Henry is quite right- running a sieve for a high-weight k higher than n= 2M is a waste of effort. It would take years for a n=100M sieve to reach anything like optimal depth, and quitting sooner negates the effect of the more efficient sieve.

Even going to 10M would be more than a decade-long project to sieve and test, for a k with no special significance. If you insist on sieving higher than you will ever test, sieve to n=2M. For a project you might finish in a few years, sieve to 1M.
-Curtis

[QUOTE=KEP;180900]k=49185 has been doublechecked to n=50K, there was a total of 79 primes. Attached is the residuals for all non-prime n's. Following 79 n's has been doublechecked as primes for n<=50K:

9, 16, 21, 24, 25, 26, 32, 33, 42, 45, 50, 53, 89, 90, 92, 93, 100, 120, 123, 167, 179, 192, 201, 204, 234, 276, 303, 340, 346, 371, 402, 587, 596, 674, 687, 741, 779, 803, 881, 903, 956, 962, 963, 1041, 1065, 1085, 1636, 1673, 1731, 1979, 2217, 2465, 2594, 2607, 3090, 3234, 3353, 3551, 4324, 5421, 7081, 7130, 7297, 7304, 7444, 9107, 11010, 11955, 12562, 13173, 13779, 14421, 18025, 20850, 21771, 22674, 22769, 26771, 46553

Next update will be early august when primes for n>50K to n<=100K will be added aswell as the residuals for the doublechecked non-prime n's will be added.

Regards

KEP[/QUOTE]

For some weird reason, the primes for n 1, 2, 3 and 4, which makes a total of 83 primes for n<=50K, wasn't listed. Else did everything else checks out.

Regards

KEP

[QUOTE=VBCurtis;199604]Henry is quite right- running a sieve for a high-weight k higher than n= 2M is a waste of effort. It would take years for a n=100M sieve to reach anything like optimal depth, and quitting sooner negates the effect of the more efficient sieve.

Even going to 10M would be more than a decade-long project to sieve and test, for a k with no special significance. If you insist on sieving higher than you will ever test, sieve to n=2M. For a project you might finish in a few years, sieve to 1M.
-Curtis[/QUOTE]

I did some testing on a k=49185, n=5M sieve range, and sieving that range compared to an n=100M range, is only 3 times faster, while it should have been more than 60 times faster, if it should be more efficient to sieve an n=5M range than to sieve an n=100M range. So I will stick with sieving the n=100M range and then break of parts as I get to have them reach optimal sievedepth :smile:

KEP

[QUOTE=KEP;199717]I did some testing on a k=49185, n=5M sieve range, and sieving that range compared to an n=100M range, is only 3 times faster, while it should have been more than 60 times faster, if it should be more efficient to sieve an n=5M range than to sieve an n=100M range. So I will stick with sieving the n=100M range and then break of parts as I get to have them reach optimal sievedepth :smile:

KEP[/QUOTE]

I don't think you understand- nobody said the sieve would be more efficient for a smaller range. We said you'll never LLR the range above 2M, so there is no *reason* to sieve it. Even k's for which we intend to test very high, such as k<31, we have no sieve higher than 5M running. Do a little math to calculate how long it would take to test 500k-1M, 1M-2M, 2M-3M; realize it will be something like 2030 before you finish 2M-3M working on your own. The efficiency you gain sieving only matters when you test ALL of the candidates in the sieve- and 100M will *never* be tested, let alone 10M.

Consider: compute the time to test 900k-1M range. The 9M-10M range will take one THOUSAND times longer than 900k-1M. 3 CPU months for the former, perhaps- 250 years for the latter, then. Do you really want to test only this k the rest of your life?
How is spending 5 CPU-years up front on a sieve better for finding primes than what we're suggesting?
-Curtis

[QUOTE=VBCurtis;199761]I don't think you understand- nobody said the sieve would be more efficient for a smaller range. We said you'll never LLR the range above 2M, so there is no *reason* to sieve it. Even k's for which we intend to test very high, such as k<31, we have no sieve higher than 5M running. Do a little math to calculate how long it would take to test 500k-1M, 1M-2M, 2M-3M; realize it will be something like 2030 before you finish 2M-3M working on your own. The efficiency you gain sieving only matters when you test ALL of the candidates in the sieve- and 100M will *never* be tested, let alone 10M.

Consider: compute the time to test 900k-1M range. The 9M-10M range will take one THOUSAND times longer than 900k-1M. 3 CPU months for the former, perhaps- 250 years for the latter, then. Do you really want to test only this k the rest of your life?
How is spending 5 CPU-years up front on a sieve better for finding primes than what we're suggesting?
-Curtis[/QUOTE]

A test at n=500K takes ~6 minutes to compute. But I didn't read and understand what you explained to me, the same as you meant with what you were writing. I'm close to p=1T, and maybe I'll break the n>50K to n<=100M range in several smaller ranges and consider to only continue with the lower ranges. However sieveing is still more efficient when it comes to a combined sieving effort, and maybe PrimeGrid or someone else will be interested in continuing the sieved file once I abandon it. However, I plan to test this k to a pretty high n range, so I'm undecided for now, as to how to proceed beyond p=1T. Anyway, though I misunderstood, thanks for your replys and your inputs.

Regards

KEP

Reserve 4213
 

Reserving k=4213
Already tested to 350k, planning to test to 1.6M

- Chaos13

Chaos-
Welcome to the forum. Did you find any primes for 4213 in the 10k-350k range? If so, please post them in this thread.

We use this thread for primes too small to report to top-5000 (currently around 500k); bigger, reportable primes go in the "report primes here" thread.

-Curtis

Small primes found for 4213 from 10k to 350k:

4213*2^109575-1
4213*2^198903-1
4213*2^276399-1
4213*2^277667-1

Happy hunting!

- Chaos13

At last, I finished my range 430K-500K for k=7605, with two primes previously reported.

Cheers.

k=7605
 

Nomarcland

Thanks for your report. Here are all k=7605 known primes:

7, 9, 10, 13, 14, 19, 22, 25, 33, 39, 54, 62, 80, 89, 125, 127, 141, 144, 250, 295, 297, 314, 430, 475, 489, 680, 717, 766, 840, 922, 1090, 1201, 1210, 1213, 1342, 1420, 1560, 1941, 1997, 2065, 2085, 2101, 2314, 2369, 2790, 3065, 3715, 4600, 5469, 5961, 5986, 6295, 6994, 7194, 11120, 11950, 12219, 17994, 22895, 27754, 32782, 33625, 47609, 52194, 57697, 59549, 63087, 73563, 75164, 76139, 78245, 81089, 93082 [100k], ..., 353105, 358367, 438655, 454099, 524632, 530086, 595952 [333333-670k]

Tests are in progress, and I'll reach 700k in about a month.

Can I reserve k=12121?
I checked this k to n=1000 and found these primes:
[code]12121*2^29-1
12121*2^69-1
12121*2^119-1
12121*2^131-1
12121*2^309-1
12121*2^341-1
12121*2^353-1[/code]

Nash weight is 995.

Edit: Should I tell this to Low weight stats page?