mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Software

Reply
 
Thread Tools
Old 2009-05-22, 00:55   #1
tyeken8
 
tyeken8's Avatar
 
May 2009
China

816 Posts
Question About self-compiled Prime95

Well, I have a small problem.

I downloaded the source code of Prime95 v25.9 and compiled it successfully.

But it can't report my stats to primenet successfully, prompting this:

Code:
[Comm thread May 22 08:48] URL: http://v5.mersenne.org/v5server/?v=0.95&px=GIMPS&t=uc&g=897f745f60f7ce4170b5c00f4a02a1a7&hg=a3d99cef5b3f53737f59f8b215fc426d&wg=&a=Linux,Untrusted+Prime95,v25.9,build+4&c=Genuine+Intel(R)+CPU+3.80GHz&f=Prefetch,SSE,SSE2&L1=16&L2=2048&np=1&hp=2&m=2026&s=4226&h=24&r=1000
[Comm thread May 22 08:48] RESPONSE:
[Comm thread May 22 08:48] pnErrorResult=7
[Comm thread May 22 08:48] pnErrorDetail=parameter ss: Invalid int value/precision ''
[Comm thread May 22 08:48] ==END==
[Comm thread May 22 08:48] 
[Comm thread May 22 08:48] PrimeNet error 7: Invalid parameter
[Comm thread May 22 08:48] parameter ss: Invalid int value/precision ''
[Comm thread May 22 08:48] Visit http://mersenneforum.org for help.
[Comm thread May 22 08:48] Will try contacting server again in 10 minutes.
I saw the correct HTTP request is

Code:
http://v5.mersenne.org/v5server/?v=0.95&px=GIMPS&t=uc&g=897f745f60f7ce4170b5c00f4a02a1a7&hg=a3d99cef5b3f53737f59f8b215fc426d&wg=&a=Linux,Prime95,v25.9,build+4&c=Genuine+Intel(R)+CPU+3.80GHz&f=Prefetch,SSE,SSE2&L1=16&L2=2048&np=1&hp=2&m=2026&s=4226&h=24&r=1000&ss=54683&sh=6A5D387CF8B7733715864CE015FC44CB
but mine is

Code:
http://v5.mersenne.org/v5server/?v=0.95&px=GIMPS&t=uc&g=897f745f60f7ce4170b5c00f4a02a1a7&hg=a3d99cef5b3f53737f59f8b215fc426d&wg=&a=Linux,Untrusted+Prime95,v25.9,build+4&c=Genuine+Intel(R)+CPU+3.80GHz&f=Prefetch,SSE,SSE2&L1=16&L2=2048&np=1&hp=2&m=2026&s=4228&h=24&r=1000
So what do para "ss" and "sh" mean? And how to solve the problem? Thanks.

BTW, how does the GHz-Days mean? Is it GHz*Day?
tyeken8 is offline   Reply With Quote
Old 2009-05-22, 02:24   #2
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

11101001110112 Posts
Default

I've forwarded your problem to Scott.
Prime95 is offline   Reply With Quote
Old 2009-05-22, 07:57   #3
tyeken8
 
tyeken8's Avatar
 
May 2009
China

23 Posts
Default

Quote:
Originally Posted by Prime95 View Post
I've forwarded your problem to Scott.
Thanks.

I looked up in the PrimeNet's Web API Specification
And it says that "ss" and "sh" is optional.

but my mprime said this:

Code:
parameter ss: Invalid int value/precision ''
tyeken8 is offline   Reply With Quote
Old 2009-05-22, 20:15   #4
cheesehead
 
cheesehead's Avatar
 
"Richard B. Woods"
Aug 2002
Wisconsin USA

22×3×641 Posts
Default

Quote:
Originally Posted by tyeken8 View Post
BTW, how does the GHz-Days mean? Is it GHz*Day?
Yes. 1 GHz-day is the number of clock cycles that result from a 1 GHz rate sustained for 24 hours (or 2 Ghz rate sustained for 12 hours, 500 MHz rate sustained for 48 hours, and so on...).
cheesehead is offline   Reply With Quote
Old 2009-05-23, 12:09   #5
tyeken8
 
tyeken8's Avatar
 
May 2009
China

810 Posts
Default

Quote:
Originally Posted by cheesehead View Post
Yes. 1 GHz-day is the number of clock cycles that result from a 1 GHz rate sustained for 24 hours (or 2 Ghz rate sustained for 12 hours, 500 MHz rate sustained for 48 hours, and so on...).
Well, where's my earlier post??

