Prime95 on Puppy Linux?
 

Hello,

I'm wondering if there is a version of Prime95 (or any other software that will crunch numbers for GIMPS) that works on Puppy Linux.

As I've said on other threads, I have an old Pentium-75 PC with 128MB RAM and running Windows for Workgroups 3.11 that I'd like to add to my farm, but there doesn't seem to be any version of Prime95 that works on that edition of Windows. So I've been investigating Linux alternatives, as has been suggested, and I found Puppy Linux which claims to work on very limited resources, and moreover supposedly can boot off a floppy diskette and then run off a CD, even saving files on an existing Windows hard drive.

This SOUNDS like it could be the solution, as it requires minimum disruption to the existing system, but first I'd like to ask the experts around here if Puppy Linux is an adequate OS for GIMPS purposes.

(Please, no stuff about running ancient hardware, we've been down that road before! :smile: )

Thanks very much.

Rodrigo

[QUOTE=Rodrigo;250165]As I've said on other threads, I have an old Pentium-75 PC with 128MB RAM and running Windows for Workgroups 3.11 that I'd like to add to my farm, but there doesn't seem to be any version of Prime95 that works on that edition of Windows.

(Please, no stuff about running ancient hardware, we've been down that road before! :smile: )[/QUOTE]
If I recall correctly:
To run on something less than a P90 is quite troublesome. Also, an version that will run on that may be [U]out of work that it could do[/U].

As far as I know, mprime should work fine on Puppy Linux, since it's a fully functional Linux system with (IIRC) a reasonably modern kernel. Note, however, that as Uncwilly mentioned, it may be sufficiently slow that PrimeNet won't even assign it anything if it thinks it can't finish it by the deadline--Operation Billion Digits may be a good alternative for this computer.

[QUOTE=mdettweiler;250170]Note, however, that as Uncwilly mentioned, it may be sufficiently slow that PrimeNet won't even assign it anything if it thinks it can't finish it by the deadline[/QUOTE]Actually, any version of Mprime that can run on it might be so old that it does not have the need FFT sizes etc. for any assignments that maybe available. All of the factoring assignments that it could do might have already been done. The versions of Mprime95 might be old enough to run on it could not go above 79.3, except with "advanced factor", but that might be a problem.

Doing low level clearance work (thinning out the weeds) for OBD is a possibility or slowly moving the leading edge of OBD range up one expo at a time.

[QUOTE=Uncwilly;250172]Doing low level clearance work (thinning out the weeds) for OBD is a possibility or slowly moving the leading edge of OBD range up one expo at a time.[/QUOTE]
Uncwilly (and @mdettweiler),

Thanks a bunch for the info!

Suppose, then, that we set out to put this PC to do OBD work. That would still involve using Puppy Linux, since I can't get any of the Factor_X programs for Windows to work on it. Which software should I look for to do OBD? If I understand it, there are other programs that can do OBD work.

Rodrigo

[QUOTE=Uncwilly;250172]Actually, any version of Mprime that can run on it might be so old that it does not have the need FFT sizes etc. for any assignments that maybe available. All of the factoring assignments that it could do might have already been done. The versions of Mprime95 might be old enough to run on it could not go above 79.3, except with "advanced factor", but that might be a problem.[/QUOTE]
Hmm...how so? Does Puppy use an older (2.4?) kernel or something? Which also implies the question, do the latest mprimes require a 2.6 kernel?
[QUOTE=Rodrigo;250183]Uncwilly (and @mdettweiler),

Thanks a bunch for the info!

Suppose, then, that we set out to put this PC to do OBD work. That would still involve using Puppy Linux, since I can't get any of the Factor_X programs for Windows to work on it. Which software should I look for to do OBD? If I understand it, there are other programs that can do OBD work.

Rodrigo[/QUOTE]
AFAIK the only programs that are useful for OBD are FactorX and mfaktc. Since the latter requires a modern GPU, the only option for CPU contributions is FactorX. (Not sure if there's a Factor5 build for Linux, but I'm pretty sure there is for Factor4.)

[QUOTE=mdettweiler;250226]AFAIK the only programs that are useful for OBD are FactorX and mfaktc. Since the latter requires a modern GPU, the only option for CPU contributions is FactorX. (Not sure if there's a Factor5 build for Linux, but I'm pretty sure there is for Factor4.)[/QUOTE]
mdettweiler,

The OBD downloads page ([URL]http://www.moregimps.it/billion/download1.php[/URL]) lists a "Factor4 AMD 64" in the Linux column. Short of asking ET, can you tell from the name if that might (or might not) do the trick?

