| Author |
Message |
|---|
Scott Brown Volunteer moderator
Project administrator
Volunteer tester
Project scientist
Send message
Joined: 17 Oct 05
Posts: 2329
ID: 1178
Credit: 15,614,073,567
RAC: 11,561,086
                                           
|
|
The last GFN32768 prime found a few days ago is now only 50 spots or so from falling off the Top 5000 primes list. Let's give this project a push and see how many more of these we can get in the Top 5000 list!
____________
141941*2^4299438-1 is prime!
|
|
|
|
|
|
Aaaaaaaand it's off the list. My number 5177652^32768+1 was 100 digits short of the bottom line. Hop to it, all! |
|
|
|
|
|
I just want to celebrate that user composite has found no less than twelve primes in the GFN32768 within just a couple of weeks. The b values of his finds are in the range 7663846 through 8191116.
NB! No primes in GFN32768 are in the Top 5000 list anymore. At the time of writing, that would require a b value of nearly 136,000,000,000.
/JeppeSN |
|
|
 compositeVolunteer tester
Send message
Joined: 16 Feb 10
Posts: 1022
ID: 55391
Credit: 888,486,996
RAC: 136,636
                       
|
|
The b limit with the x87 transform is 64M. That's far short of 136G needed to be near the bottom of the Top 5000 list.
I'm wondering if there's any point to continuing this project on PRPNet, other than collecting long lists of smaller primes. It has been established at this thread that results from lower N like 32768 will not be used to filter work units of higher N on PRPNet.
Is there a commitment to finish work that PRPNet has reserved? How much work is left in the reservation?
GFN32768 would be a nice shorter task for CPU-only BOINCers to attain credit toward GFN badges, but it goes against the spirit of PrimeGrid, which runs non-conjecture projects that place in Top 5000. That said, there are a lot of people that take pride in their badges, who lack patience or CPU speed for longer-deadline CPU tasks. |
|
|
|
|
[...] I'm wondering if there's any point to continuing this project on PRPNet, [...]
May I ask what your motivation has been for doing such an impressive effort on GFN32768? You are finding almost 1.0 new prime per day. Are you hoping to upgrade your badge? /JeppeSN |
|
|
compositeVolunteer tester
Send message
Joined: 16 Feb 10
Posts: 1022
ID: 55391
Credit: 888,486,996
RAC: 136,636
|
|
When I started on this port I wasn't aware that it would never reach Top 5000. But now it just seems like PSA work that has been neglected, probably because it's x87-only and it's not Top 5000.
I'm suggesting 32768 and 64556 as complementary additions to BOINC for GFN badge seekers that don't have AVX. I'll still be crunching PSA if these ports are moved to BOINC.
The current prime find rate looks impressive because the tasks are so short and the primes are locally dense. Why have people who did more tests on this port find so many fewer primes? Was a previous version of the seive not as effective or the primality test defective, or were larger b's being tested? Or is it just blind luck?
|
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13804
ID: 53948
Credit: 345,369,032
RAC: 3,564
                     
|
When I started on this port I wasn't aware that it would never reach Top 5000. But now it just seems like PSA work that has been neglected, probably because it's x87-only and it's not Top 5000.
I'm suggesting 32768 and 64556 as complementary additions to BOINC for GFN badge seekers that don't have AVX. I'll still be crunching PSA if these ports are moved to BOINC.
The current prime find rate looks impressive because the tasks are so short and the primes are locally dense. Why have people who did more tests on this port find so many fewer primes? Was a previous version of the seive not as effective or the primality test defective, or were larger b's being tested? Or is it just blind luck?
There has been no sieving done in a long time, so we're using the same sieve file as before. There's nothing to suggest that there's been anything wrong with the software. (And, by the way, I have been verifying that the PRP's you're finding with Genefer are in fact prime.)
A very big part of it is simply that you're doing a lot of tests, and have done more n=32768 tests than anyone else except Scott Brown. (Even without AVX, computers are much faster than they used to be.) You've got more primes than he does, and that's probably just luck. More interesting is that Honza has done almost as many tests as you have but only has found one prime. Again, I suspect that's luck.
____________
My lucky number is 75898524288+1 |
|
|
Scott BrownVolunteer moderator
Project administrator
Volunteer tester
Project scientist
Send message
Joined: 17 Oct 05
Posts: 2329
ID: 1178
Credit: 15,614,073,567
RAC: 11,561,086
|
|
Indeed, much of my luck (as compared to Honza's) on this port came from the fact that I did a lot of crunching as a solo cruncher while others were off to other PRPnet and/or BOINC projects. I suspect that this is why you are finding so many primes compared to others who did large amounts of work when more users were active on the port.
That said, I have thrown an old T2450 laptop back into the mix (just got a replacement battery for it)...maybe I'll get lucky, too! ;)
|
|
|
|
|
|
Also note that the GFN65536 primes are also falling out of the Top 5000. I found a way to link this search result which shows that (right now) eight GFN65536 numbers have fallen out of the Top 5000 (all were known before the birth of PrimeGrid).
/JeppeSN
|
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13804
ID: 53948
Credit: 345,369,032
RAC: 3,564
|
Also note that the GFN65536 primes are also falling out of the Top 5000. I found a way to link this search result which shows that (right now) eight GFN65536 numbers have fallen out of the Top 5000 (all were known before the birth of PrimeGrid).
/JeppeSN
I suspect that within 5 or 6 years PPSE will have pushed n=65536 completely out of the top 5000.
____________
My lucky number is 75898524288+1 |
|
|
