GPU polynomial selection, extreme edition - Page 6

banxian · 2012-11-18, 11:42

svn800.zip link broken now

jasonp · 2012-11-19, 02:36

I've corrected the link in the OP to point to the current latest.

skan · 2012-12-22, 13:22

Hello.
Culd you upload a 64bit CUDA compiled version for Windows, please? msieve 1.51 if possible.

jasonp · 2012-12-22, 13:45

I don't have a development environment that can do that; maybe Jeff Gilchrist can help?

Dubslow · 2012-12-24, 06:07

Does anyone know how/if the poly select code reacts badly to CUDA 5?

Edit: I realized I have 4.1 installed, I'll just go with that.

Edit2: @jrk: how stable/awesome is the new branch you're working on? @jasonp: How does that alpha value fix thingy affect CADO vs. Msieve poly scores? Presumably they're better matched now?

jasonp · 2012-12-24, 14:06

I don't think anyone has tested it with CUDA 5. Both CADO and Msieve needed fixes to their score calculations, they should produce E values within 10^-3 of each other. jrk's branch is insurance for the future, it won't help us for RSA896.

WraithX · 2013-03-13, 12:01

Quote:

Originally Posted by jasonp

I don't think anyone has tested it with CUDA 5. Both CADO and Msieve needed fixes to their score calculations, they should produce E values within 10^-3 of each other. jrk's branch is insurance for the future, it won't help us for RSA896.

Jason, would you like someone to test msieve gpu poly select with CUDA 5? I'm about to start a big (for me) poly select job and don't have CUDA installed yet. I'm trying to decide between 4.2 and 5.0. If 5.0 would help you out, then I can go that route.

Also, if I had questions about the poly select steps, should I post them in this thread, or start a new one?

jasonp · 2013-03-13, 13:50

I looked through the CUDA 5 manual, and think moving over should not be difficult. If you're willing, go ahead and give it a try, and if there are any complaints please post to the 1.51 feedback thread.

frmky · 2013-03-13, 20:12

I've been using CUDA 5 for a while now with no issues. As far as I see, there's no reason other than libraries for legacy compiled apps to use CUDA 4.2.

2012-12-24, 06:07	#60
Dubslow Basketry That Evening! "Bunslow the Bold" Jun 2011 40<A<43 -89<O<-88 3×29×83 Posts	Does anyone know how/if the poly select code reacts badly to CUDA 5? Edit: I realized I have 4.1 installed, I'll just go with that. Edit2: @jrk: how stable/awesome is the new branch you're working on? @jasonp: How does that alpha value fix thingy affect CADO vs. Msieve poly scores? Presumably they're better matched now? Last fiddled with by Dubslow on 2012-12-24 at 06:25

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
2^877-1 polynomial selection	fivemack	Factoring	47	2009-06-16 00:24
Polynomial selection	CRGreathouse	Factoring	2	2009-05-25 07:55
Intel Core i7 Extreme Edition Q965 on the X58 MOBO	IronBits	Hardware	17	2008-11-13 18:07
Vista 64 Ultimate Extreme Remix Limited Edition	Timber Jockey	PrimeNet	4	2008-10-20 19:39
Northwood, Prescott Or Extreme Edition?	georgekh	Hardware	13	2005-03-17 06:31

2012-11-18, 11:42	#56
banxian Oct 2012 1 Posts	svn800.zip link broken now

2012-11-19, 02:36	#57
jasonp Tribal Bullet Oct 2004 3,541 Posts	I've corrected the link in the OP to point to the current latest.

2012-12-22, 13:22	#58
skan Apr 2012 2×47 Posts	Hello. Culd you upload a 64bit CUDA compiled version for Windows, please? msieve 1.51 if possible.

2012-12-22, 13:45	#59
jasonp Tribal Bullet Oct 2004 3,541 Posts	I don't have a development environment that can do that; maybe Jeff Gilchrist can help?

2012-12-24, 14:06	#61
jasonp Tribal Bullet Oct 2004 3,541 Posts	I don't think anyone has tested it with CUDA 5. Both CADO and Msieve needed fixes to their score calculations, they should produce E values within 10^-3 of each other. jrk's branch is insurance for the future, it won't help us for RSA896.

2013-03-13, 13:50	#63
jasonp Tribal Bullet Oct 2004 3,541 Posts	I looked through the CUDA 5 manual, and think moving over should not be difficult. If you're willing, go ahead and give it a try, and if there are any complaints please post to the 1.51 feedback thread.

2013-03-13, 20:12	#64
frmky Jul 2003 So Cal 2×3⁴×13 Posts	I've been using CUDA 5 for a while now with no issues. As far as I see, there's no reason other than libraries for legacy compiled apps to use CUDA 4.2.