Prime95 version 25.7

[Prime95]

Originally Posted by petrw1

Is it imminent? Any reason not to use Build 2 in the meantime?

25.8 build 2 is safe to use. Build 3 will happen when I need someone to test a fix or a serious bug is reported that stops someone from using prime95.

I don't like to have a lot of "official" releases as most people feel compelled to upgrade, or at least investigate whether or not an upgrade is necessary. In other words, making an official release adds to everyone's workload and I don't want GIMPS to become a burden.

[ET_]

Originally Posted by Prime95

25.8 build 2 is safe to use. Build 3 will happen when I need someone to test a fix or a serious bug is reported that stops someone from using prime95.

I don't like to have a lot of "official" releases as most people feel compelled to upgrade, or at least investigate whether or not an upgrade is necessary. In other words, making an official release adds to everyone's workload and I don't want GIMPS to become a burden.

[James Heinrich]

Originally Posted by Prime95

25.8 build 2 is safe to use. Build 3 will happen when I need someone to test a fix or a serious bug is reported that stops someone from using prime95.

Are you able to release a v25.8.x 64-bit build? My current P-1 assignment is chugging along happily enough with 2500MB allocated on v25.8.2 (32-bit), but the TF on the other cores could be 50% faster if I could use the x64 version

[James Heinrich]

v25.8.2 is doing weird things with LowMemWhileRunning.

[Dec 10 19:00] Using 2468MB of memory.  Processing 10 relative primes (150 of 480 already processed).

[Main thread Dec 10 19:50] Entering a period of low memory usage because C:\PROGRAM FILES (X86)\WEBTEH\BSPLAYER\BSPLAYER.EXE is running.

[Dec 10 19:50] Restarting worker with new memory settings.
[Dec 10 19:50] Optimal P-1 factoring of M332203901 using up to 2500MB of memory.
[Dec 10 19:50] Assuming no factors below 2^75 and 2 primality tests saved if a factor is found.
[Dec 10 19:50] Optimal bounds are B1=3360000, B2=91560000
[Dec 10 19:50] Chance of finding a factor is an estimated 6.91%
[Dec 10 19:50] Using FFT length 20M, 2 threads
[Dec 10 19:50] Setting affinity to run helper thread 1 on any logical CPU.
[Dec 10 19:50] Cannot use lots of memory because C:\PROGRAM FILES (X86)\WEBTEH\BSPLAYER\BSPLAYER.EXE is running.
[Dec 10 19:50] Looking for work that uses less memory.
[Dec 10 19:50] Setting affinity to run helper thread 1 on any logical CPU.
[Dec 10 19:50] Resuming primality test of M46271419 using FFT length 2560K, 2 threads
[Dec 10 19:50] Iteration: 9108102 / 46271419 [19.68%].

Notice how the main thread notices a flagged program is running, and restarts the worker thread with (supposedly) new memory settings, but the worker goes through its whole 30-second initialization process until it also realizes a LowMemWhileRunning-flagged program is running, and then restarts itself (again), but properly this time -- it goes off and does some LL work until it can use more RAM for the P-1 again.

[James Heinrich]

I also noticed that when you tell a particular worker to start, it does it in an odd way: it tells all the workers to start then immediately stops all the workers you didn't want to start... that seems inelegant?

A small "bug" in the v25.8.4 changelog... actually a rare instance where there should be a "bug" but there isn't

8) A crash buf doing...

[James Heinrich]

I've no idea how it happened, but I managed to factor a number from 2^68 to 2^66 (yes, I mean it in that order, that's the problem):

[Tue Nov 04 04:07:43 2008]
UID: JamesHeinrich/q6600_257, M899999143 no factor from 2^59 to 2^63, Wd4: BE01652C, AID: FD7E1DC418C0EAAA0A709FF25D74FE0B
UID: JamesHeinrich/q6600_257, M899999143 no factor from 2^63 to 2^64, Wd4: BE01652C, AID: FD7E1DC418C0EAAA0A709FF25D74FE0B
[Tue Nov 04 04:12:50 2008]
UID: JamesHeinrich/q6600_257, M899999143 no factor from 2^64 to 2^65, Wd4: BE01652C, AID: FD7E1DC418C0EAAA0A709FF25D74FE0B
[Tue Nov 04 04:19:09 2008]
UID: JamesHeinrich/q6600_257, M899999143 no factor from 2^65 to 2^66, Wd4: BE01652C, AID: FD7E1DC418C0EAAA0A709FF25D74FE0B
[Tue Nov 04 04:31:43 2008]
UID: JamesHeinrich/q6600_257, M899999143 no factor from 2^66 to 2^67, Wd4: BE01652C, AID: FD7E1DC418C0EAAA0A709FF25D74FE0B
[Tue Nov 04 04:56:46 2008]
UID: JamesHeinrich/q6600_257, M899999143 no factor from 2^67 to 2^68, Wd4: BE01652C, AID: FD7E1DC418C0EAAA0A709FF25D74FE0B
[Tue Nov 04 21:07:04 2008]
UID: JamesHeinrich/q6600_257, M899999143 no factor from 2^68 to 2^66, Wd4: BE01652C, AID: FD7E1DC418C0EAAA0A709FF25D74FE0B

Even managed to make it through to PrimeNet:
http://v5www.mersenne.org/report_exp...p_lo=899999143

I only caught it when my own code found 1 TF assignment (out of 3.2 million I have a record of) valued at absolute-zero GHz-days.

[richs]

I mistakenly unreserved an exponent that one of my computers is currently double checking. It has not been reassigned yet. How do I re-reserve that number?

[Prime95]

Originally Posted by richs

I mistakenly unreserved an exponent that one of my computers is currently double checking. It has not been reassigned yet. How do I re-reserve that number?

Exit prime95. Add it back to the worktodo.txt. Start prime95. Advanced/manual communication, send new end dates.

[JuanTutors]

I just want to clarify, can my 100M exponents actually get reassigned after a year?

[Prime95]

No, 100M assignments are safe - we have plenty to go around :)

[lycorn]

Originally Posted by James Heinrich

I've no idea how it happened, but I managed to factor a number from 2^68 to 2^66 (yes, I mean it in that order, that's the problem).

That´s a good one. It deserves one
Seriously, have you had any explanation for that?
Does it correspond to TF from 2^68 to 2^69?