations, and onward to M50*.

My numerologue for the new Mersenne Prime's exponent (mod 10):

Knowing this is incredibly calming to personal involvement.

So glad I stuck around.

[KyleAskine]

Congratulations everyone! Great work by all of you!

[ixfd64]

https://www.ucmo.edu/news/cooper.primenumber2016.cfm

[dbaugh]

Curtis has done over 500 times as much GIMPS LL work as I have. As a consequence it is very unlikely that I will find a Mersenne prime before he finds his next one. It is also likely that the University of Central Missouri will spend in the neighborhood of $500K on additional electricity for the search. I think it is worth it and would if I could. I would like to see Amazon or Google devote the resources to find the next one. How many LL GHz-days did Curtis contribute between M48 and M49? How many LL GHz-days were contributed by everyone else over the same time period? That should give some big boy an idea of what resources it would take to find M50 before 2016 is done.

[ATH]

Originally Posted by dbaugh

How many LL GHz-days did Curtis contribute between M48 and M49? How many LL GHz-days were contributed by everyone else over the same time period?

Curtis did 5.3M Ghz-days the last year
http://www.mersenne.org/report_top_500_ll/

But that must have been more than previous years due to new computers. 3 years * 5.3M ~ 16M Ghz-days but his lifetime total is 18.9M:
http://www.mersenne.org/report_top_5...1-01&end_date=

So it is probably around 10M Ghz-days for the last 3 years since M48.

Originally Posted by dbaugh

How many LL GHz-days were contributed by everyone else over the same time period?

We should really have a rank 0 in all these reports showing the total GIMPS throughput in that category, or show the total throughput shown at the bottom of the report. Is that something that might be worth adding?

[Dubslow]

Originally Posted by dbaugh

That should give some big boy an idea of what resources it would take to find M50 before 2016 is done.

I think you are underestimating the amount of time and computing power it will take to find the next largest Mersenne prime, by at least 1 if not 2 orders of magnitude.

http://mersenneforum.org/showpost.ph...7&postcount=14

[airsquirrels]

Originally Posted by Dubslow

I think you are underestimating the amount of time and computing power it will take to find the next largest Mersenne prime, by at least 1 if not 2 orders of magnitude.

http://mersenneforum.org/showpost.ph...7&postcount=14

There are some big potential tech changes that could speed this up considerably. 512bit instructions for one. Also my analysis of the state of LL on GPU is that some (large) optimization effort alone will probably improve its performance 2-3x, enough to bring some GPUs over from the TF front. Current decent GPUs can get 50-60 GhzDay/Day on LL vs 700 on TF, if that improves to 200GhzDay/Day or more on GPU it will change the LL power output of the project significantly.

I'm betting we see another one found by late 2017, though I agree with Madpoo that there is a decent chance it is actually found in DC land.