mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Riesel Prime Search (https://www.mersenneforum.org/forumdisplay.php?f=59)
-   -   The Third Megabit Drive (https://www.mersenneforum.org/showthread.php?t=19398)

pinhodecarlos 2014-05-30 23:10

The Third Megabit Drive
 
1 Attachment(s)
We are starting the Third Megabit Drive.
The k's on this current drive, as of 30 May 2014, are at level n=1.4M therefore the goal of this drive is to take them up to 2M. Goal can be prior extended.

The k's of this drive are:

203, 205, 219, 227, 235, 257, 259

[B]Update (June 03, 2014):[/B] adding k=275 and k=295 to the drive from n=1680k. Range n=1400-1680k for both k added as "batch 2".

[B]Update (June 08, 2014):[/B] batch 2 completed.

[B]Update (July 09, 2014):[/B] The drive is completed to n=2M.

[B] Update (June 16, 2015):[/B] Reopening the drive from n=2M.

Please help yourself and happy hunting! :smile:

[B]Status[/B]

Batch 1 (7 k's to n=1680k and all 9 k's from n=1680k)
[CODE]
Range of n Tested by Status
1,400,000-1,404,000 - Carlos - Complete
1,404,000-1,408,000 - Trilo - Complete
1,408,000-1,412,000 - Thomas - Complete
1,412,000-1,418,000 - Carlos - Complete
1,418,000-1,450,000 - kracker - Complete
1,450,000-1,720,000 - Batalov - Complete (5 primes)
1,720,000-1,728,000 - Carlos - Complete
1,728,000-1,740,000 - kracker - Complete
1,740,000-1,750,000 - Carlos - Complete
1,750,000-1,800,000 - kracker - Complete
1,800,000-1,820,000 - Carlos - Complete
1,820,000-1,834,000 - Batalov - Complete (2 primes)
1,834,000-1,838,000 - Carlos - Complete
1,838,000-1,854,000 - Batalov - Complete (1 prime)
1,854,000-1,866,000 - Carlos - Complete
1,866,000-2,000,000 - Batalov - Complete
2,000,000-2,003,175 - Carlos - Complete
2,003,176-2,025,000 - Thomas11 - Complete
2,025,000-2,027,970 - Carlos - Complete
2,027,971-2,050,000 - Thomas11 - Complete
2,050,000-2,050,963 - Carlos - Complete
2,050,963-2,060,000 - Curtis - Complete
2,060,000-2,102,000 - Kosmaj - Complete
2,102,000-2,200,000 - gd_barnes - Complete (3 primes)
[/CODE]Batch 2 (k=275 and 295 from n=1400k to n=1680k)
[CODE]
Range of n Tested by Status
1,400,000-1,420,000 - Carlos - Complete
1,420,000-1,680,000 - Batalov - Complete
[/CODE][B]Test Files[/B]
Extensively sieved by Psieve (the latest sieve file of July 31, 2011)

See attachment with candidates from n=2M to 3M (split into 2k blocks, kindly prepared by Carlos).

[B]FFT lengths[/B]
Up to about n=1820k the 112k FFT will be used. AVX machines will benefit from the intermediate 120k FFT in the n=1850-1960k interval, where non-AVX machines will use the 128k FFT instead.

PS (Please attach zipped results naming like TMD_1400to1405.txt.zip and so on)


[B]Updated 2021-05-04[/B]

Found Primes:
203 [0]:
205 [2]: 1537779, 1666435
219 [1]: 2178673
227 [0]:
235 [3]: 1718787, 1824515, 2163273
257 [2]: 1658254, 2108554
259 [1]: 1695723
275 [1]: 1846390
295 [1]: 1832129

Total: 11 primes
(Found primes for 7 out of 9 k's)
.

Kosmaj 2014-05-31 01:16

Hi Carlos

Thanks for starting the drive. I'm busy right now but will join the drive soon.
Happy hunting! :smile:

pinhodecarlos 2014-05-31 01:40

Taking 1400-1404.

Edit: Fast machines can test a candidate at n=1.4M in 500 seconds or less. Come on guys!

pinhodecarlos 2014-05-31 02:33

Sorry for not splitting the input files into chunks but I don't know how to do it automatically. Also as the numbers are very fast to process, on latest Intel processors they can take less than 500 secs, I don't know what is the optimal chunk size for now for this drive, if 2k or 4k or 6k, for a period of 7-10 days of crunching as usual done on the other drives.

