Request: P-1 79M

PageFault · 2003-09-29, 22:15

I have a 79M LL test but no machine with enough ram to do P-1. Added problem is that it is beyond SSE2 limits, i.e., has to be done on a AMD.

I don't know what you guys are doing here, but if anyone has over a GB of prime95 allocated ram we could arrange something - do my P-1 test and in the meantime I'll do a block of yours on my 2 GHz tbred with 512 MB.

My test would have to be done offline as I am a TPR member and our exponents cannot be unreserved - you would have to send me the .spl file for the result to be accepted.

Any ideas ? Last thing I want to do is a year of LL on a test with a findable factor.

GP2 · 2003-09-29, 23:17

Quote:

Originally posted by PageFault
I have a 79M LL test but no machine with enough ram to do P-1. Added problem is that it is beyond SSE2 limits, i.e., has to be done on a AMD.

Perhaps it could be done on a P4 if you set
CpuSupportsSSE2=0
in local.ini? (and remember to delete it again when it finishes). That's from undoc.txt.

Sorry, I have under 1 GB myself.

Xyzzy · 2003-09-29, 23:27

How much would that slow it down?

I've been trying to figure a way to get this particular exponent P-1 factored for a long time...

PageFault · 2003-09-30, 01:06

I recall the pre SSE2 client was something like three times slower than the first SSE2 release. So, dog slow is the word, akin to a 1 GHz tbird or so. Added complication is that the client treats cpu's differently - on AMD's I noticed that, for a block of exponents of similar size, the client cut back on the bounds compared to northwood, using same ram. Doubling the ram (256 to 512) resulted in the same bounds for AMD and Intel (256 MB). I don't know if cpu speed had anything to do with it but it appears that a fully loaded AMD, i.e., 2 GB ram, would do the same depth as a Northwood with 1 GB, this most likely being satisfactory.

Xyzzy · 2003-09-30, 06:08

Prime95 tells me it might take around two weeks... That's a long time...

PageFault · 2003-10-01, 00:50

I started it for comparison:

B1=770000
B2=11742500
Prob. of finding factor = 4.54%

I started it with the normal test line, so status doesn't show how much time for P-1. First iteration, 479.486 sec. for 0.09% of stage 1 ... at that rate, more than six days to do the easy part ...

Looks like I'm going to have to hold off on doing that test, or I'm going to have to hail to the

lycorn · 2003-10-01, 19:32

Quote:

Originally posted by PageFault
I Added problem is that it is beyond SSE2 limits, i.e., has to be done on a AMD.

I did not know there was such a limit. That means exps above a certain value can´t be tested on P4s, at least taking advantage of what makes them fast?

Xyzzy · 2003-10-01, 23:13

By the time we get to the 70M range, George will have fixed that...

Paulie · 2003-11-22, 21:32

I have a DL380/G2 with 2gigs of ram (it's a dual P3/1.15 something). I think my boss will let me play with it until january sometime, so I'd be willing to run the P-1 for you....

Let me know, else I'll continue to do some other Mersenne-ary work.

pakaran · 2003-11-22, 22:16

Quote:

Originally posted by Xyzzy
How much would that slow it down?

I've been trying to figure a way to get this particular exponent P-1 factored for a long time...

It would slow it down by a factor of 3 to 3.5, probably more.

Memory for XPs is in the multiple hundreds of dollars for that size range. Otherwise I'd get some and do it for you.

pakaran · 2003-11-22, 22:17

Quote:

Originally posted by Xyzzy
By the time we get to the 70M range, George will have fixed that...

I'd hope so. I think that should be 5 years at least.

2003-09-29, 22:15	#1
PageFault Aug 2002 Dawn of the Dead 5·47 Posts	Request: P-1 79M I have a 79M LL test but no machine with enough ram to do P-1. Added problem is that it is beyond SSE2 limits, i.e., has to be done on a AMD. I don't know what you guys are doing here, but if anyone has over a GB of prime95 allocated ram we could arrange something - do my P-1 test and in the meantime I'll do a block of yours on my 2 GHz tbred with 512 MB. My test would have to be done offline as I am a TPR member and our exponents cannot be unreserved - you would have to send me the .spl file for the result to be accepted. Any ideas ? Last thing I want to do is a year of LL on a test with a findable factor.

2003-09-30, 06:08	#5
Xyzzy Aug 2002 8554₁₀ Posts	Prime95 tells me it might take around two weeks... That's a long time... Attached Thumbnails

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Request	pinhodecarlos	Lounge	3	2017-10-26 18:58
Bug/request	Dubslow	YAFU	4	2012-03-31 03:07
Odd request?	Xyzzy	Lounge	23	2011-03-08 17:50
GMP-ECM Request	rogue	GMP-ECM	4	2009-11-23 15:07
A little request	10metreh	Aliquot Sequences	8	2009-10-07 18:13

2003-09-29, 23:27	#3
Xyzzy Aug 2002 8554₁₀ Posts	How much would that slow it down? I've been trying to figure a way to get this particular exponent P-1 factored for a long time...

2003-09-30, 01:06	#4
PageFault Aug 2002 Dawn of the Dead 5·47 Posts	I recall the pre SSE2 client was something like three times slower than the first SSE2 release. So, dog slow is the word, akin to a 1 GHz tbird or so. Added complication is that the client treats cpu's differently - on AMD's I noticed that, for a block of exponents of similar size, the client cut back on the bounds compared to northwood, using same ram. Doubling the ram (256 to 512) resulted in the same bounds for AMD and Intel (256 MB). I don't know if cpu speed had anything to do with it but it appears that a fully loaded AMD, i.e., 2 GB ram, would do the same depth as a Northwood with 1 GB, this most likely being satisfactory.

2003-10-01, 00:50	#6
PageFault Aug 2002 Dawn of the Dead 5×47 Posts	I started it for comparison: B1=770000 B2=11742500 Prob. of finding factor = 4.54% I started it with the normal test line, so status doesn't show how much time for P-1. First iteration, 479.486 sec. for 0.09% of stage 1 ... at that rate, more than six days to do the easy part ... Looks like I'm going to have to hold off on doing that test, or I'm going to have to hail to the

2003-10-01, 23:13	#8
Xyzzy Aug 2002 10000101101010₂ Posts	By the time we get to the 70M range, George will have fixed that...

2003-11-22, 21:32	#9
Paulie Aug 2002 223 Posts	I have a DL380/G2 with 2gigs of ram (it's a dual P3/1.15 something). I think my boss will let me play with it until january sometime, so I'd be willing to run the P-1 for you.... Let me know, else I'll continue to do some other Mersenne-ary work.