Prime95 version 25.7

[Prime95]

THIS IS A BETA VERSION - USE WITH CAUTION

Version of 25.7 is now available.

If you are running version 25.6 you can install the new version on top of the old version. If you were using version 25.6 to talk to the v4 server, then you will need to add the line "UseV4=1" at the front of prime.txt (if it isn't already there).

Steps to install version 25.7
---------------------------

1) DO NOT INSTALL OVER VERSION 24. If you have problems, you want to be able to revert to testing using version 24.
2) Create a new directory, download version 25.7 from the links below, unzip into this new directory.
3) Run version 25.7. Follow the link in the dialog box to create a userid.
4) Find problems. Report them to me.

Links to 25.7 versions
---------------------
Windows 32-bit - ftp://mersenne.org/gimps/p95v257.zip
Linux - ftp://mersenne.org/gimps/mprime257.tar.gz
Linux 64-bit - ftp://mersenne.org/gimps/mprime257-linux64.tar.gz
Windows 64-bit - ftp://mersenne.org/gimps/p64v257.zip
Mac OS X - ftp://mersenne.org/gimps/mprime257-MacOSX.tar.gz

What's new
-----------------

This is primarily a bug fix release. Several small fixes have accumulated over the months.

The main new feature is a rewrite of the memory allocation scheme. Hopefully, available memory is allocated more intelligently among multiple workers with fewer restarts. There are more undoc.txt items for controlling prime95 in unusually situations (PauseWhileRunning, LowMemWhileRunning, Memory).

The "Time=" feature of version 24 is officially dead. There is a new during/else syntax that only works with a few prime.txt and local.txt options. If you have a compelling reason for supporting during/else on another option, let me know.

[Prime95]

The Mac OS X version should work on OS X 10.4 but I cannot test that. If anyone has access to such a system, I'd appreciate a quick test.

If anyone has any idea how to create a staticly linked Linux version, I'd be grateful for the help. Using -lstatic whines that using gethostname (and others) means it cannot be linked staticly.

Version 25.7 sources are available too.

[Mini-Geek]

Originally Posted by Prime95

THIS IS A BETA VERSION - USE WITH CAUTION

Compare to 25.6's

Originally Posted by Prime95

THIS IS A PRE-BETA VERSION - NOT FOR EVERYDAY USE

So is that a hint that it's getting more stable and closer to release, or are you just happening to use a different wording?

[Prime95]

Originally Posted by Mini-Geek

So is that a hint that it's getting more stable and closer to release, or are you just happening to use a different wording?

Don't read too much into that! Since many are using it every day without problems, I thought it better to reduce the level of warning.

[harlee]

Tried running the Windwos 32-bit version and got the following errors

[Oct 7 16:27] Updating computer information on the server
[Oct 7 16:27] URL: http://v5.mersenne.org/v5server/?v=0...g=cc6e171616a0
[Oct 7 16:27] RESPONSE:
[Oct 7 16:27] pnErrorResult=9
[Oct 7 16:27] pnErrorDetail=Untrusted program versions currently excluded by PrimeNet
[Oct 7 16:27] ==END==
[Oct 7 16:27]
[Oct 7 16:27] PrimeNet error 9: Access denied
[Oct 7 16:27] Untrusted program versions currently excluded by PrimeNet
[Oct 7 16:27] Visit http://mersenneforum.org for help.
[Oct 7 16:27] Will try contacting server again in 15 minutes.

[James Heinrich]

Found a problem with memory allocation:

[Oct 7 19:13] Worker starting
[Oct 7 19:13] Setting affinity to run worker on any logical CPU.
[Oct 7 19:13] Using FFT length 96K
[Oct 7 19:13] ECM on M1836271: curve #2 with s=1768286443529228, B1=50000, B2=5000000
[Oct 7 19:13] Using 2009MB of memory in stage 2.
[Oct 7 19:13] Stage 2 init complete. 18570 transforms, 2 modular inverses. Time: 33.027 sec.
[Oct 7 19:14] Out of memory!
[Oct 7 19:14] Worker stopped.

Identical failure on both 32- and 64-bit Windows versions. System is Vista64 with 4GB physical RAM, Prime95 started with 1.0GB RAM in use; RAM use goes up to ~2.75GB and then fails (stage2 init seems to take about 1GB of RAM, the other 750GB is allocated post-init but it fails somewhere in that section. I'm not sure if a more verbose debug level would reveal anything?

