Bases 33-100 reservations/statuses/primes

[mdettweiler]

Sierp. base 33 is complete to n=300K, no primes. Results are attached for 100K-300K; releasing.

[gd_barnes]

Originally Posted by mdettweiler

Sierp. base 33 is complete to n=300K, no primes. Results are attached for 100K-300K; releasing.

I was about to say "Wow, that was a lot of work fast!". Then I looked in the file and see that there was only just over 2,000 tests or ~1% of the entire n-range. Holy cow, that is a LOW weight k! Typical low weight k's here that are one of the final few k's remaining on a base usually have about 2-3% of the entire n-range remaining after sieving to the optimum depth.

Just curious: How long did that take you? I'm assuming you used a full quad running a PRPnet server.

With only ~1,000 tests for every n=100K range and the chance of prime falling well below 1 in 15,000 at n=300K, we may be searching this one to n=5M-10M to find the final prime as the odds of prime keep getting smaller.

[gd_barnes]

Originally Posted by MyDogBuster

Reserving Sierp Base 49 (all 7 k's) from n=25K-40K

Ian,

I'm confused. You reserved base 49 and not base 48 (that I could find). But you searched base 48. Are you still going to search base 49? You stated the correct # of k's for base 49, which leads me to think you're still going to do it.

One more thing: I closely checked the results file and it does appear that you searched all k's for base 48. I was somewhat surprised that there was only one prime. Let me know if anything seems out of sorts here.


Gary

[MyDogBuster]

I'm confused. You reserved base 49 and not base 48 (that I could find). But you searched base 48. Are you still going to search base 49? You stated the correct # of k's for base 49, which leads me to think you're still going to do it.

One more thing: I closely checked the results file and it does appear that you searched all k's for base 48. I was somewhat surprised that there was only one prime. Let me know if anything seems out of sorts here.

I will do Base 49 soon. I have it sieved, just nothing to run it on yet.

I did re-check my results for Base 48 and found nothing out of the ordinary.

I did have a PRPNET barf on it about halfway, but when I checked it out, the candidates file looked just fine. Nothing missing. 5975 original tests, 5713 complete. I'm assuming the rest was k359 being zapped by PRPNET. I'd chalk it up to one our infamous galactic holes. Someone doing n=40K-70K may find and bunch of the remaining.

[mdettweiler]

Originally Posted by gd_barnes

I was about to say "Wow, that was a lot of work fast!". Then I looked in the file and see that there was only just over 2,000 tests or ~1% of the entire n-range. Holy cow, that is a LOW weight k! Typical low weight k's here that are one of the final few k's remaining on a base usually have about 2-3% of the entire n-range remaining after sieving to the optimum depth.

Just curious: How long did that take you? I'm assuming you used a full quad running a PRPnet server.

With only ~1,000 tests for every n=100K range and the chance of prime falling well below 1 in 15,000 at n=300K, we may be searching this one to n=5M-10M to find the final prime as the odds of prime keep getting smaller.

According to the timestamps on the first and last results as returned to my PRPnet servers, it took me close to exactly 14 days to do the range. That was, as you said, with a full quad (Q6600 overclocked to 2.8Ghz).

I actually hadn't realized this was quite so low-weight. The time it took to run didn't seem too particularly strange to me at first. But now that you mention it, you're right, when I compare it to the approximate figures I remember for other bases I've done, it does seem rather surprisingly low-weight. (And that's considering that the bases I'm comparing it to are also going to be on the lower-weight side since they were also the last k or last two k's of a conjecture. )

Hmm, if I'd realized it was quite this light, I would have probably sieved much higher in terms of n-range; probably to 500K rather than 300K.

[mdettweiler]

Originally Posted by MyDogBuster

I will do Base 49 soon. I have it sieved, just nothing to run it on yet.

I did re-check my results for Base 48 and found nothing out of the ordinary.

I did have a PRPNET barf on it about halfway, but when I checked it out, the candidates file looked just fine. Nothing missing. 5975 original tests, 5713 complete. I'm assuming the rest was k359 being zapped by PRPNET. I'd chalk it up to one our infamous galactic holes. Someone doing n=40K-70K may find and bunch of the remaining.

