mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > FermatSearch

Reply
 
Thread Tools
Old 2018-08-04, 19:58   #23
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

22×23×67 Posts
Default

Quote:
Originally Posted by ET_ View Post
Done

Would you mind spending 2 words explaining why adding N values helps deepen the search? Would extending the k range have the same result, or the opposite?
Adding n means at most one addition and one bit shift per n per prime. This is much faster than sieving two consecutive n separately as each n would require one powmod per prime. In this case I have replaced a powmod with an add and bit shift.

Adding k only increases the memory needed for sieving, but won't slow down the sieve once you get p > max k.
rogue is offline   Reply With Quote
Old 2018-08-13, 12:55   #24
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

22·23·67 Posts
Default

Both of these range are done. Nothing new to report beyond the Fermat divisor that I already reported.

n: 5100-5199 k:1200e6-2500e6
n: 5100-5299 k:1200e6-2500e6
rogue is offline   Reply With Quote
Old 2018-08-19, 21:54   #25
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

22·23·67 Posts
Default

I have to redo

n: 5000-5399 k:1200e6-2500e6

because of the bug gfndsieve had with ABCD files which I fixed with the 1.7.2 release of mtsieve in July. Although that bug only impacts starting gfndsieve from an previous save point, I suspect that the other large ranges I have sieved with it need to be re-run. n 5400-5599 are fine. There is another large gap I did a few months ago that I will re-run. I doubt any of the smaller ranges I have done are impacted, but they should be "thrown back into the pool" to be redone. Unfortunately for me this is about 6 months of rework.

If anyone used a build of gfndsieve that is prior than the 1.7.2 release of mtseive and you stopped and restarted sieving from a saved file, you will need to redo that range. If you did not stop and restart then you are fine.

Last fiddled with by rogue on 2018-08-19 at 22:03
rogue is offline   Reply With Quote
Old 2018-12-13, 16:52   #26
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

22·23·67 Posts
Default

I finished the recheck of n 5000-5399 and completed n 5400-5599 for k 1200e6-2500e6. Here are my finds:

2222488721*2^5349+1 is a Factor of xGF(5347,7,3)!!!!
2055442709*2^5431+1 is a Factor of GF(5429,11)!!!!
1994861771*2^5517+1 is a Factor of xGF(5512,7,2)!!!!
1321981497*2^5592+1 is a Factor of xGF(5590,7,5)!!!!

Nothing was missed in 5000-5299.

I have rested all other ranges that I had completed previously with the the exception of n: 10003-10999 k: 100e6 269e6. I expect that to complete in February. Based upon sieving, I know that range was impacted by the bug in gfndsieve.
rogue is offline   Reply With Quote
Old 2019-11-22, 10:14   #27
ET_
Banned
 
ET_'s Avatar
 
"Luigi"
Aug 2002
Team Italia

479010 Posts
Default Some new "Most wanted" ranges

Quote:
Originally Posted by ATH View Post
I admit I did some factoring without making a reservation but I did check the Reservation thread and here: http://www.fermatsearch.org/stat/running.php

I was running mmff-0.28 near the maximum range which is n<=223 and k*2n+1 <= 2252. I completed the maximum available ranges that mmff-0.28 can do:


Code:
no factor for k*2^205+1 in k range: 130000000000000 to 140737488355327 (252-bit factors) [mmff 0.28 mfaktc_barrett252_F192_223gs]

no factor for k*2^204+1 in k range: 130000000000000 to 140737488355327 (251-bit factors) [mmff 0.28 mfaktc_barrett252_F192_223gs]
no factor for k*2^204+1 in k range: 140737488355328 to 281474976710655 (252-bit factors) [mmff 0.28 mfaktc_barrett252_F192_223gs]

no factor for k*2^203+1 in k range: 130000000000000 to 140737488355327 (250-bit factors) [mmff 0.28 mfaktc_barrett252_F192_223gs]
no factor for k*2^203+1 in k range: 140737488355328 to 281474976710655 (251-bit factors) [mmff 0.28 mfaktc_barrett252_F192_223gs]
no factor for k*2^203+1 in k range: 281474976710656 to 562949953421311 (252-bit factors) [mmff 0.28 mfaktc_barrett252_F192_223gs]

no factor for k*2^202+1 in k range: 130000000000000 to 140737488355327 (249-bit factors) [mmff 0.28 mfaktc_barrett249_F192_223gs]
no factor for k*2^202+1 in k range: 140737488355328 to 281474976710655 (250-bit factors) [mmff 0.28 mfaktc_barrett252_F192_223gs]
no factor for k*2^202+1 in k range: 281474976710656 to 562949953421311 (251-bit factors) [mmff 0.28 mfaktc_barrett252_F192_223gs]
no factor for k*2^202+1 in k range: 562949953421312 to 1125899906842623 (252-bit factors) [mmff 0.28 mfaktc_barrett252_F192_223gs]
ATH did a terrific work, taking the N ranges between 202 and 205 to their maximum allowed k limits for the actual mmff software