[James Heinrich]

Most features from my initial v25.6 feature request post have been addressed (thanks George! ), these are the remaining ones:

• When doing multi-threaded ECM it shows "Using FFT length 20K, 2 threads", but for single-threaded it just says "Using FFT length 20K". Can you make it explicity say "Using FFT length 20K, 1 thread", please?
• ability to specify more than one type of worktype preference; for example I'd choose to any of TF, ECM or P-1. Ideally you could set a percentage distribution between the worktypes (e.g. spend 50% of time on TF, 30% on ECM and 20% on P-1) or have a "whatever-makes-most-sense" checkbox to auto-distribute the work among the selected types (taking into account how many real and hyperthreaded cores are in the current system, etc).
• When starting stage2 could Prime95 print a message to the effect of "max efficiency needs <1234>MB of RAM; <768>MB available; using <620>MB"
• If one thread is idle because it has run out of work to do that doesn't require a lot of RAM, see if it can borrow or steal a worktodo entry from another thread. For example, if thread #1 is doing P-1 work, but it's done all the stage1 work it can for now but needs to wait until more memory is available, it should be able to look at the queued work for other threads and steal (or borrow) low memory work (TF, P-1 stage 1, LL, etc) to do while it waits for more memory to be available.
• I'd like to see all worktodo automagically assigned to available CPUs (unless the user has explicitly requested that certain cores only do certain work). There's no reason for core[1] to be idle when workdoto[2] has work waiting.
  An extension to the above: If one (or more) cores are idle, and another core is working on something that can reasonably exploit more multithreading, respawn the other work with the appropriate number of threads to fully utilize the CPU. I'm currently doing a hack version of this: I have a dual-core CPU running TF-LMH on one core and LL on the other. Problem is that every now and then the server goes down and I can't get any work for the TF thread to do. Normally that would mean one core idle for hours (or days), but I have the LL set to run in 2 threads. Perhaps it's not ideal when TF has work (I'm not sure the of exact inefficiency ratio) but overall my throughput is definitely higher due to capturing all the otherwise-idle time on the TF core.

[Prime95]

Originally Posted by harlee

Tried running the Windwos 32-bit version and got the following errors
[Oct 7 16:27] pnErrorResult=9.

Fixed. The program is supposed to be self-registering on the server. However, Scott apparently allows this only if I run the client. Anyway, v25.7 build 3 is now registered with the server for all 5 ports.

[WraithX]

Well, I've just created an account at mersenne.org and am running the 64-bit windows version of Prime95 25.7. I've done some work with Prime95 25.6 64-bit and that is saved in results.txt. I just told it to connect to the primenet servers and it looks like it got an LL test to do. However, in the "worker #1" window it is talking about doing a P-1 test. Is anything wrong here? Here are the outputs: (also, at the bottom you'll see I've selected 5 threads, but it looks like it is only using 4)