I said:

I (as a noob in Linux) am running mprime (also Prime95 on Win) for more than 10 days but only gained about 10 GHz-days, what does that mean?

BTW #2, I wonder if everyone's thread should be checked manually by admin? Different from most of other forums... Or just because of my Join Date or number of posts?
tyeken8 is offline   Reply With Quote
Old 2009-05-23, 12:10   #6
tyeken8
 
tyeken8's Avatar
 
May 2009
China

10002 Posts
Default

Hey, It successes~
tyeken8 is offline   Reply With Quote
Old 2009-05-23, 12:13   #7
henryzz
Just call me Henry
 
henryzz's Avatar
 
"David"
Sep 2007
Cambridge (GMT/BST)

5,869 Posts
Default

Quote:
Originally Posted by tyeken8 View Post
BTW #2, I wonder if everyone's thread should be checked manually by admin? Different from most of other forums... Or just because of my Join Date or number of posts?
from vague memory it is the first 5 posts that need verifying
henryzz is offline   Reply With Quote
Old 2009-05-23, 23:23   #8
tyeken8
 
tyeken8's Avatar
 
May 2009
China

108 Posts
Default

Quote:
Originally Posted by henryzz View Post
from vague memory it is the first 5 posts that need verifying
O, I see. Thx.
tyeken8 is offline   Reply With Quote
Old 2009-05-25, 00:27   #9
cheesehead
 
cheesehead's Avatar
 
"Richard B. Woods"
Aug 2002
Wisconsin USA

22×3×641 Posts
Default

Quote:
Originally Posted by tyeken8 View Post
I (as a noob in Linux) am running mprime (also Prime95 on Win) for more than 10 days but only gained about 10 GHz-days, what does that mean?
GIMPS credits are based on the speed with which a reference system performed the calculations. That is, 1.0 GHz-day of credit means the calculations for that result would have taken 1.0 GHz-day on the reference system. (So, L-L on exponent 98765431 will get the same amount of credit no matter what type of system you actually use.)

Two possible reasons why the amount of credit would be less than your system's elapsed time (after adjusting for the difference in CPU clock speed between your system and the reference system, of course) for the calculation are:

Your system's CPU type may differ from the reference system's CPU type. Different CPU types have different efficiencies in performing the kinds of computation done for GIMPS assignments. Some types will require more (or less) than 1.0 GHz-day to complete a calculation that the reference system performed in exactly 1.0 GHz-day.

or

Other tasks running on your system would have lengthened the total elapsed time on your system, compared to how long it would have taken if only GIMPS client, nothing else, had been running all that time. The reference system's speed was measured with nothing else running on it.

Because the inner computation loop in mprime is identical to the inner computation loop in prime95 (both are coded in assembly language), there shouldn't be any difference in credit because of that. It is possible that differences in how the two operating systems perform support tasks for necessary functions of the GIMPS client could make a difference between mprime times and prime95 (which is what the reference system used) times, but that's probably very small.

Maybe there's something else, but I can't think of one now.

Last fiddled with by cheesehead on 2009-05-25 at 00:31
cheesehead is offline   Reply With Quote
Old 2009-05-25, 11:56   #10
tyeken8
 
tyeken8's Avatar
 
May 2009
China

23 Posts
Default

Quote:
Originally Posted by cheesehead View Post
GIMPS credits are based on the speed with which a reference system performed the calculations. That is, 1.0 GHz-day of credit means the calculations for that result would have taken 1.0 GHz-day on the reference system. (So, L-L on exponent 98765431 will get the same amount of credit no matter what type of system you actually use.)

