[quote=Prime95;133198]I just recorded the M1117009 result. You can delete the assignment.

Report the single M1061 curve in the M1061 factoring thread. My database is a bit behind in keeping up with the results reported in that thread.[/quote]
Okay, I unreserved the M1117009 assignment, and reported the M1061 curve in the M1061 factoring thread. :smile:

IS there somewhere a 'request list'?
Because I'm not really ready to go through all the posts.

Request: (I guess it's already asked...)
I'd like to halt a single worker thread when I have 4. P-1 takes a hell of a lot of RAM and slows down the computer... I'd like to halt just the P-1 when I'm working. Setting the day-RAM low isn't really 'nice', the program will be working too slow when I'm out.

[QUOTE=sonjohan;133217]IS there somewhere a 'request list'?[/QUOTE]I have my own request list that I keep pestering George to look at and include in future versions :smile:
[url]http://www.mersenneforum.org/showpost.php?p=121318&postcount=7[/url]

On there is basically your request(s) in more detailed form (stop individual threads, auto-set low RAM when certain programs are running, more complex RAM scheduling, etc).

[QUOTE=Prime95;132386]Plenty of UI, communication, and memory allocation rough spots.[/QUOTE]The memory allocation issues are really frustrating me. It seems to be an issue if a thread allocates ~2+GB for itself. If I set memory limit to 1800MB I never have issues. If I set it >2GB I have immediate failures on the start of Stage2 work, like the attached screenshot. You can see from the screenshot that Windows (Vista64) is using ~1GB, add the 2GB for Prime95 (25.6.6-x64) and you've still got almost a GB of free physical RAM left.

What it looks like to me is that Prime95 (32-bit version) doesn't know it's limited to 2GB per application, and the same code has been copied over to the 64-bit version of 25.6, since >2GB mem on 24.x works fine. Incidentally, Prime95-25.6 32-bit version performs exactly the same (as the attached screenshot). On a similar note, the 32-bit version of Prime95 should prevent the user from allocating more than 2GB memory.

I'm sure you're busy (if you weren't you would've already fixed it, I assume :smile:), but do you have any rough ETA of when you'll be able to take a look at (and hopefully fix) the memory-allocation issues of 25.6?

On my computer at work I frequently get a "Error creating worktodo.txt file" at least once a day (error occurs between 5 hours to 2-3 days of starting Prime95). I always need to keep checking on Prime95 to see if it hung... kill the process and restart it. There is no log on this error in my prime.log file, it is only outputed in my results.txt file.

Setup:

Atlon X2 Dual 4400+
2GB of Ram
Windows XP SP2
Prime92: 25.6.1.0

I currently run Factoring to low limit on both thread. I thaught it was a deadlock problem where both thread tried to write to worktodo.txt. But this problem never occured at home on my Core2 Quad running 4 thread TFLM.

[QUOTE=diamonddave;134643]On my computer at work I frequently get a "Error creating worktodo.txt file" at least once a day .[/QUOTE]

Do you have any utilities that show the number of open file handles for a process? I'm wondering if I'm running out due to some code forgetting to close an open file.

[QUOTE=Prime95;134654]Do you have any utilities that show the number of open file handles for a process? I'm wondering if I'm running out due to some code forgetting to close an open file.[/QUOTE]

I Only have File Monitor and Process Explorer from SysInternals. I couldn't find in either the number of file handle held by a process. If you know of some other utility?

However when tracking the Handle Count on Prime95 it stay within 90 and 100 handle count. Well below some other process, DevEnv 2400, Explorer, 1200 and Excel 1100 to name a few.

While tracking the file activity with filemon it looked like Prime95 is looking for worktodo.ini. Perhaps the error is there?

