Oops - New Prime! (M49 related)

[Mini-Geek]

Originally Posted by ixfd64

Does anyone know how much faster GPU to 72 allowed us to find this new prime?

Originally Posted by 0PolarBearsHere

It looks like it was TFed to 75 bits, not sure how much was via GPUto72 though. Chris should be able to tell us.
http://www.mersenne.ca/exponent.php?...tails=74207281

You'd have to take into account factors found, and that CPUs could spend their time LLing instead of TFing. I wouldn't be surprised if looking at an individual candidate that went to LL, GPU to 72 actually (apparently) slowed it down. It does speed the project up, but I don't know by how much.

[chalsall]

Originally Posted by ixfd64

Does anyone know how much faster GPU to 72 allowed us to find this new prime?

It's impossible to say exactly, but a back-of-the-envelope guestimate is approximately 6% faster (based on our now going to 75 bits instead of Primenet's default of 70). And, yes, this candidate was TF'ed and P-1'ed via GPU72.

It was actually because of the new Recycling Rules / Assignment Categories which let us find this as far up, and as soon, as we did; the MP was (I believe) a "Cat 4" assignment.

[chalsall]

So, quite a bit of coverage today in the "mainstream media". And we're seeing a huge surge in DC Cat 4 assignments (as expected).

I have to say it's quite amusing seeing just how wrong just about every outlet gets the "facts", even so called "respected" sources.

Makes one wonder just how wrong they get everything else....

[retina]

Originally Posted by chalsall

I have to say it's quite amusing seeing just how wrong just about every outlet gets the "facts", even so called "respected" sources.

Makes one wonder just how wrong they get everything else....

You've only just worked that out now?! They never let boring facts get in the way of a good story.

[airsquirrels]

Originally Posted by chalsall

So, quite a bit of coverage today in the "mainstream media". And we're seeing a huge surge in DC Cat 4 assignments (as expected).

I have to say it's quite amusing seeing just how wrong just about every outlet gets the "facts", even so called "respected" sources.

Makes one wonder just how wrong they get everything else....

We did a bit of fun promotion in Twitter about the prime that has generated a pretty decent click rate. I agree, every article I have read contains groan worthy statements, even things my own coworkers wrote about it and then 'tweaked' after I fact checked.

[Madpoo]

Originally Posted by chalsall

It's impossible to say exactly, but a back-of-the-envelope guestimate is approximately 6% faster (based on our now going to 75 bits instead of Primenet's default of 70). And, yes, this candidate was TF'ed and P-1'ed via GPU72.

It was actually because of the new Recycling Rules / Assignment Categories which let us find this as far up, and as soon, as we did; the MP was (I believe) a "Cat 4" assignment.

Yeah, I'm pretty sure it was cat 4 when assigned last August. I mentioned it had already been "churned" by one previous Anonymous user who got the assignment and then abandoned it, so Curtis' machine picked it up very soon after it expired (same day I think).

[Madpoo]

Originally Posted by chalsall

So, quite a bit of coverage today in the "mainstream media". And we're seeing a huge surge in DC Cat 4 assignments (as expected).

I have to say it's quite amusing seeing just how wrong just about every outlet gets the "facts", even so called "respected" sources.

Makes one wonder just how wrong they get everything else....

I looked back at M48 regarding the date of it's announcement and how new user sign-ups progressed.

Day 1 (Feb 5) saw an approx tripling over normal levels, but it was day 2 that saw a huge spike of 10-12 times the normal. Day 3 was about 9-10 times, day 4 was 6-7 times, and then days 5-14 after the announcement saw it taper off from 3x the normal amount to just somewhat above normal. By 2 weeks later things were back to around typical levels.

Going by that past record, this time around seems like it could match that. Yesterday saw an increase of about 5x typical new users, and so far today we're at 10x and climbing.

The real test is how many of those will stick it out and finish at least one test. Here's hoping it's a decent amount.

[airsquirrels]

Originally Posted by Madpoo

I looked back at M48 regarding the date of it's announcement and how new user sign-ups progressed.

Day 1 (Feb 5) saw an approx tripling over normal levels, but it was day 2 that saw a huge spike of 10-12 times the normal. Day 3 was about 9-10 times, day 4 was 6-7 times, and then days 5-14 after the announcement saw it taper off from 3x the normal amount to just somewhat above normal. By 2 weeks later things were back to around typical levels.

Going by that past record, this time around seems like it could match that. Yesterday saw an increase of about 5x typical new users, and so far today we're at 10x and climbing.

The real test is how many of those will stick it out and finish at least one test. Here's hoping it's a decent amount.

Not to keep beating this drum but, about that plan we've discussed to save checkpoints that make even churners work valuable...

[airsquirrels]

So this happened:

https://twitter.com/LargestPrimeNum

[paulunderwood]

http://www.bbc.co.uk/news/technology-35361090

[science_man_88]

Originally Posted by airsquirrels

So this happened:

https://twitter.com/LargestPrimeNum

if they keep rate this up 24/7 it will only take them 55.4 days of tweeting roughly until they completely type it good long advertisement.edit: oops for some reason I thought they'd at least hit the 140 char limit turns out they are only doing about 100 per tweet so multiply that by 1.4 and then maybe by 1.5 for the 16 hours they are more likely to do per day. so up to over 110.8 days now.