Which versions for multiple processors, Xeon?

[kdq]

For the Xeon, which version of the software should I be using to take best advantage?

For 32-bit multiple processors, which version?

[cheesehead]

Originally Posted by kdq

For the Xeon, which version of the software should I be using to take best advantage?

For 32-bit multiple processors, which version?

Partial answers:

For Intel CPUs, Prime95 Version 24 is the latest stable release, available for download in several versions for various OSes, here: http://www.mersenne.org/freesoft.htm. It does not have explicit multicore support; instead, you'll need to start multiple instances with different Affinities set. Each instance would work on a separate assignment (TF, LL, DC, whatever).

There are additional details concerning special Vista limitations, naming of files for different cores (e.g., "worktodo.ini", "work0001.ini", "work0002.ini" ...), setting up separate directories, cautions about memory contentions, and so on, that are discussed in several separate threads, primarily in the Software (http://mersenneforum.org/forumdisplay.php?f=10) or PrimeNet (http://mersenneforum.org/forumdisplay.php?f=11) subforums.

I (or someone) really need to collect the various explanations and post them in a single sticky thread or in mersennewiki.org. Right now, I don't have time to look up titles of the various relevant threads and post links here, but I'll try to do so later today or tomorrow.

Meanwhile, you could try using the forum search with appropriate keywords, but I'd have to wish you "Good Luck" with that.

Version 25, now in beta (or pre-beta) test, has multicore support (so that multiple cores can work cooperatively on a common assignment simultaneously). See sticky thread "Prime95 version 25.6" (http://mersenneforum.org/showthread.php?t=9779) in the PrimeNet subforum for the current discussion about it.

I invite anyone else to step in here to add info, especially about Xeons.

[henryzz]

Originally Posted by cheesehead

I (or someone) really need to collect the various explanations and post them in a single sticky thread or in mersennewiki.org. Right now, I don't have time to look up titles of the various relevant threads and post links here, but I'll try to do so later today or tomorrow.

that would be nice then we wont have to keep posting about vista

edit: if u have any things u want me to test on vista(if u dont have vista) just pm me

[kdq]

The only way i touch Vista is with a ten-foot pole (i.e., in a VM). Anyway, I am running 25.6 on all my machines, 2 and 4 core, x86 and Xeon. I guess I'll find out by the end of November.

Having been out of it for so long, I first grabbed four of th 100M primes for my Xeon. I turned them back in the next day; I'm patient, but I probably won't live until 2023. I hope i hand on until 2015 - that's when we break the light barrier.

[ET_]

Originally Posted by kdq

The only way i touch Vista is with a ten-foot pole (i.e., in a VM). Anyway, I am running 25.6 on all my machines, 2 and 4 core, x86 and Xeon. I guess I'll find out by the end of November.

Having been out of it for so long, I first grabbed four of th 100M primes for my Xeon. I turned them back in the next day; I'm patient, but I probably won't live until 2023. I hope i hand on until 2015 - that's when we break the light barrier.

But by the way we break the light barrier we won't have hardware restrictions anymore, computer speed will exponentially raise and you'll be able to finish your 100M exponents in few weeks...

Luigi

[kdq]

For that you want to use my complex-time computer. Since computing in both the first and third quadrants is positive, you can have your answer in advance of writing the program. I generally have the program display its listing as wqell as the computed results - it avoids the bugs that way.