Prime95 v25.3

[jasong]

Um, I screwed up. It wasn't a major screwup, but since it involved a quad-core, it's going to be a bit of work for someone. Suffice it to say, transferring the settings of the approved client over to the 25.3 version isn't wise. The following numbers may need to be reset. I have no intention of crunching them:

43190621  F  63               9.8  -7.8  52.2                   24-Jul-07 06:38  S543292        jasong01
43190633  F  63               9.8  -6.8  53.2                   24-Jul-07 06:38  S543292        jasong01
43190647  F  63               9.8  -5.8  54.2                   24-Jul-07 06:38  S543292        jasong01
43190699  F  63               9.8  -4.8  55.2                   24-Jul-07 06:38  S543292        jasong01
43190723  F  63               9.8  -3.8  56.2                   24-Jul-07 06:38  S543292        jasong01
43190827  F  63               9.8  -2.8  57.2                   24-Jul-07 06:38  S543292        jasong01
43190911  F  63               9.8  -1.8  58.2                   24-Jul-07 06:38  S543292        jasong01
43190921  F  63               9.8   0.2  60.2                   24-Jul-07 06:38  S543292        jasong01
43190993  F  63               9.8   1.2  61.2                   24-Jul-07 06:39  S543292        jasong01
43228243  F  65               2.0  -1.0  59.0                   01-Aug-07 02:57  S543292        jasong01
43228253  F  65               2.0   0.0  60.0                   01-Aug-07 02:57  S543292        jasong01
43270483  F  63               4.0  -1.0  59.0                   30-Jul-07 01:10  S543292        jasong01
43270489  F  63               4.0  -0.0  60.0                   30-Jul-07 01:10  S543292        jasong01
43293979  F  64               2.0   0.0  60.0                   01-Aug-07 03:00  S543292        jasong01
43306867  F  63               2.0   1.0  61.0                   01-Aug-07 02:57  S543292        jasong01
43306943  F  63               2.0   3.0  63.0                   01-Aug-07 02:58  S543292        jasong01
43306987  F  63               2.0   4.0  64.0                   01-Aug-07 02:58  S543292        jasong01
43306997  F  63               2.0   5.0  65.0                   01-Aug-07 02:58  S543292        jasong01
43307009  F  63               2.0   6.0  66.0                   01-Aug-07 02:58  S543292        jasong01
43307051  F  63               2.0   8.0  68.0                   01-Aug-07 02:58  S543292        jasong01
43307119  F  63               2.0  -1.0  59.0                   01-Aug-07 02:58  S543292        jasong01
43307129  F  63               2.0   0.0  60.0                   01-Aug-07 02:58  S543292        jasong01
43307141  F  63               2.0   1.0  61.0                   01-Aug-07 02:58  S543292        jasong01
43307171  F  63               2.0   3.0  63.0                   01-Aug-07 02:58  S543292        jasong01
43307227  F  63               2.0   4.0  64.0                   01-Aug-07 02:58  S543292        jasong01
43307249  F  63               2.0   5.0  65.0                   01-Aug-07 02:59  S543292        jasong01
43307261  F  63               2.0   7.0  67.0                   01-Aug-07 02:59  S543292        jasong01
43307309  F  63               2.0   8.0  68.0                   01-Aug-07 02:59  S543292        jasong01
43307359  F  63               2.0  -1.0  59.0                   01-Aug-07 02:59  S543292        jasong01
43307387  F  63               2.0   0.0  60.0                   01-Aug-07 02:59  S543292        jasong01
43307399  F  63               2.0   1.0  61.0                   01-Aug-07 02:59  S543292        jasong01
43307513  F  63               2.0   3.0  63.0                   01-Aug-07 02:59  S543292        jasong01
43307549  F  63               2.0   4.0  64.0                   01-Aug-07 02:59  S543292        jasong01
43307639  F  63               2.0   5.0  65.0                   01-Aug-07 02:59  S543292        jasong01
43307647  F  63               2.0   7.0  67.0                   01-Aug-07 02:59  S543292        jasong01
43307659  F  63               2.0   8.0  68.0                   01-Aug-07 03:00  S543292        jasong01
43307681  F  63               2.0  -1.0  59.0                   01-Aug-07 03:00  S543292        jasong01
43307699  F  63               2.0   1.0  61.0                   01-Aug-07 03:02  S543292        jasong01
43307729  F  63               2.0   3.0  63.0                   01-Aug-07 03:02  S543292        jasong01
43307753  F  63               2.0   4.0  64.0                   01-Aug-07 03:03  S543292        jasong01
43307801  F  63               2.0   5.0  65.0                   01-Aug-07 03:03  S543292        jasong01
43307807  F  63               2.0   7.0  67.0                   01-Aug-07 03:03  S543292        jasong01
43307861  F  63               2.0   8.0  68.0                   01-Aug-07 03:03  S543292        jasong01

