Expanding on lfm's answer:

[quote=asmfan;196151]Actually is there a strict formulae for memory allocations? Or how strict the manual memory limit is regarded?[/quote](You're referring to the "Daytime available memory" and "Nighttime available memory", I presume. If not, please clarify.)

Prime95 automatically allocates what it needs, no more or less, in all cases except stage 2 of P-1 (and stage 2 of ECM, but you won't be assigned that unless you ask for ECM). Only in those two stages can Prime95 use as many auxiliary memory areas as the user wants to allow it, and that's the only time when "Daytime available memory" and "Nighttime available memory" apply. They don't apply to LL or TF testing, or to stage 1 of P-1 or ECM.

(I wish they were labeled something like "Daytime auxiliary memory for stage 2 P-1/ECM" and "Nighttime auxiliary memory for stage 2 P-1/ECM".)

A Test= assignment may do P-1 on an exponent before it does the LL. This "implied" P-1 will run just fine with both day and night "available memory" at the default setting of 8. If you want to set the day and night "available memory" to some higher figures, go ahead -- it [I]will[/I] make the P-1 test more effective, and you [I]will[/I] be contributing more to GIMPS -- but it's not a big deal unless you're a P-1 geek.

[quote]I think it doesnt matter which OS but process statistics matters. Everyone can look at process memory statistics in taskmanager (ctrl+shift+esc on Win and select proper columns) with its peak memory level and exceeeds manual limits a lot.[/quote]That's the sort of confusion that results from having such a too-vague label as "available memory" for those two settings.

[I]Don't bother comparing "available memory" to what the Task Manager shows, unless Prime95 is doing P-1 stage 2 or ECM stage 2 [U]at that very time[/U]. [/I]Such a comparison doesn't mean anything at any other time.

[QUOTE=cheesehead;196374](I wish they were labeled something like "Daytime optional memory for stage 2 P-1/ECM" and "Nighttime optional memory for stage 2 P-1/ECM".) [/QUOTE]

They are in 25.11. I don't ignore [I]all[/I] of your suggestions :smile:

I apologize for (a) not paying attention, and (b) not having upgraded recently.

Cheesehead (et al.), I think asmfan is referring to the memory setting available when the custom torture test is selected.

[quote=Prime95;181057]Known bugs and fixes:
...
[/quote]

The linux 64 load average stuff isn't working. In 25.9 it would stop running but never resume. In 25.11 it never stops running.

Is it known and fixed or should I take a stab at it?

[QUOTE=andrewburgess;198655]Is it known and fixed or should I take a stab at it?[/QUOTE]

This is the first I've heard of the problem. Please attempt a fix - I'll be happy to incorporate it.

sumout errors
 

While I was using Prime95 Win ver 25.11b2 I frequently but not always had sumout errors during P-1 factoring Phase 1. After exiting and changing the executive to 25.8b1 The exponent I was working on went back and restarted testing instead of continuing from where I had left off(~ 20% of stage 2). With 25.8 no sumout error occurred even though I was doing as much as possible during the first stage to see if I could force a sumout error and none occurred. Maybe that was the wrong approach. My usual expereience with any errors always traced back to memory problems but even with very conservative settings occassionally a sumout error would still occur.

I wondered if it possibly could be related to [URL="http://www.mersenneforum.org/showthread.php?t=12814"]Segfault errors in Mprime[/URL] when a factor is found. All I know is that memory allocation is a tricky business when you don´t know what you´re doing as would be my situation. So please don´t jump on me for making an erroneous troubleshooting statement. For now the problem seems to be resolved using an older version.

The System I´m using for P-1 is a 3.00Ghz PIV Prescott with 1 MB Cache at 3.08 Ghz and 2 GB Main memory. All drivers are up to date and temperatures are normal (for presshot that is). P-1 is set to run on any logical CPU since th CPU is hyperthreaded. Earlier I had been multi threading P-1 but could not measure any appreciable gains in speed.

I try to live with the windows size not being saved when changes are made but since I restart my systems weekly it begins to be a nuisance. As I recall 25.7 saved window adjustments properly. And the bugger of the bottom line continually scrolling up until the top of the window is reached may be caused by how information is "PRINT"ed to the screen ie. CRLF is unneccesary at the end of a line because every call to output has a CRLF automatically attached to the beginning of a new line when using the standard entry point.

nelson

I suffer from the 100%(+) progress bug, again:

