mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   XYYXF Project (https://www.mersenneforum.org/forumdisplay.php?f=110)
-   -   Leyland Primes (x^y+y^x primes) (https://www.mersenneforum.org/showthread.php?t=19347)

rogue 2014-07-11 17:44

1 Attachment(s)
Here is v1.0.3. I fixed a calculation error in the stats that I inadvertently added in the previous build. I also made the following changes:

1) Delete the .ckpt file when done.
2) Write to the output file whenever the .ckpt file is written.
3) Apply factors from xyyxsieve.log to the input file before processing it.

These were designed to make it safer to stop and restart without losing factors.

NorbSchneider 2014-07-12 20:03

I tested the program with command line:
xyyxsievecl64.exe -v --platform=1 -x4000 -X8020 -y8001 -Y8020 -P100000 -t8 -s2000 -S+ -oxyyx_8020p.out -f1 -F1

When i use -S+ then only the x^y+y^x numbers were sieved,
when -S- then x^y+y^x and x^y-y^x numbers were sieved,
when -Sb then error: Fatal Error: sign must be specified.

In the help file is: -S --sign=+/-/b sign to sieve for

I would like to use xyyxsievecl64.exe for only x^y-y^x numbers sieve.

rogue 2014-07-19 00:30

My double-check for all x/y has yielded this:

1467^4076+4076^1467 is 3-PRP!

The range containing this PRP was searched prior to the creation of MultiSieve.

I'm still double-checking the range where both x and y are <= 10000. I'm about 50% complete.

rogue 2014-07-19 00:34

1 Attachment(s)
[QUOTE=NorbSchneider;377975]I tested the program with command line:
xyyxsievecl64.exe -v --platform=1 -x4000 -X8020 -y8001 -Y8020 -P100000 -t8 -s2000 -S+ -oxyyx_8020p.out -f1 -F1

When i use -S+ then only the x^y+y^x numbers were sieved,
when -S- then x^y+y^x and x^y-y^x numbers were sieved,
when -Sb then error: Fatal Error: sign must be specified.

In the help file is: -S --sign=+/-/b sign to sieve for

I would like to use xyyxsievecl64.exe for only x^y-y^x numbers sieve.[/QUOTE]

Sorry that I missed your post. It is fixed in the attached. BTW you can use scientific notation for most parameters. In other words you can use -P1e5 instead of -P100000.

You're probably just testing, but I recommend using a larger range for y due to optimizations in the code. You might also want to increase -s and decrease -t to improve performance, but I'll leave that for you to figure out what works best on your machine.

NorbSchneider 2014-07-21 12:31

[QUOTE=rogue;378542]Sorry that I missed your post. It is fixed in the attached. BTW you can use scientific notation for most parameters. In other words you can use -P1e5 instead of -P100000.
[/QUOTE]

Thank You!
[SIZE=2]It works perfect. I can now sieve only XY+, only XY- PRPs and the two simultaneously.[/SIZE]

rogue 2014-07-24 12:59

[QUOTE=rogue;378541]My double-check for all x/y has yielded this:

1467^4076+4076^1467 is 3-PRP!

The range containing this PRP was searched prior to the creation of MultiSieve.

I'm still double-checking the range where both x and y are <= 10000. I'm about 50% complete.[/QUOTE]

I completed a double-check of the range. That was the only missing PRP. I'm continuing to double-check for x and y <= 12500. Unfortunately I didn't sieve as deeply as I would have liked because I took a Windows patch and now none of my OpenCL applications works on it. I have other computers, but most have a slower GPU. The one with the fastest GPU (a Mac) broke earlier this year when I installed Mavericks. I don't know what is going on with that, but I suspect that I need to re-install the OS to fix the problem and since it is my main computer I don't want to take it down for an extended period of time to fix the problem. Eventually I will or I'll talk my wife into an upgrade (of the computer, not her :grin:).

rogue 2014-07-27 21:21

Here are two more PRPS that were missed. These were probably missed due to the bug in MultiSieve.

1490^11487+11487^1490
2239^12462+12462^2239

I've finished the double check where y <= 2500 and x <= 12500. I have a ways to go before finishing the rest of y <= 12500.

Batalov 2014-07-27 21:31

Intriguingly, the second one is in factordb since July 11, 2014

P.S. EDIT: And in prptop (Norbert Schneider, 07/2014)

rogue 2014-07-28 02:18

[QUOTE=Batalov;379175]Intriguingly, the second one is in factordb since July 11, 2014

P.S. EDIT: And in prptop (Norbert Schneider, 07/2014)[/QUOTE]

I wonder why they are missing on xyyxf's website.

NorbSchneider 2014-07-28 06:22

I'm Norbert Schneider.

On July 26, 2014 have I finished my search in the interval
12,051<=x<=12,500, 2001<=y<=x-1.

My last new PRPs have i sent to Andrey Kulsha, but he has
so far the website not updated.

The last PRPs:
6330^12421+12421^6330, 47218 digits,
4569^12448+12448^4569, 45558 digits,
8602^12451+12451^8602, 48990 digits,
2239^12462+12462^2239, 41749 digits.

XYYXF 2014-08-01 23:11

[QUOTE=NorbSchneider;379199]but he has so far the website not updated.[/QUOTE]I'm waiting for two C168 GNFS tasks, just a few days :)


All times are UTC. The time now is 07:34.

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