Now I need some volunteering effort to close the gaps of the range 200-209 and take each N to the same level...
The request is addressed to everyone,and duplicated in the "Results" thread

I know Oleg Naryshkin has been working on the range 200-249 for a long time, and will advice him as well about the change of the k limits ofthese N. Anyway, any help on these single ranges is welcome.

Luigi

p.s. just asked Oleg if he plans to update the limits, and started N=205 from 140737488355327 to 150000000000000 and N=204 from 281474976710655 to 300000000000000 using Feromant_CUDA
pps: I am also working on 201, 202 and 203 to get a "more round" k limit.

Last fiddled with by ET_ on 2019-11-23 at 10:11
ET_ is offline   Reply With Quote
Old 2021-01-02, 16:18   #28
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

22×23×67 Posts
Default

Here is the current "wanted" list per my program. I noticed that some of these reservations are from 2019.

Code:
type 1 means that n was tested from k1 to k2, but k1 != 1
type 2 means that n was tested from k1 to k2 and k3 to k4, but not between k2 and k3
type 3 means that that some n in the group were not tested to the same k as others
            30 <= n <    100, each group has 1 n
           100 <= n <   1000, each group has 10 n from xx0 to xx9
          1000 <= n <  10000, each group has 100 n from xx00 to xx99
         10000 <= n < 100000, each group has 1000 n from xx000 to xx999
        100000 <= n are not included