[m_f_h]

Originally Posted by Prime95

Hmm. 25.3 should be able to read version 24 save files, but I did not put any effort into reading 25.1 save files as very few people are running that version. The only solution is to complete the assignments in 25.1.

wouldn't it be a matter of seconds to write a small tool converting one format to another (or, since this seems the most useful thing, converting from the old to the new format)
esentially by replacing e.g. in the old source code the save routine by the new one, and replacing the main program by something like
load_file(argv[1]); save_file(argv[2]);
or so? or do these routines make use of many "common" routines which are not common between versions ?

[jasong]

Would it be possible to come up with some sort of method to tell Prime95 when work is and isn't related to GIMPS? I was absent-mindedly messing with the Prime95 25.3, and I had a heart stopping moment where I thought everything in the worktodo was going to be deleted.

I'm thinking of two possibilities:

(1) Bracketed tags. Something like [ non-GIMPS ] [ /non-GIMPS ] (had to add spaces to get it to post it properly)
(2) a code word that sends Prime95 to look at a different file.
non-GIMPS=sobworktodo.txt would have it going to a file 'sobworktodo.txt' in the same directory. non-GIMPS=c:\program files\dc\sob\worktodo.txt would send it to look at that location.

I'm thinking the bracketed tabs approach would be easist, especially if you made the rule that the tags had to have their own line.

[jasong]

Sorry for double post. Although this is posted much later.

I've been stopping and restarting version 25.3 for various reasons. One effect of this is that thread 2 ran out of work while thread 1 ended up with about 18 hours worth, even though they started out even.

I think rather than assigning work to specific threads, it would be better(though complex to implement :( ) to make it where if there are 2 or more work threads that another thread is added just to give out work. This other special thread would be able to assign work more dynamically than what we have now.

I haven't programmed in years, but a possibility is that when the program is told it can go to work, it could edit the worktodo file and possibly rearrange stuff before it starts. Clicking 'Status' could cause it to do this, although these features might deserve a warning that asks permission, first to do it that time, and second a check box that would give permission to do it whenever necessary.

[AES]

Will there be a "Receive e-mail" checkbox in 25.x? I believe it sets the value to "NO" by default on the v4 server.

[Prime95]

Originally Posted by AES

Will there be a "Receive e-mail" checkbox in 25.x?

No. There hasn't been a mass emailing in years. Those interested in email news can either read this forum or join the low-traffic Mersenne mailing list.

[RichJacot]

Hello,

I hope this is the right place to post this....

I'm trying to run Mersenne Prime Test Program, Version 25.3.8 on RHEL5 (64bit) and I want to make sure I have it setup properly on two dual core cpu's before deploying it to 10 blades with four dual core cpu's.