Ian, if you don't already have something comparable, I can send you my script that I use for checking to see whether a given results file contains all the same results as a sieve file. That can be somewhat more reliable than checking the # of tests; on a couple of occasions I've actually encountered situations where the same number of tests that were missing were exactly compensated for by duplicates elsewhere in the file, so it's definitely important to check.

[MyDogBuster]

Ian, if you don't already have something comparable, I can send you my script that I use for checking to see whether a given results file contains all the same results as a sieve file. That can be somewhat more reliable than checking the # of tests; on a couple of occasions I've actually encountered situations where the same number of tests that were missing were exactly compensated for by duplicates elsewhere in the file, so it's definitely important to check.

Thanks Max, but I'm in the process of writing one as we speak.

That barf on Base 48 was with the old version of PRPNET. 2.4.0 seems to have fixed that problem.

[gd_barnes]

I got a PM from Karsten on Riesel base 35. He has now searched it to n=10094 and reported the following primes for n=9230-10094:

196828 9235
115886 9292
129350 9316
155900 9318
86234 9340
257762 9340
208204 9345
239732 9372
31222 9423
248602 9425
159902 9442
131444 9452
243584 9456
255758 9474
83966 9476
117854 9520
98300 9570
203470 9579
1426 9607
92546 9614
59006 9648
97932 9654
257570 9664
228672 9666
29570 9668
109252 9683
198788 9684
167348 9692
262756 9701
181868 9708
225928 9727
83578 9731
282548 9782
40154 9784
176302 9787
205570 9803
201556 9821
240912 9849
162974 9856
114986 9908
10504 9951
67876 9965
93634 9967
143206 9981
279458 9986
81748 10001
105964 10049
181190 10078
227594 10082
193332 10089

Karsten, I removed 69002*35^9222-1 from your PM list. You had already reported it and it was already shown on the pages.

A total of 50 primes. This makes 1155 k's remaining at n=10094. Thanks for the update Karsten.

With this update, we have now accomplished an amazing feat: All Riesel bases < 39 with the exception of huge bases 3/7/15 have been searched to n>=10K! To take it a step further, Ian has already taken on the dubious task of taking Riesel base 39 up to n=10K. Conjectured at k=1352534, he has it at n=7K with no less than 6182 k's remaining! Base 40 is the next true monster conjectured at k=3386517. Past that, we have all bases 41 to 50 at n=25K. Base 51 conjectured at k=8632534 is the next one that looks extremely tough at this point.

The only thing stopping us from the same on the Sierp side now is base 35. Although a very large conjecture of k=214018, it is smaller than the Riesel base 35 conjecture of k=287860. Bases 39 and 40 are also conjectured as much smaller than their Riesel counterparts. The Sierp side is much more doable to base 50/n=10K then the Riesel side since bases 35, 39, & 40 all have much smaller conjectures.

I'm not advocating starting these extremely tough bases any time soon. The direction of the project is excellent right now with many different varied efforts being completed by people of all tastes. I just thought I'd mention them for future reference. From an admin standpoint, huge new bases can be a big headache.


Gary

[Siemelink]

Originally Posted by gd_barnes

Base 40 is the next true monster conjectured at k=3386517.

Gary

Ah yes, I've just taken this one to n = 5000. I'll post what I have in the weekend.

Cheers, Willem.

[gd_barnes]

Originally Posted by Siemelink

Ah yes, I've just taken this one to n = 5000. I'll post what I have in the weekend.

Cheers, Willem.

NO!! Not 10's of thousands of k's remaining. HELP!! lol

[rogue]

More primes:

12056*58^35062-1
642*58^35088-1
58082*58^35515-1
20826*58^35518-1

38823*58^36929-1

56337*58^37370-1
24204*58^37967-1

90212*58^38591-1
87707*58^38783-1

61779*58^39060-1
64112*58^39243-1

They are starting to get a little thin. I need about 30 more to get under 300 for this base. I intend to stop at 50,000, which is as far as I had sieved.