Rodrigo

[QUOTE=Rodrigo;250269]mdettweiler,

The OBD downloads page ([URL]http://www.moregimps.it/billion/download1.php[/URL]) lists a "Factor4 AMD 64" in the Linux column. Short of asking ET, can you tell from the name if that might (or might not) do the trick?

Rodrigo[/QUOTE]

I'm afraid it uses the 64-bit libraries and SSE2 extensions provided... :no:

Luigi

[QUOTE=ET_;250282]I'm afraid it uses the 64-bit libraries and SSE2 extensions provided... :no:

Luigi[/QUOTE]
Luigi,

Rats! I guess I'm outta luck, unless you or somebody else knows of some other software that'll work on that old machine.

Thanks for the scoop.

Rodrigo

[QUOTE=Rodrigo;250286]Luigi,

Rats! I guess I'm outta luck, unless you or somebody else knows of some other software that'll work on that old machine.

Thanks for the scoop.

Rodrigo[/QUOTE]

I should search an old Pentium PC, install Linux on it and recompile Factor.
Or [COLOR="Red"]look for someone on the forum having a Pentium plain vanilla with Linux and GMP installed[/COLOR], and giving advice on how to compile Factor (HINT, HINT!)

Anybody willing to help out there?

Luigi

[QUOTE=ET_;250306]I should search an old Pentium PC, install Linux on it and recompile Factor.
Or [COLOR=Red]look for someone on the forum having a Pentium plain vanilla with Linux and GMP installed[/COLOR], and giving advice on how to compile Factor (HINT, HINT!)

Anybody willing to help out there?

Luigi[/QUOTE]
Isn't it possible to cross compile for that system? Or has gcc dropped support for something so old?

Crosscompiling is not enough, more important will be glibc compatibility. Also in case of possible dynamic linkage to other libraries their versions should be checked to match.
Gcc has flags for all flavours of x86, i386 up.

[QUOTE] -mtune=cpu-type
Tune to cpu-type everything applicable about the generated code,
except for the ABI and the set of available instructions. The
choices for cpu-type are:

generic
Produce code optimized for the most common IA32/AMD64/EM64T
processors. If you know the CPU on which your code will run,
then you should use the corresponding -mtune option instead of
-mtune=generic. But, if you do not know exactly what CPU users
of your application will have, then you should use this option.

As new processors are deployed in the marketplace, the behavior
of this option will change. Therefore, if you upgrade to a
newer version of GCC, the code generated option will change to
reflect the processors that were most common when that version
of GCC was released.

There is no -march=generic option because -march indicates the
instruction set the compiler can use, and there is no generic
instruction set applicable to all processors. In contrast,
-mtune indicates the processor (or, in this case, collection of
processors) for which the code is optimized.

native
This selects the CPU to tune for at compilation time by deter-
mining the processor type of the compiling machine. Using
-mtune=native will produce code optimized for the local machine
under the constraints of the selected instruction set. Using
-march=native will enable all instruction subsets supported by
the local machine (hence the result might not run on different
machines).

i386
Original Intel's i386 CPU.

i486
Intel's i486 CPU. (No scheduling is implemented for this
chip.)[/QUOTE]

I have a notion that you are worrying too much. Just try it and we'll see if something (if anything) won't work.

Isn't a static build possible?

In general ? Sure, just link against static libraries (.a) and not the dynamic ones (.so)

Speaking about Windows, shouldn't [URL="http://www.sourceforge.net/projects/kernelex"]KernelEx[/URL] or something similar help you out?

Guys,

Thanks a bunch for all the ideas! I'm going to reveal my utter and nearly complete ignorance here, but would it be right to say that GCC might make it possible to create (compile) a version of FactorX that would work on that old edition of Windows?

Obviously I've got a lot of learning to do...

Rodrigo

[QUOTE=Rodrigo;250429]Guys,

Thanks a bunch for all the ideas! I'm going to reveal my utter and nearly complete ignorance here, but would it be right to say that GCC might make it possible to create (compile) a version of FactorX that would work on that old edition of Windows?

Obviously I've got a lot of learning to do...

Rodrigo[/QUOTE]

Meanwhile, I probably found an old AMD k6-350 (faster than P233 in integer math) with Linux installed.
If I get it to work, I'll try to compile a generic Factor5 for all Pentium platforms. Keep your fingers crossed... :smile:

Luigi

[QUOTE=ET_;250455]Meanwhile, I probably found an old AMD k6-350 (faster than P233 in integer math) with Linux installed.
If I get it to work, I'll try to compile a generic Factor5 for all Pentium platforms. Keep your fingers crossed... :smile:

Luigi[/QUOTE]
Luigi,

That would be fantastic!! Good luck!

Rodrigo