[code]
[Worker #4 Jan 24 02:05] M79000903 stage 2 is 95.14% complete. Time: 1844.460 sec.
[Worker #4 Jan 24 02:17] Available memory is 524MB.
[Worker #4 Jan 24 02:17] Using 501MB of memory. Processing 7 relative primes (405 of 480 already processed).
[Worker #4 Jan 24 02:35] M79000903 stage 2 is 95.55% complete. Time: 1802.655 sec.
[Worker #4 Jan 24 03:06] M79000903 stage 2 is 95.97% complete. Time: 1813.133 sec.
[Worker #4 Jan 24 03:36] M79000903 stage 2 is 96.39% complete. Time: 1822.239 sec.
[Worker #4 Jan 24 04:07] M79000903 stage 2 is 96.81% complete. Time: 1861.264 sec.
[Worker #4 Jan 24 04:17] Available memory is 524MB.
[Worker #4 Jan 24 04:17] Using 501MB of memory. Processing 7 relative primes (412 of 480 already processed).
[Worker #4 Jan 24 04:37] M79000903 stage 2 is 97.22% complete. Time: 1826.023 sec.
[Worker #4 Jan 24 05:09] M79000903 stage 2 is 97.65% complete. Time: 1868.391 sec.
[Worker #4 Jan 24 05:39] M79000903 stage 2 is 98.07% complete. Time: 1814.402 sec.
[Worker #4 Jan 24 06:09] M79000903 stage 2 is 98.49% complete. Time: 1827.020 sec.
[Worker #4 Jan 24 06:17] Available memory is 524MB.
[Worker #4 Jan 24 06:17] Using 501MB of memory. Processing 7 relative primes (419 of 480 already processed).
[Worker #4 Jan 24 06:40] M79000903 stage 2 is 98.90% complete. Time: 1837.607 sec.
[Worker #4 Jan 24 07:11] M79000903 stage 2 is 99.32% complete. Time: 1857.618 sec.
[Worker #4 Jan 24 07:42] M79000903 stage 2 is 99.74% complete. Time: 1843.310 sec.
[Worker #4 Jan 24 08:14] M79000903 stage 2 is 100.00% complete. Time: 1960.750 sec.
[Worker #4 Jan 24 08:21] Available memory is 524MB.
[Worker #4 Jan 24 08:21] Using 501MB of memory. Processing 7 relative primes (426 of 480 already processed).
[Worker #4 Jan 24 08:49] M79000903 stage 2 is 100.00% complete. Time: 2075.418 sec.
[Worker #4 Jan 24 09:29] M79000903 stage 2 is 100.00% complete. Time: 2388.387 sec.
[Worker #4 Jan 24 10:16] M79000903 stage 2 is 100.00% complete. Time: 2846.276 sec.
[Worker #4 Jan 24 10:53] M79000903 stage 2 is 100.00% complete. Time: 2217.720 sec.
[Worker #4 Jan 24 10:57] Available memory is 524MB.
[Worker #4 Jan 24 10:57] Using 501MB of memory. Processing 7 relative primes (433 of 480 already processed).
[Worker #4 Jan 24 11:30] M79000903 stage 2 is 100.00% complete. Time: 2210.961 sec.
[Worker #4 Jan 24 12:22] M79000903 stage 2 is 100.00% complete. Time: 3103.092 sec.
[Worker #4 Jan 24 13:15] M79000903 stage 2 is 100.00% complete. Time: 3218.025 sec.
[Worker #4 Jan 24 13:55] M79000903 stage 2 is 100.00% complete. Time: 2383.107 sec.
[Worker #4 Jan 24 13:55] Available memory is 524MB.
[Worker #4 Jan 24 13:55] Using 501MB of memory. Processing 7 relative primes (440 of 480 already processed).
[Worker #4 Jan 24 14:35] M79000903 stage 2 is 100.00% complete. Time: 2386.610 sec.
[Worker #4 Jan 24 15:12] M79000903 stage 2 is 100.00% complete. Time: 2253.594 sec.
[Worker #4 Jan 24 15:53] M79000903 stage 2 is 100.00% complete. Time: 2424.657 sec.
[Worker #4 Jan 24 16:29] Available memory is 524MB.
[Worker #4 Jan 24 16:29] Using 501MB of memory. Processing 7 relative primes (447 of 480 already processed).
[Worker #4 Jan 24 16:30] M79000903 stage 2 is 100.00% complete. Time: 2262.537 sec.
[/code]If it keeps the pace of 42 relative primes per 14:12 it might actually finish tonight, only some 36 hours after the ETA it projected yesterday... :big grin:

Seriously. Since it is obviously hard to fix this 100%(+) bug, can we at least have yet another fix which corrects the progress percentage whenever a new set of relative primes is started? :innocent:

It's somewhat annoying to have to use a calculator to compute ETAs...

[quote=Prime95;196375]They are in 25.11. I don't ignore [I]all[/I] of your suggestions :smile:[/quote]

Do you have a statical linked version of the linux mprime 64 bit executable?

[quote=joblack;208308]Do you have a statical linked version of the linux mprime 64 bit executable?[/quote]
I could have sworn that the stock 25.11 binaries [I]are[/I] statically linked. (Unless that's just the 32-bit version.)

[quote=mdettweiler;208319]I could have sworn that the stock 25.11 binaries [I]are[/I] statically linked. (Unless that's just the 32-bit version.)[/quote]

You're right - I accidentally took the version from Mersenne.org. Perhaps it's time to relink it to the official page.