Carlos

Trilo 2014-06-01 04:01

Taking 1404 to 1408

Trilo 2014-06-01 04:03

About automatically splitting the files
 
I may be able to help you out by automatically splitting the files if you want. Do you need them in splits of 1000 or 2000?

Thomas11 2014-06-01 09:04

I splitted the 1.4-2M input file into 2k blocks (see attachment). Each block contains about 400 test candidates.

The attached ZIP file also contains a file called "url_list.txt" indicating the FFT lengths and percentages (for non-AVX machines). This file is only relevant for Kosmaj when uploading the files to his server...

[I]Edit:[/I] Attachment removed. Please use the file in the top post of this thread...

Thomas11 2014-06-01 09:05

Taking 1408-1412.

pinhodecarlos 2014-06-01 12:17

Taking 1412-1418.

kracker 2014-06-01 14:16

Taking 1418-1430.

kracker 2014-06-02 01:33

Hmm... going though them faster than expected, 2k done. Taking 1430-1450. (separate reservation if possible please)

pinhodecarlos 2014-06-02 01:43

Yep, they run fast. My first reservation is 60 % complete and I think from now on until we reach next FFT jump I will start reserving a 10k block per core.

Carlos

Trilo 2014-06-02 02:31

I'm running it at 2600 sec a candidate with 4 cores, my fast computer is working on the 8th drive. I'll have my range done in the next 6 days or so.

Batalov 2014-06-02 09:06

Taking 1450-1600.

pinhodecarlos 2014-06-02 09:28

[QUOTE=Batalov;374843]Taking 1430-1600.[/QUOTE]

kracker reserved from 1430-1450.

Batalov 2014-06-02 09:30

Ok, then.

Now you know that it is a full-time job to be a "manager" (or a "make-surerer", take your pick) of a drive. Please be fast and update the reservation post fast so that the users are not confused. You have been in the shoes of a user for too long. Now, learn to be an effective manager. ;-)

pinhodecarlos 2014-06-02 09:34

[QUOTE=Batalov;374845]Ok, then[/QUOTE]

Now you just need to update first post.
kracker reservation was in the middle of a conversation and therefore probably unnoticeable.


[QUOTE=Batalov;374845]Now you know that it is a full-time job to be a "manager" (or a "make-surerer", take your pick) of a drive. Please be fast and update the reservation post fast so that the users are not confused. You have been in the shoes of a user for too long. Now, learn to be an effective manager. ;-)[/QUOTE]

It is not my fault...I am a simple cruncher not moderator.

Batalov 2014-06-02 09:38

[QUOTE=pinhodecarlos;374846]I am a simple cruncher not moderator.[/QUOTE]
Talk about being confused.

Why do you start a drive that you have no intention of managing, then?

pinhodecarlos 2014-06-02 09:40

[QUOTE=Batalov;374847]Talk about being confused.

Why do you start a drive that you have no intention of managing, then?[/QUOTE]

As many started here by others. Kosmaj gave me permission to start it. Also the current three moderators can handle the work although I think it misses one from side of America's (north and south of America). One is in the UK, other in Japan and the last one in Germany.

Batalov 2014-06-03 09:13

1450-1600 is done.
Carrying on to 1600-1680.
Results will be in one file.

pinhodecarlos 2014-06-03 12:17

Thank you Batalov for the push.
Meanwhile here are the Nash Weight of the k's:

k=203 - NW=1687
k=205 - NW=1351
k=219 - NW=1069
k=227 - NW=2550
k=235 - NW=1241
k=257 - NW=1867
k=259 - NW=1561

Carlos

EDIT: Damn, I didn't see the k=275 still in n=1.4M. It should be on this drive too.

Trilo 2014-06-03 13:37

[QUOTE=pinhodecarlos;374933]Thank you Batalov for the push.
Meanwhile here are the Nash Weight of the k's:

k=203 - NW=1687
k=205 - NW=1351
k=219 - NW=1069
k=227 - NW=2550
k=235 - NW=1241
k=257 - NW=1867
k=259 - NW=1561

Carlos

EDIT: Damn, I didn't see the k=275 still in n=1.4M. It should be on this drive too.[/QUOTE]

I can run 275 from 1.4M to 1.6M then we can enter it into the drive past there.

pinhodecarlos 2014-06-03 13:51

[QUOTE=Trilo;374936]I can run 275 from 1.4M to 1.6M then we can enter it into the drive past there.[/QUOTE]

The files slitted in 2k blocks were already prepared so I think it is better to run it alone until 2M. How many cores do you have? Do you have the file?