type 2: n 40            500e14-510e14        12 days with mmff              Dylan Parsons 2020/04/29
type 2: n 41            800e14-900e14        129 days with mmff             Liam McGonegal 2020/08/18
type 2: n 150-159       1000e12-1100e12      70 days with FeromantCUDA      Alex_soldier 2020/10/06
type 3: n 205-209       190e12-1130e12       1207 days with FeromantCUDA    Roman Maznichenko 2020/12/12
type 3: n 203-204       610e12-1130e12       668 days with FeromantCUDA     available 
type 2: n 500-509       1000e10-2000e10      45 days with FeromantCUDA      Ryan Propper 2019/05/01
type 2: n 510-519       1000e10-2000e10      45 days with FeromantCUDA      Ryan Propper 2019/05/01
type 2: n 520-529       1000e10-2000e10      45 days with FeromantCUDA      Ryan Propper 2019/05/01
type 2: n 531-539       870e10-2000e10       50 days with FeromantCUDA      Ryan Propper 2019/04/25
type 2: n 530           1000e10-2000e10      45 days with FeromantCUDA      Ryan Propper 2019/04/25
type 2: n 540-549       870e10-2000e10       50 days with FeromantCUDA      Ryan Propper 2019/05/01
type 2: n 550-559       870e10-2000e10       50 days with FeromantCUDA      Ryan Propper 2019/05/01
type 2: n 560-569       870e10-2000e10       50 days with FeromantCUDA      Ryan Propper 2019/05/01
type 2: n 570-579       350e10-2000e10       73 days with FeromantCUDA      Peter Strasser 2019/04/23
type 2: n 580-589       350e10-2000e10       73 days with FeromantCUDA      Peter Strasser 2019/04/23
type 2: n 590-599       350e10-2000e10       73 days with FeromantCUDA      Peter Strasser 2019/04/23
type 2: n 601-609       250e10-300e10        210 days with Feromant         Pacher Maximilian 2020/04/21
type 3: n 600           250e10-400e10        630 days with Feromant         Gary Gostin 2020/12/23
type 2: n 610-619       250e10-300e10        210 days with Feromant         Pacher Maximilian 2020/04/21
type 2: n 620-629       250e10-300e10        210 days with Feromant         Pacher Maximilian 2020/04/21
type 2: n 630-639       250e10-300e10        210 days with Feromant         Pacher Maximilian 2020/04/21
type 2: n 660-669       200e10-220e10        84 days with Feromant          Roman Maznichenko 2020/12/12
type 2: n 670-679       200e10-220e10        84 days with Feromant          Roman Maznichenko 2020/12/12
type 2: n 680-689       200e10-220e10        84 days with Feromant          Roman Maznichenko 2020/12/12
type 2: n 750-759       100e10-120e10        84 days with Feromant          Roman Maznichenko 2020/12/12
type 2: n 760-769       100e10-120e10        84 days with Feromant          Roman Maznichenko 2020/12/12
type 2: n 770-779       100e10-120e10        84 days with Feromant          Roman Maznichenko 2020/12/12
type 2: n 780-789       100e10-120e10        84 days with Feromant          Roman Maznichenko 2020/12/12
type 2: n 790-799       100e10-120e10        84 days with Feromant          Roman Maznichenko 2020/12/12
type 2: n 4500-4599     7e8-10e8             209 days with Feromant         available 
type 2: n 4500          45e8-50e8            348 days with Feromant         Franz Fritz 2020/10/21
type 2: n 4501-4599     45e8-60e8            1043 days with Feromant        Franz Fritz 2020/10/21
type 3: n 4501-4599     65e8-75e8            695 days with Feromant         available 
type 2: n 4600-4699     7e8-10e8             209 days with Feromant         available 
type 2: n 4600-4699     45e8-60e8            1043 days with Feromant        Franz Fritz 2020/10/21
type 2: n 4700-4799     7e8-10e8             209 days with Feromant         available 
type 2: n 4700-4799     45e8-60e8            1043 days with Feromant        Franz Fritz 2020/10/21
type 2: n 4800-4899     7e8-10e8             209 days with Feromant         available 
type 2: n 4800-4899     45e8-60e8            1043 days with Feromant        Franz Fritz 2020/10/21
type 2: n 4900-4999     7e8-10e8             209 days with Feromant         available 
type 2: n 4900-4999     45e8-60e8            1043 days with Feromant        Franz Fritz 2020/10/21
type 2: n 5000          450e7-600e7          1737 days with gfndsieve+pfgw  available 
type 3: n 5001-5099     450e7-650e7          2315 days with gfndsieve+pfgw  available 
type 3: n 5501-5599     400e7-450e7          579 days with gfndsieve+pfgw   available 
type 3: n 5701-5799     300e7-400e7          1158 days with gfndsieve+pfgw  Ryan Propper 2019/07/10
type 3: n 6001-6099     300e7-400e7          1158 days with gfndsieve+pfgw  available 
type 3: n 6501-6599     200e7-300e7          1158 days with gfndsieve+pfgw  available 
type 2: n 7000-7099     80e7-100e7           232 days with gfndsieve+pfgw   Pacher Maximilian 2020/04/25
type 2: n 7000          120e7-150e7          348 days with gfndsieve+pfgw   available 
type 3: n 7001-7099     120e7-200e7          926 days with gfndsieve+pfgw   available 
type 2: n 7100-7199     80e7-100e7           232 days with gfndsieve+pfgw   Pacher Maximilian 2020/04/25
type 2: n 7200-7299     80e7-100e7           232 days with gfndsieve+pfgw   Pacher Maximilian 2020/04/25
type 2: n 7300-7399     80e7-100e7           232 days with gfndsieve+pfgw   Pacher Maximilian 2020/04/25
type 2: n 7400-7499     80e7-100e7           232 days with gfndsieve+pfgw   Pacher Maximilian 2020/04/25
type 2: n 7500-7599     80e7-100e7           232 days with gfndsieve+pfgw   Pacher Maximilian 2020/04/25
type 2: n 7600-7699     80e7-100e7           232 days with gfndsieve+pfgw   Pacher Maximilian 2020/04/25
type 2: n 7700-7799     80e7-100e7           232 days with gfndsieve+pfgw   Pacher Maximilian 2020/04/25
type 2: n 7800-7899     80e7-100e7           232 days with gfndsieve+pfgw   Pacher Maximilian 2020/04/25
type 2: n 7900-7999     80e7-100e7           232 days with gfndsieve+pfgw   Pacher Maximilian 2020/04/25
type 2: n 8000          80e7-100e7           232 days with gfndsieve+pfgw   Pacher Maximilian 2020/04/25
type 3: n 8001-8099     80e7-120e7           463 days with gfndsieve+pfgw   available 
type 3: n 9001-9099     60e7-80e7            232 days with gfndsieve+pfgw   available 
type 2: n 10000         420e6-450e6          695 days with gfndsieve+pfgw   available 
type 3: n 10001-10999   420e6-600e6          4167 days with gfndsieve+pfgw  available 
type 2: n 11000         250e6-300e6          1158 days with gfndsieve+pfgw  available 
type 3: n 11001-11999   250e6-420e6          3936 days with gfndsieve+pfgw  available 
type 2: n 12000         150e6-160e6          232 days with gfndsieve+pfgw   available 
type 3: n 12001-12999   150e6-250e6          2315 days with gfndsieve+pfgw  available 
type 2: n 14000-14999   70e6-100e6           695 days with gfndsieve+pfgw   Adolf Nordin 2019/05/30
type 3: n 14001-14999   120e6-150e6          695 days with gfndsieve+pfgw   available 
type 2: n 16000         60e6-70e6            232 days with gfndsieve+pfgw   available 
type 3: n 16001-16999   60e6-120e6           1389 days with gfndsieve+pfgw  available 
type 3: n 18001-18999   50e6-60e6            232 days with gfndsieve+pfgw   available 
type 3: n 20001-20999   40e6-50e6            724 days with gfndsieve+pfgw   available 
type 3: n 22001-22999   32e6-40e6            579 days with gfndsieve+pfgw   available 
type 3: n 23001-23999   20e6-32e6            869 days with gfndsieve+pfgw   available 
type 3: n 25001-25999   12e6-20e6            579 days with gfndsieve+pfgw   available 
type 3: n 30001-30999   8e6-12e6             1544 days with gfndsieve+pfgw  available 
type 3: n 35001-35999   4e6-8e6              1544 days with gfndsieve+pfgw  available 
type 3: n 40001-40999   300e4-400e4          386 days with gfndsieve+pfgw   available 
type 3: n 45001-45999   220e4-300e4          309 days with gfndsieve+pfgw   available 
type 3: n 50001-50999   180e4-220e4          4630 days with gfndsieve+pfgw  available 
type 3: n 55001-55999   70e4-180e4           12732 days with gfndsieve+pfgw Gary Gostin 2020/11/29
type 3: n 60001-60999   50e4-70e4            2315 days with gfndsieve+pfgw  Gary Gostin 2020/11/29
type 3: n 65001-65999   40e4-50e4            1158 days with gfndsieve+pfgw  Gary Gostin 2020/11/29
type 3: n 70001-70999   35e4-40e4            579 days with gfndsieve+pfgw   Gary Gostin 2020/11/29
type 3: n 75001-75999   30e4-35e4            579 days with gfndsieve+pfgw   Gary Gostin 2020/11/29
type 3: n 80001-80999   25e4-30e4            579 days with gfndsieve+pfgw   Dylan Delgado 2020/12/16
type 3: n 90001-90999   20e4-25e4            579 days with gfndsieve+pfgw   Dylan Delgado 2020/12/16
rogue is offline   Reply With Quote
Old 2021-01-02, 18:18   #29
ET_
Banned
 