<Communication thread - Inactive>
[Main thread Oct 7 19:22] Mersenne number primality test program version 25.7
[Main thread Oct 7 19:27] Starting worker.
[Comm thread Oct 7 19:27] Updating computer information on the server
[Comm thread Oct 7 19:27] URL: http://v5.mersenne.org/v5server/?v=0...071d34af06&wg=
[Comm thread Oct 7 19:27] RESPONSE:
[Comm thread Oct 7 19:27] pnErrorResult=0
[Comm thread Oct 7 19:27] pnErrorDetail=SUCCESS
[Comm thread Oct 7 19:27] g=e9a62e8cdeb42af4cfabb1c636f543ad
[Comm thread Oct 7 19:27] u=WraithX
[Comm thread Oct 7 19:27] un=WraithX
[Comm thread Oct 7 19:27] cn=
[Comm thread Oct 7 19:27] od=0
[Comm thread Oct 7 19:27] ==END==
[Comm thread Oct 7 19:27]
[Comm thread Oct 7 19:27] Exchanging program options with server
[Comm thread Oct 7 19:27] URL: http://v5.mersenne.org/v5server/?v=0...fWork=5&DayMem
[Comm thread Oct 7 19:27] RESPONSE:
[Comm thread Oct 7 19:27] pnErrorResult=0
[Comm thread Oct 7 19:27] pnErrorDetail=SUCCESS
[Comm thread Oct 7 19:27] od=1
[Comm thread Oct 7 19:27] ==END==
[Comm thread Oct 7 19:27]
[Comm thread Oct 7 19:27] Getting assignment from server
[Comm thread Oct 7 19:27] URL: http://v5.mersenne.org/v5server/?v=0...1ED2973BD9E904
[Comm thread Oct 7 19:27] RESPONSE:
[Comm thread Oct 7 19:27] pnErrorResult=0
[Comm thread Oct 7 19:27] pnErrorDetail=Server assigned Lucas Lehmer primality test work.
[Comm thread Oct 7 19:27] g=e9a62e8cdeb42af4cfabb1c636f543ad
[Comm thread Oct 7 19:27] k=3363D04DC68C6DEF90C523A2F197144D
[Comm thread Oct 7 19:27] A=1
[Comm thread Oct 7 19:27] b=2
[Comm thread Oct 7 19:27] n=46112389
[Comm thread Oct 7 19:27] c=-1
[Comm thread Oct 7 19:27] w=100
[Comm thread Oct 7 19:27] sf=68
[Comm thread Oct 7 19:27] p1=0
[Comm thread Oct 7 19:27] ==END==
[Comm thread Oct 7 19:27]
[Comm thread Oct 7 19:27] PrimeNet success code with additional info:
[Comm thread Oct 7 19:27] Server assigned Lucas Lehmer primality test work.
[Comm thread Oct 7 19:27] Got assignment 3363D04DC68C6DEF90C523A2F197144D: LL M46112389
[Comm thread Oct 7 19:27] Sending expected completion date for M46112389: Oct 25 2008
[Comm thread Oct 7 19:27] URL: http://v5.mersenne.org/v5server/?v=0...2F197144D&c=0&
[Comm thread Oct 7 19:27] RESPONSE:
[Comm thread Oct 7 19:27] pnErrorResult=0
[Comm thread Oct 7 19:27] pnErrorDetail=SUCCESS
[Comm thread Oct 7 19:27] ==END==
[Comm thread Oct 7 19:27]
[Comm thread Oct 7 19:27] Done communicating with server.


<Worker #1 - 0.00% of M46112389 P-1 stage 1>
[Oct 7 19:27] Worker starting
[Oct 7 19:27] Setting affinity to run worker on any logical CPU.
[Oct 7 19:27] No work to do at the present time. Waiting.
[Oct 7 19:27] Resuming.
[Oct 7 19:27] Optimal P-1 factoring of M46112389 using up to 2048MB of memory.
[Oct 7 19:27] Assuming no factors below 2^68 and 2 primality tests saved if a factor is found.
[Oct 7 19:27] Optimal bounds are B1=560000, B2=16520000
[Oct 7 19:27] Chance of finding a factor is an estimated 6.53%
[Oct 7 19:27] Setting affinity to run helper thread 1 on any logical CPU.
[Oct 7 19:27] Setting affinity to run helper thread 2 on any logical CPU.
[Oct 7 19:27] Setting affinity to run helper thread 3 on any logical CPU.
[Oct 7 19:27] Using FFT length 2560K, 5 threads
[Oct 7 19:27] Setting affinity to run helper thread 4 on any logical CPU.
[Oct 7 19:35] M46112389 stage 1 is 1.23% complete. Time: 497.406 sec.

[Prime95]

Originally Posted by WraithX

Well, I've just created an account at mersenne.org and am running the 64-bit windows version of Prime95 25.7. I've done some work with Prime95 25.6 64-bit and that is saved in results.txt. I just told it to connect to the primenet servers and it looks like it got an LL test to do. However, in the "worker #1" window it is talking about doing a P-1 test. Is anything wrong here?

That all looked good. Workers always use 1 thread. Since you selected multithreading using 5 threads you get the 1 worker thread and 4 helper threads.

[James Heinrich]

Originally Posted by WraithX

...looks like it got an LL test to do. However, in the "worker #1" window it is talking about doing a P-1 test. Is anything wrong here?

Sounds normal. Testing normally follows the sequence:

1. trial factoring (up to around 2^68)
2. P-1 factoring
3. LL testing
4. LL double-check (not by you, by someone else at a later date)