Trilo 2014-06-03 14:08

I will have 4 cores available once my reservation on 1404-1408 is finished. I can get the files from primegrid's sieving files.

Trilo 2014-06-03 14:14

If you are not going to add it to the drive then I won't test it, but if you want to add it to the drive then I will test it.

pinhodecarlos 2014-06-03 14:16

Let's wait for Kosmaj or Thomas reply.

Trilo 2014-06-03 14:18

[QUOTE=pinhodecarlos;374941]Let's wait for Kosmaj or Thomas reply.[/QUOTE]

That's fine. 275 has a weight of 1581 so it won't take too long to test.

Thomas11 2014-06-03 15:39

[QUOTE=pinhodecarlos;374941]Let's wait for Kosmaj or Thomas reply.[/QUOTE]

I've added k=275 to the input files starting from n=1680k.

The ZIP file in the first post of this thread has been replaced and now contains the whole file (rps_M3.txt, n=1680-2000k) as well as the split files (2k blocks).
There is also the file for k=275 included (t17_b2_k275.npg, n=1000-2000k) for individual testing the n=1400-1680k range.

pinhodecarlos 2014-06-03 15:44

Thomas, k=295 is also stopped at 1.4M. Let's add this one too?

Thomas11 2014-06-03 16:08

[QUOTE=pinhodecarlos;374951]Thomas, k=295 is also stopped at 1.4M. Let's add this one too?[/QUOTE]

Okay, it's been added too.
The ranges n=1400-1680k for both k (275 and 295) are open for individual testing. Please post your reservations and results to this thread.

Batalov 2014-06-03 16:33

1 Attachment(s)
1450-1680 is done, 3 primes.

pinhodecarlos 2014-06-03 19:36

1 Attachment(s)
1402-1404 complete, no primes. (7 k's)

pinhodecarlos 2014-06-03 19:39

Taking 275 and 295 from 1400 to 1420.

pinhodecarlos 2014-06-03 19:53

1 Attachment(s)
1400-1402 complete, no primes. (7' k's)

Batalov 2014-06-04 00:54

Ok, I'll catch up the batch 2 from 1.42 to 1.68.

...and taking 1.68-1.72 (all k)

kracker 2014-06-04 15:01

1 Attachment(s)
1418-1430 no primes.

Batalov 2014-06-04 18:50

1 Attachment(s)
[QUOTE=Batalov;374995]Ok, I'll catch up the batch 2 from 1.42 to 1.68.

...and taking 1.68-1.72 (all k)[/QUOTE]
Both done and attached. 2 primes.

Trilo 2014-06-04 21:00

Now that the 2 extra k's have been all caught up, I think we should remove the "batch 2" part of the reservations to avoid confusion when Carlos is done with his range.

pinhodecarlos 2014-06-04 21:01

[QUOTE=Trilo;375073]Now that the 2 extra k's have been all caught up, I think we should remove the "batch 2" part of the reservations to avoid confusion.[/QUOTE]

Not yet, I am still doing a little part. More 4 days.

Trilo 2014-06-04 21:03

[QUOTE=pinhodecarlos;375074]Not yet, I am still doing a little part. More 4 days.[/QUOTE]

Yeah just realized you were not done with your reservation and edited my post right as you said this :max:

pinhodecarlos 2014-06-04 22:01

My reservation is 25 % complete. The server has 6 cores available that I could use but I don't know if I should. There's no queue work system, work is inserted manually.

pinhodecarlos 2014-06-05 16:36

1 Attachment(s)
1412-1414 complete, no primes. (7' k's)

pinhodecarlos 2014-06-07 01:30

Taking 1720-1728. (9 k's)

kracker 2014-06-07 21:06

1 Attachment(s)
1430 to 1450, no primes.

pinhodecarlos 2014-06-08 06:04

1 Attachment(s)
1400-1420 complete, no primes. (2 k's - batch 2)
Maybe clean on first post in order?!

pinhodecarlos 2014-06-08 12:15

I was wondering if it isn't better to switch the order of the status progress by putting the batch 2 information in first. Thoughts please?