ET_'s Avatar
 
"Luigi"
Aug 2002
Team Italia

2·5·479 Posts
Default

Thank you Mark, Time for a close call to the older assignments.
ET_ is offline   Reply With Quote
Old 2021-01-03, 04:52   #30
Gary
 
Gary's Avatar
 
"Gary"
Aug 2015
Texas

22·3·5 Posts
Default

Quote:
Originally Posted by rogue View Post
Here is the current "wanted" list per my program. I noticed that some of these reservations are from 2019.

Code:
type 1 means that n was tested from k1 to k2, but k1 != 1
type 2 means that n was tested from k1 to k2 and k3 to k4, but not between k2 and k3
type 3 means that that some n in the group were not tested to the same k as others
            30 <= n <    100, each group has 1 n
           100 <= n <   1000, each group has 10 n from xx0 to xx9
          1000 <= n <  10000, each group has 100 n from xx00 to xx99
         10000 <= n < 100000, each group has 1000 n from xx000 to xx999
        100000 <= n are not included

...
Luigi recently posted a request that search ranges which could discover GFN factors be of the form n = xxx0 - xxx0:

https://www.mersenneforum.org/showpo...69&postcount=9

This would apply to most ranges with n >= ~5000 where gfndsieve + pfgw is the fastest approach and are usually Fermat+GFN searches. This request is for consistency with the GFN ranges tracked at ProthSearch, and is the result of some discussion between Wilfrid, Luigi and myself. So far there has been no community feedback / debate on this request. Mark and others, what are your thoughts? Mark, this differs from the xx0 - xx9 convention that you have advocated in the past and is incorporated in your program.
Gary is offline   Reply With Quote
Old 2021-01-03, 14:16   #31
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

