Prime95 version 25.5

[Prime95]

Originally Posted by Bundu

The minimise button on the windows seems to have stopped working?

Yes, so is the close button. I see little need for minimizing or closing MDI windows.

The ECM memory fix forced multiple ECM workers in stage 2 share the available memory. I'll investigate your particular single-worker ECM stage 2 problem.

[Bundu]

cool thanks

[ixfd64]

Someone might want to update the changelog.

By the way, George, do you know when the next "public" version of Prime95 will be released?

[Prime95]

Originally Posted by S00113

I'm getting SUMOUT errors at once when ECM factoring with the 64bit Linux client

That was a nasty bug to find. Please follow the new line to a 64-bit Linux executable.

[Prime95]

Originally Posted by ixfd64

By the way, George, do you know when the next "public" version of Prime95 will be released?

We are a long, long, way from that wonderful day

[Prime95]

Does anyone have a good reason why I should keep the -B command line argument for Linux?

[James Heinrich]

Observations for x64 version under Vista:

PrimeNet username setting is still broken:

[Sep 25 21:08] Updating computer information on the server
[Sep 25 21:08] URL: http://v5.mersenne.org/v5server/?v=0.95&px=GIMPS&t=uc&g=c4c1132ea5fa74fce4c0e3712cb9ed2d&hg=a558bb4549a8a940bfa2390cf9355e97&wg=ca82aec0c07
[Sep 25 21:08] RESPONSE:
[Sep 25 21:08] pnErrorResult=0
[Sep 25 21:08] pnErrorDetail=SUCCESS
[Sep 25 21:08] g=c4c1132ea5fa74fce4c0e3712cb9ed2d
[Sep 25 21:08] u=admin_user_anon
[Sep 25 21:08] un=ANONYMOUS
[Sep 25 21:08] cn=3600x2v255
[Sep 25 21:08] od=12
[Sep 25 21:08] ==END==
[Sep 25 21:08] 
[Sep 25 21:08] Done communicating with server.

I'd like to see an (additional) option to merge Main and Communication threads into one window and leave the worker threads each in their own window.

So far (12 minutes runtime) the ECM memory usage seems to be behaving itself. I'll report back if it misbehaves.

[StarQwest]

I understand that this new version can make use of multiple cores at once. When I am testing M39,584,257, the per iteration time when using one core is 0.101 seconds and when using two cores, it is 0.054 seconds, an increase of 87%. However, when testing M332,192,831, the per iteration time using one core is 0.960 seconds and when using two cores, it is 0.670 seconds, an increase of only 43%. How come the increase is so little? I would think that when using both cores (I have a Pentium D processor), the time would decrease to 0.480 seconds. Am I doing something wrong?

[jbristow]

My available memory is set to 1100MB for 25.5. The program crashed while using 1500MB, running 3 ECM workers and a TF. This is in Vista.

[Xyzzy]

I understand that this new version can make use of multiple cores at once. When I am testing M39,584,257, the per iteration time when using one core is 0.101 seconds and when using two cores, it is 0.054 seconds, an increase of 87%. However, when testing M332,192,831, the per iteration time using one core is 0.960 seconds and when using two cores, it is 0.670 seconds, an increase of only 43%. How come the increase is so little? I would think that when using both cores (I have a Pentium D processor), the time would decrease to 0.480 seconds. Am I doing something wrong?

You are using a much larger FFT size.

How much L2 cache do you have?

[ixfd64]

Originally Posted by Prime95

We are a long, long, way from that wonderful day

I'm guessing that there are still some issues that need to be resolved before the next stable version is released. The last "public" release happened in August 2005.