kracker 2014-06-08 14:54

taking 1728 to 1740

pinhodecarlos 2014-06-08 19:54

1 Attachment(s)
1414-1416 complete, no primes. (7 k's)

Trilo 2014-06-09 15:54

1404-1408 complete, no primes

Trilo 2014-06-09 15:55

[QUOTE=pinhodecarlos;375348]I was wondering if it isn't better to switch the order of the status progress by putting the batch 2 information in first. Thoughts please?[/QUOTE]

Now that batch 2 is complete, I say we remove it from the list entirely.

pinhodecarlos 2014-06-09 16:34

[QUOTE=Trilo;375421]1404-1408 complete, no primes[/QUOTE]

Please attach the results.

pinhodecarlos 2014-06-09 18:07

Taking 1740-1750.

Thomas11 2014-06-10 14:04

1 Attachment(s)
1408-1412 complete, no primes.

kracker 2014-06-11 15:03

1 Attachment(s)
1728to1740 no primes.

kracker 2014-06-11 15:04

taking 1750 to 1800.

pinhodecarlos 2014-06-12 01:01

Taking 1800-1804.

pinhodecarlos 2014-06-12 09:31

1 Attachment(s)
1416-1418 complete, no primes (7 k's)

Trilo 2014-06-14 03:02

1 Attachment(s)
[QUOTE=pinhodecarlos;375431]Please attach the results.[/QUOTE]

I split it into a few separate files so I have attached them.

pinhodecarlos 2014-06-14 14:31

Taking 1804-1806.

pinhodecarlos 2014-06-15 10:55

1 Attachment(s)
1720-1722 complete, no primes.

pinhodecarlos 2014-06-15 21:28

Taking 1806-1810.

pinhodecarlos 2014-06-16 00:26

Candidates left from 1810-2000: 46135.

pinhodecarlos 2014-06-18 01:32

Taking 1810-1814.

pinhodecarlos 2014-06-18 01:40

Taking 1814-1854.
All separate reservations.

pinhodecarlos 2014-06-18 04:24

1 Attachment(s)
1804-1806 complete, no primes.

pinhodecarlos 2014-06-19 01:56

1 Attachment(s)
1800-1802 complete, no primes.

pinhodecarlos 2014-06-21 12:08

Taking 1854-1856.

pinhodecarlos 2014-06-22 00:00

1 Attachment(s)
1810-1812 complete, no primes.
Taking 1856-1858.

pinhodecarlos 2014-06-22 11:55

1 Attachment(s)
1722-1728 complete, no primes.

pinhodecarlos 2014-06-22 21:22

1 Attachment(s)
1806-1808 complete, no primes.

pinhodecarlos 2014-06-24 12:56

1 Attachment(s)
1812-1814 complete, no primes.

pinhodecarlos 2014-06-24 12:57

Taking 1858-1860.

pinhodecarlos 2014-06-25 15:49

1 Attachment(s)
1740-1750 complete, no primes.

pinhodecarlos 2014-06-25 15:51

Taking 1860-1862.

pinhodecarlos 2014-06-25 15:57

1 Attachment(s)
1802-1804 complete, no primes.

pinhodecarlos 2014-06-27 10:38

1 Attachment(s)
1814-1816 complete, no primes.

pinhodecarlos 2014-06-29 01:46

1 Attachment(s)
[QUOTE=pinhodecarlos;376351]Taking 1854-1856.[/QUOTE]

Done. No primes.

pinhodecarlos 2014-06-29 05:24

1 Attachment(s)
1856-1858 complete, no primes.

pinhodecarlos 2014-06-29 10:15

1 Attachment(s)
1808-1810 complete, no primes.
So 1740-1750 is done as well as 1800-1816.

kracker 2014-06-30 00:17

1 Attachment(s)
1750to1800 no primes.

pinhodecarlos 2014-07-01 12:02

Taking 1862-1864 and 1864-1866. (4 cores)