Two possible reasons why the amount of credit would be less than your system's elapsed time (after adjusting for the difference in CPU clock speed between your system and the reference system, of course) for the calculation are:

Your system's CPU type may differ from the reference system's CPU type. Different CPU types have different efficiencies in performing the kinds of computation done for GIMPS assignments. Some types will require more (or less) than 1.0 GHz-day to complete a calculation that the reference system performed in exactly 1.0 GHz-day.

or

Other tasks running on your system would have lengthened the total elapsed time on your system, compared to how long it would have taken if only GIMPS client, nothing else, had been running all that time. The reference system's speed was measured with nothing else running on it.

Because the inner computation loop in mprime is identical to the inner computation loop in prime95 (both are coded in assembly language), there shouldn't be any difference in credit because of that. It is possible that differences in how the two operating systems perform support tasks for necessary functions of the GIMPS client could make a difference between mprime times and prime95 (which is what the reference system used) times, but that's probably very small.

Maybe there's something else, but I can't think of one now.
Thx.

Well, maybe my computer is out-of-date...

Quote:
Other tasks running on your system would have lengthened the total elapsed time on your system, compared to how long it would have taken if only GIMPS client, nothing else, had been running all that time.
I think it can be ignored. I saw the CPUs (HyperThread) are used more than 190% in "top", which can equal 8.4GHz or a bit less.

I will buy a new computer next month, and I'm considering to buy an i7 cpu, does anyone knows if it's good for GIMPS?

Quote:
Because the inner computation loop in mprime is identical to the inner computation loop in prime95 (both are coded in assembly language), there shouldn't be any difference in credit because of that. It is possible that differences in how the two operating systems perform support tasks for necessary functions of the GIMPS client could make a difference between mprime times and prime95 (which is what the reference system used) times, but that's probably very small.
Yes, I've mentioned it. It's just a little bit rate difference to prime95 (both of them are slow, oh, my CPU..).

Thanks again.

Tydus
tyeken8 is offline   Reply With Quote
Old 2009-05-26, 02:33   #11
markr
 
markr's Avatar
 
"Mark"
Feb 2003
Sydney

3×191 Posts
Default

Quote:
Originally Posted by tyeken8
I (as a noob in Linux) am running mprime (also Prime95 on Win) for more than 10 days but only gained about 10 GHz-days, what does that mean?
Quote:
Originally Posted by tyeken8
I saw the CPUs (HyperThread) are used more than 190% in "top", which can equal 8.4GHz or a bit less.

I will buy a new computer next month, and I'm considering to buy an i7 cpu, does anyone knows if it's good for GIMPS?
Welcome! A couple more things to consider (although you may well already be aware of them).

Primenet credit is only given for completed assignments. So it comes in lumps, not continuously. This often catches newcomers out.

I could be really wrong, but even though the OS says you're getting 190%, I think your effective output will be less than that. (Assuming a single-core hyperthreaded CPU. If you actually have two CPUs or cores, ignore all this.) The kind of assignment will affect it, too.
[Cue: somebody who actually knows what they're talking about should enter now.]

An i7 would be good. Consider faster/quality RAM, too, if you're taking this project into account in your choices. There should be useful stuff in the hardware section in this forum.
markr is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Experimental lasieve4_64, compiled with MinGW-w64! Dan Ee Factoring 40 2016-02-08 20:32
Looking for windows-compiled gnfs-lasieve4I15e Andi47 Factoring 15 2008-02-21 19:15
How I get a compiled version of GMP-ECM? MatWur-S530113 GMP-ECM 12 2007-10-22 20:33
compiled msieve on PS3 bigmic Msieve 2 2007-03-29 20:06

All times are UTC. The time now is 14:30.

Fri May 7 14:30:37 UTC 2021 up 29 days, 9:11, 0 users, load averages: 4.93, 4.38, 3.72

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

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.