[CODE]5:33:04 PM prime95.exe:3184 OPEN C:\Documents and Settings\dcampeau\Mersenne 25.5\f874361027 NOT FOUND Options: Open Access: 00010080
5:33:04 PM prime95.exe:3184 QUERY INFORMATION C:\Documents and Settings\dcampeau\Mersenne 25.5 SUCCESS FileNameInformation
5:33:04 PM prime95.exe:3184 OPEN C:\Documents and Settings\dcampeau\Mersenne 25.5\f874361027 NOT FOUND Options: Open Access: 00100080
5:33:04 PM prime95.exe:3184 CREATE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Options: OverwriteIf Access: 00120196
5:33:04 PM prime95.exe:3184 QUERY INFORMATION C:\Documents and Settings\dcampeau\Mersenne 25.5 SUCCESS FileNameInformation
5:33:04 PM prime95.exe:3184 OPEN C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Options: Open Access: 00100080
5:33:04 PM prime95.exe:3184 QUERY INFORMATION C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS FileInternalInformation
5:33:04 PM prime95.exe:3184 CLOSE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS
5:33:04 PM prime95.exe:3184 OPEN C:\Documents and Settings\dcampeau\Mersenne 25.5\ SUCCESS Options: Open Directory Access: 00100000
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 0 Length: 13
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 13 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 69 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 125 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 181 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 237 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 293 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 349 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 405 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 461 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 517 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 573 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 629 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 685 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 741 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 797 Length: 2
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 799 Length: 13
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 812 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 868 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 924 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 980 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 1036 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 1092 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 1148 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 1204 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 1260 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 1316 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 1372 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 1428 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 1484 Length: 56
5:33:04 PM prime95.exe:3184 WRITE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS Offset: 1540 Length: 56
5:33:04 PM prime95.exe:3184 CLOSE C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.txt SUCCESS
5:33:04 PM prime95.exe:3184 OPEN C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.ini NOT FOUND Options: Open Access: 00010080
5:33:04 PM prime95.exe:3184 QUERY INFORMATION C:\Documents and Settings\dcampeau\Mersenne 25.5 SUCCESS FileNameInformation
5:33:04 PM prime95.exe:3184 OPEN C:\Documents and Settings\dcampeau\Mersenne 25.5\worktodo.ini NOT FOUND Options: Open Access: 00100080
5:33:04 PM prime95.exe:3184 OPEN C:\Documents and Settings\dcampeau\Mersenne 25.5\p874361207 NOT FOUND Options: Open Access: Read
5:33:04 PM prime95.exe:3184 QUERY INFORMATION C:\Documents and Settings\dcampeau\Mersenne 25.5 SUCCESS FileNameInformation
5:33:04 PM prime95.exe:3184 OPEN C:\Documents and Settings\dcampeau\Mersenne 25.5\p874361207 NOT FOUND Options: Open Access: 00100001
5:33:04 PM prime95.exe:3184 OPEN C:\Documents and Settings\dcampeau\Mersenne 25.5\f874361207 NOT FOUND Options: Open Access: Read
5:33:04 PM prime95.exe:3184 QUERY INFORMATION C:\Documents and Settings\dcampeau\Mersenne 25.5 SUCCESS FileNameInformation
5:33:04 PM prime95.exe:3184 OPEN C:\Documents and Settings\dcampeau\Mersenne 25.5\f874361207 NOT FOUND Options: Open Access: 00100001
[/CODE]

[QUOTE=diamonddave;134658]While tracking the file activity with filemon it looked like Prime95 is looking for worktodo.ini. Perhaps the error is there?[/QUOTE]

Perhaps. Did prime95 fail immediately after the failed attempt to open worktodo.ini?

I looked at the code and could not see a way this failure could happen. It only attempts to open worktodo.ini if opening worktodo.txt fails.

Nonwithstanding the bug, a paliative measure...
Maybe adding this to prime.txt will help (before [PrimeNet] section) -
[FONT=Courier New]WellBehavedWork=1 [/FONT]

(or maybe not... I just remembered that for my excercises with the 800-999M TF'ing I've created __huge__ worktodo.txt files with prescreened factorable exponents. The hard disk was going absolutely nuts when two threads were crunching on such a work assignment, trying to remove cleared assignments from the file. Then the HD got much happier and went silent -- when I added that line. But your client gets assignments from server! anyway, my guess is that you have this line at home and not at work)

I'm sure this has already been suggested somewhere, but:

Would it be possible to make Prime95/mprime do more frequent GCD's during a P-1 task? The GCD only takes a minute or two on my computers, so I can't help but think that putting more of them in the task would make P-1 more efficient (by finding factors sooner on average).

Maybe just keep the one GCD in stage 1, and during stage 2 do it 4 or 5 times.

[quote=jrk;134872]Would it be possible to make Prime95/mprime do more frequent GCD's during a P-1 task? The GCD only takes a minute or two on my computers, so I can't help but think that putting more of them in the task would make P-1 more efficient (by finding factors sooner on average).[/quote]The probability of P-1 finding a factor goes up [U]very[/U] slowly as B1/B2 go up, once you get past low limits such as B1=exponent/1000 and B2=exponent/100. I think that in all ordinary cases the extra time taken by the extra GCDs will completely wipe out any gain due to the rare instances of finding factors in GCDs before going all the way to the specified B1/B2 limits.

A few years ago, I did extensive P-1 time simulations, inspired by the same idea you have. I tried many combinations of B1/B2 and limits at which GCDs were performed. The only cases in which the average time to find a factor was improved by performing GCDs earlier than the standard times were with limit combinations that were bizarrely outside the ranges normally calculated by Prime95. Even then, if I added as little as 30 seconds to the time for each run to represent the extra manual intervention required because the combination could not be set up with the existing Prime95 algorithm, any improvement in efficiency just disappeared. It would have been better to just use the B1/B2 limits computed by the existing Prime95 algorithm.

So, adding sufficient complexity to the Prime95 algorithm for setting interim times to perform GCDs seems unjustified -- the maximum gain in the small percentage of cases where there was a gain was only a few seconds. In almost all cases, the extra GCD time (plus, remember, extra time for reinitializing work areas to proceed to higher limits if the GCD finds nothing*) is a pure waste, slowing down the overall effort.

However, it's quite possible that my simulations were faulty. I'd welcome any independent investigation.

- - -

* - I notice that when I give Prime95 (v24) 1.6 GB for P-1 stage 2 workareas, and use (Windows) Task Manager to watch storage allocations dynamically change, it takes about a minute (on a 2.8 GHz CPU) to initialize all the work areas.