I received a new computer at work, an i5 running Windows 7. I moved the Prime95 directory from my old P-4 machine to the new machine, downloaded the latest version of Prime95 and restarted. All worked well until I logged off for the night. On my old machine with Windows XP, Prime95 kept running overnight or weekends when logged off. On my new machine, Prime95 stops when I log off. Any advice to keep it running?

[QUOTE=richs;287267]I received a new computer at work, an i5 running Windows 7. I moved the Prime95 directory from my old P-4 machine to the new machine, downloaded the latest version of Prime95 and restarted. All worked well until I logged off for the night. On my old machine with Windows XP, Prime95 kept running overnight or weekends when logged off. On my new machine, Prime95 stops when I log off. Any advice to keep it running?[/QUOTE]

Windows Vista killed off prime95's ability to run as a service. The only work around I know of is to run the NT service version.

[QUOTE=Prime95;287280]Windows Vista killed off prime95's ability to run as a service. The only work around I know of is to run the NT service version.[/QUOTE]

Actually, it *can* still be done (that is, run Prime95/Prime64 as a service). It just requires some legwork. See my post with detailed instructions at [URL]http://www.mersenneforum.org/showpost.php?p=240247&postcount=7[/URL].

While I'm running MPrime, when a P-1 S2 worker finishes (or is almost done and has reduced mem usage) my system monitor still reports that MPrime is using the maxumum memory allocated to it, even if no memory is in use (besides the basics few MB required for all worktypes).

I noticed this abnormal detection of hyperthreaded cores:[quote]Mersenne number primality test program version 26.6
Optimizing for CPU architecture: Unknown Intel, L2 cache size: 256 KB, L3 cache size: 12 MB
Logical CPUs 1,2 form one physical CPU.
Logical CPUs [b]3,7[/b] form one physical CPU.
Logical CPUs [b]4,12[/b] form one physical CPU.
Logical CPUs 5,6 form one physical CPU.
Logical CPUs [b]8,10[/b] form one physical CPU.
Logical CPUs [b]9,11[/b] form one physical CPU.[/quote]It usually gets it right on this CPU (i7-3930K), either with detection or default-assumption. This is the only time I've noticed it confused like this.

Using 27.2, I just had the following:

[code][Feb 13 15:51] Iteration: 21330000 / 26166731 [81.51%]. Per iteration time: 0.021 sec.
[Feb 13 15:52] Iteration: 21335053/26166731, ERROR: ROUND OFF (0.40625) > 0.40
[Feb 13 15:52] Continuing from last save file.
[Feb 13 15:52] Resuming primality test of M26166731 using AVX Core2 type-3 FFT length 1344K, Pass1=448, Pass2=3K
[Feb 13 15:52] Iteration: 21321044 / 26166731 [81.48%].
[Feb 13 15:52] Possible hardware errors have occurred during the test! 1 ROUNDOFF > 0.4.
[Feb 13 15:52] Confidence in final result is fair.
[Feb 13 15:56] Iteration: 21330000 / 26166731 [81.51%]. Per iteration time: 0.021 sec.
[Feb 13 15:56] Possible hardware errors have occurred during the test! 1 ROUNDOFF > 0.4.
[Feb 13 15:56] Confidence in final result is fair.
[Feb 13 15:57] Disregard last error. Result is reproducible and thus not a hardware problem.
[Feb 13 15:57] For added safety, redoing iteration using a slower, more reliable method.
[Feb 13 15:57] Continuing from last save file.
[Feb 13 15:57] Resuming primality test of M26166731 using AVX Core2 type-3 FFT length 1344K, Pass1=448, Pass2=3K
[Feb 13 15:57] Iteration: 21335047 / 26166731 [81.53%].
[Feb 13 15:59] Iteration: 21340000 / 26166731 [81.55%]. Per iteration time: 0.021 sec.[/code]

I wasn't sure if it was something that needed looking into, and it happened long enough ago that I could not grab the check file.

[QUOTE=bcp19;289318]Using 27.2, I just had the following:
[/quote]Perhaps the 27.2 thread then? Sorry :P

[QUOTE=bcp19;289318]...

I wasn't sure if it was something that needed looking into, and it happened long enough ago that I could not grab the check file.[/QUOTE]
You can find the time of the error in results.txt log file. Look for the same text - it will be there only once (in the dialog window, it keeps repeating - "Lest on too close sight the users miss the darling illusion").

One reproducible round off error where the error is barely above 0.4 is of little concern.

Max Memory
 

Does anyone know if there is a maximum amount of memory that Prime95 x64 can use? I installed 96Gb of memory in a system and when it contacts PrimeNet, is says:

[CODE]
[Fri Feb 24 18:34:49 2012 - ver 26.6]
Exchanging program options with server
PrimeNet error 7: Invalid parameter
parameter DayMemory: Invalid int value/precision '88459'
[/CODE]

I know the server is having some 'problems' today, so maybe that's it, but I thought I'd ask anyway.

Thanks.

[QUOTE=flashjh;290780]Does anyone know if there is a maximum amount of memory that Prime95 x64 can use? I installed 96Gb of memory in a system and when it contacts PrimeNet, is says:[CODE]Invalid int value/precision '88459'[/CODE][/QUOTE]I'll hazard a guess that Prime95 can handle whatever you can throw at it (for now), you're more likely to run into Windows memory limits (e.g. Win7-Home = 16GB; Win7-Pro=192GB). PrimeNet, on the other hand, from what you just posted I would hazard a guess it stores the value in a 16-bit-unsigned field. Try setting a value of 65530 and 65540; if the first succeed and the second fails, that's very likely the problem. But this shouldn't affect how Prime95 itself runs, since locally it understand and accepts the larger amount (based on my hypothesis and no substantiated proof). It [i]may[/i] affect what assignments PrimeNet wants to give you: if it stores the value as zero then it probably won't hand out any P-1 assignments; if it caps it at 65535 then it should hand out assignments just fine.

I'd be curious to know the result of the above/below 2^16-1 limit test.