22·23·67 Posts
Default

Quote:
Originally Posted by Gary View Post
Luigi recently posted a request that search ranges which could discover GFN factors be of the form n = xxx0 - xxx0:

https://www.mersenneforum.org/showpo...69&postcount=9

This would apply to most ranges with n >= ~5000 where gfndsieve + pfgw is the fastest approach and are usually Fermat+GFN searches. This request is for consistency with the GFN ranges tracked at ProthSearch, and is the result of some discussion between Wilfrid, Luigi and myself. So far there has been no community feedback / debate on this request. Mark and others, what are your thoughts? Mark, this differs from the xx0 - xx9 convention that you have advocated in the past and is incorporated in your program.
At ProthSearch it used to be searches for a single k, but a range of n, so testing thru n=xx0 makes more sense, but FermatSearch does it differently. Are these projects overlapping the ranges they are searching? Maybe it makes sense for ProthSearch to search for large n (whatever value that is) or FermatSearch for small n.

IMO, it makes more sense to reserve a <= n < b where both a and b are xx0. This way neither a nor b are handled twice.
rogue is offline   Reply With Quote
Old 2021-01-03, 18:14   #32
Gary
 
Gary's Avatar
 
"Gary"
Aug 2015
Texas

22·3·5 Posts
Default

Quote:
Originally Posted by rogue View Post
At ProthSearch it used to be searches for a single k, but a range of n, so testing thru n=xx0 makes more sense, but FermatSearch does it differently. Are these projects overlapping the ranges they are searching? Maybe it makes sense for ProthSearch to search for large n (whatever value that is) or FermatSearch for small n.

IMO, it makes more sense to reserve a <= n < b where both a and b are xx0. This way neither a nor b are handled twice.
Most ProthSearch ranges are like FermatSearch ranges: a range of N * a range of K. There is a good summary table at http://www.prothsearch.com/GFNsrch.txt. I would call the search ranges of the two projects "overlapping but coordinated". A range which may find either a new Fermat or GFN factor (Fermat+GFN range) is reserved at FermatSearch. Once the range is done, it is recorded as complete at both FermatSearch and ProthSearch. This applies to most ranges with n >= ~5000 where the incremental time to search for GFN factors is small.

Separately, I have been doing some GFN-only searches with n <= 3000, where the K wavefront for GFN is much lower than the Fermat wavefront. For these I have been reserving ranges and reporting results only through ProthSearch. Currently I seem to be the only person active in these GFN-only ranges.

Wilfrid prefers tracking GFN ranges as xx1 - xx0 since that is consistent with the original GFN paper by Bjorn and Riesel. nf = xx0 is the most important of the two endpoints since it determines the "shape" of that table at http://www.prothsearch.com/GFNsrch.txt. So I have started using nf = xx0 for all my Fermat+GFN and GFN-only ranges. For ni on GFN-only ranges with small N, Wilfrid and I agree on using ni = xx1 to avoid overlap. For ni on Fermat+GFN ranges with larger N, I was also originally thinking ni = xx1. However Wilfrid pointed out that, historically, GFN search ranges have been a complex "patchwork" of N/K rectangles with different conventions for ni and nf. This makes it difficult to track ranges and ensure no gaps have been left, which I believe he does by hand. For Fermat+GFN ranges with large N, Wilfrid suggested using ni = xx0 and nf = xx0 since it is "safe" (will not leave a gap relative to a prior range with nf = xx9 below it or ni = xx1 above it) and the incremental runtime is small. For example, searching n = 5000 - 5500 instead of n = 5001 - 5500 adds only 0.2% to the runtime. This seems acceptably small to me, so I am now using ni = xx0 for all my Fermat+GFN ranges.
Gary is offline   Reply With Quote
Old 2021-01-04, 17:08   #33
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

140248 Posts
Default

The list I created is generated via the attached source code which I haven't changed in over 18 months. Since I have not actively participated in a long time, maybe someone else will be willing to make the necessary changes to support xx0. It should be fairly easy to do. This code will help identify gaps based upon the completed ranges reported at FermatSearch.
Attached Files
File Type: c flist.c (29.5 KB, 5 views)
rogue is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
New wanted ranges! ET_ FermatSearch 12 2016-11-09 13:31
P-1 formula (help wanted) Prime95 Data 18 2012-02-12 18:07
Most wanted kar_bon Riesel Prime Data Collecting (k*2^n-1) 15 2011-08-09 16:50
Wanted Lists R.D. Silverman Cunningham Tables 1 2010-09-21 16:16
100 Most Wanted Citrix Factoring 24 2004-02-22 01:05

All times are UTC. The time now is 08:11.

Mon Jan 18 08:11:43 UTC 2021 up 46 days, 4:23, 0 users, load averages: 2.25, 2.03, 2.05

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.