mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Riesel Prime Search (https://www.mersenneforum.org/forumdisplay.php?f=59)
-   -   Post small primes and tell us about your progress here (https://www.mersenneforum.org/showthread.php?t=2150)

Happy5214 2021-02-02 09:52

[QUOTE=Happy5214;570625]Extracting it actually locked up my desktop (ran out of memory), forcing me to reboot, and killed my mprime P-1 job on my new 64 GB RAM laptop (same reason), so I had to make the necessary trimming in the standard Unix line editor ed (which still took up over 6 GB of RAM even with no visual display).[/QUOTE]

[QUOTE=storm5510;570660]I wouldn't think you would be having problems with 64GB or RAM. I can open any of the archives and view the details with 16GB of RAM on my Windows 10 machine. You mention running a server of some type. If this is on the same system, then it may be using up a lot of RAM.[/QUOTE]

I guess I should clarify. The PRPNet servers (most of the RAM hit is actually the MySQL server) run on the desktop, which is 11 years old and only has 8 GB of RAM. I tried to open it using a graphical editor, and it locked the machine up. (i.e. It wasn't the extraction, it was opening the text file that froze it.) While I was waiting for it to respond (it never did, so I rebooted it), I re-downloaded the file to the laptop (with 64 GB of RAM), extracted it, and tried to open it in the graphical editor (Kate, to be specific, since these are Kubuntu systems). It ballooned to about 45 GB of RAM usage, killing the mprime instance running a P-1 job (which was allocated 32 GB) in the process, before I killed the editor and ended up using ed. I didn't actually have a copy of srfile on that laptop (I just set it up a week ago and haven't gotten around to copying all of the important executables), so I figured it was an excuse to use ed (which I have an odd obsession with; I'm on their mailing list and maintain a GitHub mirror of the code, despite only having a basic proficiency with it).

Long story short, please do use srfile.

storm5510 2021-02-02 15:40

[QUOTE=kar_bon;570704]To give more hints:
- the link Happy given is the thread, so you have to go to the last post in it from 2021-01-19 (although the suspended sieves didn't changed since then in the first post)
- next download rsp1M_20150702.abcd (suspended sieves): this file contains [B]all[/B] sieves for Riesel 3<k<10000 and n<1M
- copy in a directory, unzip it and use srfile.exe to call:
srfile -g rsp1M_20150702.abcd
(output looks: Read 145955447 terms for 4998 sequences from ABCD format file `rsp1M_20150702.abcd'.)
to create every sieve file for all k's separately
- grap the t17_b2_k6927.npg to use this as input for LLR
it contains 23784 n-values <1M for k=6927

For higher n-values use the "rsp2M_xxx", "rsp3M_xxx" and so on.[/QUOTE]

Excellent! Thank you so much. :smile:

In the wiki, [I]k[/I]=6927 shows [I]n[/I]=400e3 in the "max [I]n[/I]" column. The largest value in the "primes for[I] n[/I]" column is 743,481. Removing everything from the sieve < this value leaves 6,077 terms. This is where I will start. If this is not correct, then let me know.

storm5510 2021-02-03 01:56

[QUOTE=storm5510;570732].[COLOR=Gray].In the wiki, [I]k[/I]=6927 shows [I]n[/I]=400e3 in the "max [I]n[/I]" column. The largest value in the "primes for[I] n[/I]" column is 743,481. Removing everything from the sieve < this value leaves 6,077 terms. This is where I will start. If this is not correct, then let me know.[/COLOR][/QUOTE]

[U]A correction[/U]: I went back to the "max [I]n[/I]" in the wiki, 400e3, and started there.

manrus 2021-03-29 10:29

Does it mean if I sieve k=3317 n=4e6-43e5 up to 10e12 then primegrid sieve files aren't needed?

gd_barnes 2021-03-29 20:37

[QUOTE=manrus;574716]Does it mean if I sieve k=3317 n=4e6-43e5 up to 10e12 then primegrid sieve files aren't needed?[/QUOTE]


You should use the PrimeGrid sieve file for testing. All k<10000 for n=1 to 3M are sieved to P=100P (100e15). The file will save you a lot of time in sieving and testing.

manrus 2021-03-30 07:21

[QUOTE=gd_barnes;574762]You should use the PrimeGrid sieve file for testing. All k<10000 for n=1 to 3M are sieved to P=100P (100e15). The file will save you a lot of time in sieving and testing.[/QUOTE]

So k=3317 n>= 4M sieved up to 400e15?

I opened rsp5M_20150702.abcd and see "..// Sieved to 1000000000 with srsieve". What it is mean?

gd_barnes 2021-03-30 07:30

[QUOTE=manrus;574775]So k=3317 n>= 4M sieved up to 400e15?

I opened rsp5M_20150702.abcd and see "..// Sieved to 1000000000 with srsieve". What it is mean?[/QUOTE]

rsp4M = 3M to 4M
rsp5M = 4M to 5M
rsp6M = 5M to 6M

It is n=3M to 6M that are sieved to 400e15.
(n=0 to 3M are sieved to 100e15.)

Ignore the header line in the sieve file. It means nothing. Pay attention to what it says in that post. You can tell by the fewer number of tests in each range than you would get by sieving yourself.

manrus 2021-03-30 07:50

[B]gd_barnes[/B], Thank you a lot


All times are UTC. The time now is 04:00.

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