I don't think I have it setup properly because when I check my individual account report it has:

38614073     75  38614073     5.9  60.1  59.1  12-Aug-07 20:29  08-Aug-07 15:41  rhadmin01    2612 v19/v20
38614133     75  38614133     5.9  60.1  59.1  12-Aug-07 17:27  08-Aug-07 15:41  rhadmin01    2612 v19/v20
38614351  *  68         3     5.9  60.1  59.1  12-Aug-07 16:39  08-Aug-07 15:41  rhadmin01    2612 v19/v20
38614363  *  68         2     5.9  60.1  59.1  12-Aug-07 16:39  08-Aug-07 15:41  rhadmin01    2612 v19/v20

Notice the second two don't have much for current iteration. My prime.txt looks like:

V24OptionsConverted=1
SendAllFactorData=1
StressTester=0
UsePrimenet=1
DialUp=0
UserID=RichJacot
DaysBetweenCheckins=1
UserPwd=removed password
UserName=Rich Jacot
AskedAboutMemory=1
WorkPreference=100
V5UserID=
OutputIterations=10000
ResultsFileIterations=999999999
DiskWriteTime=30
NetworkRetryTime=2
NetworkRetryTime2=70
DaysOfWork=5
TwoBackupFiles=1
SilentVictory=1

[PrimeNet]
Debug=1
ProxyHost=

My local.txt looks like:

OldCpuSpeed=2614
ComputerGUID=5a298c09ef8c4bd6e87dc8b218cd47d3
ComputerID=rhadmin01
DayMemory=2048
NightMemory=2048
DayStartTime=450
DayEndTime=1410
Affinity=100
ThreadsPerTest=1
Pid=4232
SrvrP00=5
SrvrPO1=100
SrvrPO2=1
SrvrPO3=5
SrvrPO4=273
SrvrPO5=273
SrvrPO6=450
SrvrPO7=1410
SrvrPO8=1
LastEndDatesSent=1186936742
RollingHash=1806291708
RollingStartTime=1187055722
RollingCompleteTime=22041904
RollingAverage=955

I haven't started anything with -A_ because I saw something earlier in this thread that said to use threads instead of affinity. Please let me know if you need to see any other files or information.

TIA

[chappjc]

With multiple worker threads now possible during a torture test, any hardware failure will likely occur on only one thread -- initially. I would suggest that it is logical to automatically terminate/stop all threads if an error occurs in any one of them. This way, if you are overclocking, you can take the load off the CPU long enough to keep the OS stable to reboot and/or tweak the hardware. Could this be a feature in the next version possibly?

[Prime95]

Originally Posted by RichJacot

I don't think I have it setup properly because when I check my individual account report it has:

38614073     75  38614073     5.9  60.1  59.1  12-Aug-07 20:29  08-Aug-07 15:41  rhadmin01    2612 v19/v20
38614133     75  38614133     5.9  60.1  59.1  12-Aug-07 17:27  08-Aug-07 15:41  rhadmin01    2612 v19/v20
38614351  *  68         3     5.9  60.1  59.1  12-Aug-07 16:39  08-Aug-07 15:41  rhadmin01    2612 v19/v20
38614363  *  68         2     5.9  60.1  59.1  12-Aug-07 16:39  08-Aug-07 15:41  rhadmin01    2612 v19/v20

Notice the second two don't have much for current iteration.

I haven't started anything with -A_ because I saw something earlier in this thread that said to use threads instead of affinity.

You look OK to me. The "current iteration" value is garbage when version 25.3 does trial factoring. To make sure mprime is using all CPU cores, do an "uptime" and make sure the load stays near 4.0 or larger.

Yes, the -A_ switch is obsolute in version 25.

There is a bug in P-1 memory allocation, so I'll try to build a version 25.4 today.

[jasong]

Sorry if this has been mentioned, I searched the thread for 'libcurl' and 'curl' and didn't find anything.

Anyway, I'm about to do p-1 over at PrimeGrid and would appreciate it libcurl.so.3 were implemented ASAP. Either that, or have the user Prime95 or Old Man Primenet give us a direct link to the correct file, so we won't download crap.

Thanks people, I appreciate everything you do and realize that 32-bit is better than nothing. But it never hurts to ask. :)

Edit: The 32-bit 25.5 download gets an error. (My quad-core has been down and out for about a week and a half, and now that it's fixed, I want to put it work right NOW, lol)

[jasong]

D'oh, wrong thread.