new sieving program?

[Keroberts1]

Is there any update of the "Super Siever" or the new version of proth sieve that will be able to go past 2^52? Just curious because we are making signifigant progress towards the limit of the current siever.

[engracio]

Yes any update? My plan is to have the latest reservation to be my last sieve range and start moving the boxes to prp. It was nice while it lasted. Thanks.

e:)

[Nuri]

Thanks for your contribution engracio...



Hope to see you again with the super sieve client.. ;)

[engracio]

Nuri,

Not really bailing out of the project. Just happen to focused on sieving for a while cuz it needed a booze. Now its prp's turn for a while. Hopefully the "super siever" gets here soon. I think I'm good for another Prime or so.


e:)

[jasong]

Work is still going on with the super-siever. According to the update over at Riesel Sieve, it will be able to go all the way up to 1000T. I'm not sure if that just applies to Riesel Sieve or not, I think it applies to both.

It wasn't much of an update, and he didn't comment on speed this time, but there are probably going to be some new features in the super-siever, including the ability for the program to self-edit the unsieved values file.

[Joe O]

In the mean time, you might want to check out these executables. They are based on Prothsieve, but should be faster. Consider these as BETA.

JJsieve.exe is for all Intel Pentium Processors or AMD processors. It does not rely on CMOV, SSE, or SSE2.
Others will follow.


Updated to version 0.03D Now handles Riesel.dat

[b2uc]

Hey Joe,

Running this sieve in Riesel mode I get 'An unhandled win32 exception occured in JJsieve.exe [1428] after about 8 seconds of run time. At first I thought it was because I was continuing a range from proth_sieve and the rieselstatus.dat file was screwing with it...but I moved just the dat and the program to another folder and I still get the same error. Proth_sieve isn't running. AMD-2400+ with 2GB DDR and WinXP PRo.

Last screen output is:

k: 129007 a: 16 (16 * 1) : 10 diff. a/diff: 1.6 factor: 1.6

sieve range was 116872 to 116874

Thanks Joe,

Lee Stephens
Riesel Sieve

[em99010pepe]

I get that too.

Carlos

[b2uc]

A little more info from visual studios debugger:

Unhandled exception at 0x0040e30b in JJsieve.exe: 0xC0000094: Integer division by zero.

The thread 'Win32 Thread' (0xcc8) has exited with code -1073741676 (0xc0000094).
The program '[1428] JJsieve.exe: Native' has exited with code -1073741676 (0xc0000094).

[b2uc]

Joe, After chatting with someone behind the scenes.....can you check my riesel.dat and see if it is set up in a non-compatible way with your jjsieve? I use the +1 n incrementation method and my lines are in the unix format instead of Dos.

[Keroberts1]

is there a new copy of the .dat for 991-50,000,000? just wondering beause the copy i have is from june and the new program takes up alot more memory when running.

[Joe O]

Originally posted by Keroberts1
is there a new copy of the .dat for 991-50,000,000? just wondering beause the copy i have is from june and the new program takes up alot more memory when running.
A new copy of the dat is being prepared. I am waiting on two people to submit their completed ranges.

[Joe O]

Originally posted by b2uc
Hey Joe,

Running this sieve in Riesel mode I get 'An unhandled win32 exception occured in JJsieve.exe [1428] after about 8 seconds of run time. At first I thought it was because I was continuing a range from proth_sieve and the rieselstatus.dat file was screwing with it...but I moved just the dat and the program to another folder and I still get the same error. Proth_sieve isn't running. AMD-2400+ with 2GB DDR and WinXP PRo.

Last screen output is:

k: 129007 a: 16 (16 * 1) : 10 diff. a/diff: 1.6 factor: 1.6

sieve range was 116872 to 116874

Thanks Joe,

Lee Stephens
Riesel Sieve

That's why I called it beta. I've been testing it, for the most part only with Sierpinski sieve data. I will have to test all the places where the Riesel code is different from the Sierpinski code. But it worked prior to the last set of optimizations. It looks like I have to enlarge an array to handle the larger number of k's in the Riesel dat.