pinhodecarlos 2014-07-02 10:04

1 Attachment(s)
1858-1860 complete, no primes.

pinhodecarlos 2014-07-02 10:38

1 Attachment(s)
1860-1862 complete, no primes.
Means:

[code]
1,800,000-1,816,000 - Carlos - Complete
1,816,000-1,854,000 - Carlos - In Progress
1,854,000-1,856,000 - Carlos - Complete
1,856,000-1,862,000 - Carlos - Complete
1,862,000-1,864,000 - Carlos - In Progress
1,864,000-1,866,000 - Carlos - In Progress
[/code]

pinhodecarlos 2014-07-03 10:38

1 Attachment(s)
1816-1818 complete, no primes.

Batalov 2014-07-05 01:03

Taking 1.866-2M

pinhodecarlos 2014-07-05 01:30

1 Attachment(s)
[B]Update (July 04, 2014):[/B] extending drive to 3M.

241689 candidates in the attachment extensively sieved by Psieve (the latest sieve file of July 31, 2011)

pinhodecarlos 2014-07-05 01:35

Batalov,

My range 1818-1854 is in progress like this:

1818-1820 in progress
1820-1834 in queue
1834-1838 complete
1838-1854 in queue.
1862-1866 almost complete.

Do you want to crunch the ranges in queue? If so please let me know so I can add work from Second Megabit Drive, where I want to concentrate from now on my efforts.

Carlos

Batalov 2014-07-05 01:39

No, let's keep it simple, as is.

pinhodecarlos 2014-07-05 01:44

[QUOTE=Batalov;377412]No, let's keep it simple, as is.[/QUOTE]

Thank you and happy crunching!

My ranges below 1854 will take at least one month, I am now down to 2 cores at home and so far 10 at work. 1862-1866 will be done within max 2 days, so of the 10 cores more 4 will be down soon, leaving 6 to finish my current SMD reservations. By Monday all 10 cores will complete their jobs.

Kosmaj 2014-07-05 02:21

Hi Carlsos,

Please consider releasing your second queue (1838-1854), thus speeding up the processing of the first one and giving a chance to others to complete the range of the second queue and the whole Drive to 2M faster.

Thanks.

pinhodecarlos 2014-07-05 02:26

[QUOTE=Kosmaj;377419]Hi Carlsos,

Please consider releasing your second queue (1838-1854), thus speeding up the processing of the first one and giving a chance to others to complete the range of the second queue and the whole Drive to 2M faster.

Thanks.[/QUOTE]

I prefer to release both queues.

So the board will be look like this

[code]
1,750,000-1,800,000 - kracker - Complete
1,800,000-1,818,000 - Carlos - Complete
1,818,000-1,820,000 - Carlos - In progress
[B]1,820,000-1,834,000 - Available - Free[/B]
1,834,000-1,838,000 - Carlos - In progress
[B]1,838,000-1,854,000 - Available - Free[/B]
1,854,000-1,862,000 - Carlos - Complete
1,862,000-1,864,000 - Carlos - In Progress
1,864,000-1,866,000 - Carlos - In Progress
1,866,000-2,000,000 - Batalov - In Progress
[/code]I need an answer, almost midnight here, I want to sleep but before I go I need to reserve a SMD range to one of the cores.

pinhodecarlos 2014-07-05 02:47

Releasing the ranges in bold from previous post. I think it will be quicker if done by others rather by only two of my laptop cores. Tomorrow I will publish 1834-1838 results.

Going now to bed. Bye.

Carlos

Kosmaj 2014-07-05 04:10

Thanks!

I just uploaded remaining test files to my server, and added links to them in the first post of the thread.

pinhodecarlos 2014-07-05 06:03

1 Attachment(s)
1834-1838 complete, no primes.

pinhodecarlos 2014-07-05 06:06

1 Attachment(s)
1862-1864 complete, no primes.

pinhodecarlos 2014-07-05 11:39

I am going to dedicate one core here one core on SMD.
I ask for some help to help to clean the drive up to 2M. On my laptop each candidate is tested in less than 1200 secs so I suppose with processors with AVX 2.0 it will crunch in less than 900 secs.
Within 2 days I will reserve another 2k chunk.
Meanwhile 1864-1865 is complete, waiting for 1865-1866 (5 candidates left), and another core down for me.

Carlos

pinhodecarlos 2014-07-05 13:51

1 Attachment(s)
1864-1866 complete, no primes.


All times are UTC. The time now is 15:37.

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