Bases 101-250 reservations/statuses/primes

[MyDogBuster]

Hello Gary,
Sorry for the lack on information. I can understand that a lot of people sign up and never complete their work.
I am Phillip and I live in Virginia, USA. I enjoy searching for prime numbers in my spare time. I have access to approximately 25-30 four-core machines, when they are available, which I use for mathematical computation. I am using srsieve for sieving (complete) and pfgw for testing/proving. It appears that these are popular programs and I have been using pfgw for many years. I thought that 25k was small enough that some progress could be made fairly quickly ( currently, I have found 35 primes between 25k and 50k and 30 more primes between 50k and 100k). Is there a certain time that one looks for in reserving a range? I predict 3-5 months to complete this assignment. Is that reasonable?

Welcome Phillip. I'm MyDogBuster (Ian) and I also Admin this project and NPLB. Just wanted to say hi (I'm quite close to you, I'm in Delaware) If you have any technical questions, we have enough people here that are more than willing to help. I'll let Gary answer your questions. Have fun.

[gd_barnes]

Originally Posted by MagentaSunrise

Hello Gary,
Sorry for the lack on information. I can understand that a lot of people sign up and never complete their work.
I am Phillip and I live in Virginia, USA. I enjoy searching for prime numbers in my spare time. I have access to approximately 25-30 four-core machines, when they are available, which I use for mathematical computation. I am using srsieve for sieving (complete) and pfgw for testing/proving. It appears that these are popular programs and I have been using pfgw for many years. I thought that 25k was small enough that some progress could be made fairly quickly ( currently, I have found 35 primes between 25k and 50k and 30 more primes between 50k and 100k). Is there a certain time that one looks for in reserving a range? I predict 3-5 months to complete this assignment. Is that reasonable?

I would like to reserve S151 to 200K.

Hi Phillip,

Once again, welcome to CRUS and thank you for the detailed info! It does sound like you know what you are doing so I will "officially" reserve the range for you. Yes, we do sometimes have new people who make reservations way out of line with their resources and/or their knowledge of how to search them. It sounds like you have well more than enough resources to tackle this range so we greatly appreciate your contribution. There is no specific timeline for completing reservations. I generally request that people "check in" with a status update at least once every 2-3 months so that we know that they are progressing.

What I will need from you when you are done is the primes and the residues files, i.e. the pfgw.out file from PFGW.

The only warning that I'll make is the "boredom factor". 3-5 months is a long time to let 100+ cores work on one effort in which only "smallish" primes are found so it will require some stamina on your part. I also think that you'll find that it may take longer than that starting from 175 k's. n=50K-100K likely took you ~6 times as long as n=25K-50K took and n=100K-200K will likely take you ~6 times as long as n=50K-100K assuming the removal of a normal # of k's as you go. (If primes did not remove k's, it would be almost exactly 8 times as long.) Alas, that is OK, you can take a year or more on it if you wish. This is just a heads up. If you end up choosing to stop at n=150K, that is OK with us. Actually I would be impressed with anyone who has the stamina to take so many k's to n=200K.

Good luck and may the prime gods be with you.


Gary

[rogue]

Originally Posted by MagentaSunrise

I have access to approximately 25-30 four-core machines, when they are available, which I use for mathematical computation. I am using srsieve for sieving (complete) and pfgw for testing/proving. It appears that these are popular programs and I have been using pfgw for many years.

I would like to reserve S151 to 200K.

If all of these machines are on the same network, you should consider running PRPNet. This way all of the work can be managed from a central location. Because all test results are collected on one machine, thus don't need to access all machines to get residues. You can also track progress via webpages and get an idea how much work is really left.

[MyDogBuster]

Reserving R136 & R199 to n=50K

[MagentaSunrise]

Originally Posted by rogue

If all of these machines are on the same network, you should consider running PRPNet. This way all of the work can be managed from a central location. Because all test results are collected on one machine, thus don't need to access all machines to get residues. You can also track progress via webpages and get an idea how much work is really left.

Thanks rogue. These machines are all on the same network -- all files are stored on the same server -- although I currently must ssh to each machine separately to start a new file. I will look at PRPNet to see if I can get it to work on this system and save some of the time spent starting instances of the program.

[MyDogBuster]

R199 tested n=25K-50K - 13 primes found - 52 remain

4034*199^25022-1
11366*199^26223-1
10386*199^26735-1
7706*199^30077-1
10674*199^32804-1
6306*199^35135-1
5604*199^37222-1
10424*199^37736-1
7796*199^41145-1
7944*199^41480-1
9174*199^40860-1
6654*199^44344-1
7544*199^49834-1

Results emailed - base released

[henryzz]

Wouldn't sr2sieve be better than srsieve for S151? How many ks is the max for sr2sieve?

[rogue]

Originally Posted by henryzz

Wouldn't sr2sieve be better than srsieve for S151? How many ks is the max for sr2sieve?

You would have to use the -x parameter and by doing so would slow down sr2sieve. That would probably still be faster than srsieve.

[henryzz]

Originally Posted by rogue

You would have to use the -x parameter and by doing so would slow down sr2sieve. That would probably still be faster than srsieve.

I got bored and tested it. I very lightly sieved with srsieve and then did some tests starting at 1e11.

srsieve 32-bit:       140000 p/sec
srsieve 64-bit:       210000 p/sec
sr2sieve 64-bit -x:   350000 p/sec
sr2sieve 64-bit -c:   580000 p/sec

As it turns out working out the legendre tables only took around 3 minutes. I think this is because the ks are relatively small. The speed noticably slowed down for the larger ks.

Something worth noting is that 60516=246^2.

[MyDogBuster]

Reserving S136 all k's to n=50K

[MyDogBuster]

S136 tested n=25K-50K
26 primes found - 74 remain
Results emailed - Base released