EDIT: I have a new version that should handle Riesel sieve, Version 0.03C

[Joe O]

Originally posted by b2uc
Joe, After chatting with someone behind the scenes.....can you check my riesel.dat and see if it is set up in a non-compatible way with your jjsieve? I use the +1 n incrementation method and my lines are in the unix format instead of Dos.

I have the 24.6 MB file last modified June 8, 2005. If there is another one you want me to use, please send it to me at factrange at yahoo dot com. I have written the input routines to handle cr, lf, and crlf line terminations. They also have been written to handle n's and n, +diff file formats.

[Joe O]

In the mean time, you might want to check out these executables. They are based on Prothsieve, but should be faster. Consider these as BETA.

JJsieveCMOV6.exe is for Intel Pentium Pro, P II, and P III or AMD processors. It relies on CMOV and SSE, but not on SSE2.
Others will follow

Updated to version 0.03D Now handles Riesel.dat

[Joe O]

In the mean time, you might want to check out these executables. They are based on Prothsieve, but should be faster. Consider these as BETA.

JJsieveCMOV7.exe is for Intel Pentium IV or AMD processors. It relies on CMOV and SSE, but not on SSE2.
One more will follow.

Updated to version 0.03D Now handles Riesel.dat

[Joe O]

In the mean time, you might want to check out these executables. They are based on Prothsieve, but should be faster. Consider these as BETA.

JJsieveSSE2.exe is for Intel Pentium IV or AMD64 processors. It relies on CMOV, SSE, and SSE2.

Updated to version 0.03D Now handles Riesel.dat

[hhh]

Super Siever, does it mean that it can sieve Sierpinski and Riesel in the same time with a speed gain? I might be annoying, but in this case I would consider to sieve together...

(Kind of déjÃ_ vu...)

Yours H.

[Joe O]

Originally posted by hhh
Super Siever, does it mean that it can sieve Sierpinski and Riesel in the same time with a speed gain? I might be annoying, but in this case I would consider to sieve together...

(Kind of déjß vu...)

Yours H.

Not Yet!

[Joe O]

All four versions:
jjsieve.exe
jjsieveCMOV6.exe
jjsieveCMOV7.exe
jjsieveSSE2.exe

have been updated to version 0.03d

The only change was to an array dimension to handle one of the Riesel k's that prevented Riesel sievers from using the program.

[jasong]

Originally posted by hhh
Super Siever, does it mean that it can sieve Sierpinski and Riesel in the same time with a speed gain? I might be annoying, but in this case I would consider to sieve together...

(Kind of déj�_ vu...)

Yours H.
Two observations: It will probably be easier to convince Riesel Sievers to do this than than SoB people(Riesel people will see about a 15% increase in numbers, while SoB will see about a 600% increase in numbers. BIIIIIIGGGG difference)

2nd observation: If the overall points increase at a faster rate or, in the case of SoB, come more often, my first point might not matter.

I wouldn't mind the switch one bit, but other people might feel differently.

I'm a Riesel siever at the moment, by the way.

[hhh]

I got the impression that the version before the Riesel changes significantly slows down the working speed of my P4 3.06 GHz. Perhaps the priority is not set to idle automatically? BTW it is working at 173 instead of 169 for the joint effort.
EDIT indeed, the priority ws set to normal.


Is the new siever working beyond 2^50? If so, I think the easiest way would be to separate first an secondpass sieving for SoB, and to let do the secondpasssieving together with PSP and Riesel firstpass in a joint sieving project, when it is possible. There would be PSP and Riesel people, and from time to time SoB people helping out so that it is about fair.
I agree that combining the firstpass effort of SoB as well would be a little harsh to make understand.

Not yet does mean about when?

Amazed, Yours H.

[cedricvonck]

Any updates about the new sieve program?

Regards,
Cedric

[vjs]

I know Joe is away this week... does someone have to leave for a week to have this happen.

BTW where is MikeH anyways.

[cedricvonck]

Sorry didn't know that.

[MikeH]

BTW where is MikeH anyways. I'm still here. Just being a bit quiet. Taking a bit of a backseat. Watching the potential 4847 prime with much interest.

[vjs]

Darn it maybe there is no prime.

Isn't there some correlation between you being out of town and a prime occurance???

Perhaps Joe took your place on this one, I think he was MIA during prime n=9.18M as well.