[deathstar13]

http://superpi.radeonx.com/super_pi_mod-1.4.zip

this program is my favorite for stability testing. and would mainly just need the 32meg option to loop.

im sure this would be a big hit with oc'ers far and near

[Sucka]

Great idea. Got my support

[Drec]

Originally Posted by Sucka

Great idea. Got my support

ditto, whoever pulls it off will be considered a god of pi stableness.

[Scott9027]

Since it's a project that involves altering the program, maybe you should put this in the Programming section if it is still active.

[deathstar13]

Originally Posted by Scott9027

Since it's a project that involves altering the program, maybe you should put this in the Programming section if it is still active.

ill pm a mod and see if they think it would be better there.
tbh it was dead,dead,dead there and i thought for sure it would never get seen.
yeh i know not exactly following the rules but ill turn myself in

btw there a few i saw that were mods who ill pm that looked like they do programming so mabey we will get lucky.

[deathstar13]

i will be pushing this further till i find someone to do it! i posted on another geek forum for help and will be sending more pm's out today.

i really want this to happen

[Drec]

Originally Posted by deathstar13

i will be pushing this further till i find someone to do it! i posted on another geek forum for help and will be sending more pm's out today.

i really want this to happen

I'm pretty sure it's not very hard to do, you wouldn't even have to alter the program you can just make some sort of seperate program that goes and selects 32m for you everytime it is finished, maybe with an option where you can select how many times you want to run it...don't know hardware is more my thing. always found programming to be boring .

And ima talk to some buds of mine that do this sorta thing maybe someone can help us out.

[deathstar13]

Originally Posted by Drec

I'm pretty sure it's not very hard to do, you wouldn't even have to alter the program you can just make some sort of seperate program that goes and selects 32m for you everytime it is finished, maybe with an option where you can select how many times you want to run it...don't know hardware is more my thing. always found programming to be boring .

And ima talk to some buds of mine that do this sorta thing maybe someone can help us out.

i guess thats possible that way also. i dont care how its done long as it wroks

looks like my pm about moving this worked also.
i have many pm's sent out and even posted this in a programming forum at another site.
ill cross my fingers!

[CCUABIDExORxDIE]

ill back you 100% in this, if only i knew how to code this. hopefully someone will do this for us, then we wont be bound to Prime95 for hours of stability. Were gonna hafta come up with a good name for it though. my votes for SuperPi95

[DaWiper]

Does Superpi take command line options? If it can take commandline options to do 32m and quit after it's done it would be doable to make a basic-based program to control it.

[deathstar13]

Originally Posted by DaWiper

Does Superpi take command line options? If it can take commandline options to do 32m and quit after it's done it would be doable to make a basic-based program to control it.

i have no clue about what or how the program runs,i dont know programming at all.i linked to it above.

[DaWiper]

The readme's in the archive doesn't say anything about commandline options... If someone can provide me with commandline options for telling the program to do 32m and to quit when done I can try to make a control program for infinite loop or x number of loops.

[deathstar13]

i wish i could help more but mabey this can.i found a place with many other pi programs with the source codes and binary for super pi.
ill try a few and see if they can loop.
i did find one that tests to 1024m wich may be at least a 24 hour run might be our program alredy done.

http://www.myownlittleworld.com/pi/pitable.html

[macklin01]

Thanks for the note, deathstar. I do believe the thread belongs here.

As for how, it really all depends. Command-line seems like the best way to go, if possible. I looked at the program (version 1.1e), and it seemd terribly windows-centric.

My next idea would be to edit it's source code. Unfortunately, it doesn't appear to be open-source. So, you'll need some sort of wrapper program, instead. (And that's a bit beyond my programming expertise, as I generally don't do much Windows programming.) Especially because you have to manually click OK to start the calculation. (A very bad design decision on their part.) -- Paul

[Mr.Guvernment]

well

how long does one instance take to run?

could you sue task manager to start a new task every X minutes / hours ?

[Drec]

you guys this

http://numbers.computation.free.fr/C...ram/pifast.zip

runs with command line, it also says its max is 12000m which is wayyy more then enough.

EDIT, i can't get it to run though. lol

Originally Posted by Mr.Guvernment

well

how long does one instance take to run?

could you sue task manager to start a new task every X minutes / hours ?

that wouldn't work cause it would just open up superpi not run the calculation thing.

[DaWiper]

What if we made a program that would make the same calculations 10, 100 or even 1 million times comparing those one milline(or whatever) results with each other to check for errors. Doing just that in a infinite or selected number of loops. That would be really easy to make. Even I could probably pull it off in Blitz Basic. It would probably be extremly easy to make it in0 C/C++ for those that can do it.

This is a "universial" basic version of the simplest stability check prg. that I can think of. Why does it have be so "advanced"? A cpu doesn't care if it's hard or not. Bits are bits anyway...

Universal basic example;



loop:
randomize timer; REM Whatever the current basic language require to randomize RND to the clock timer.
e#=rnd(1) ;REM This selects a number floatingpoint number between 0 and 1 var. 1
e2#=rnd(1) ; REM This selects a number floatingpoint number between 0 and 1 var. 2
for tell=1 to 10000000 ; REM start a counter for 10 million repeats
res#=e#+e2# ; REM add the two values
res2#=e#+e2# ; REM add the same values again
If re2s#=res# ; REM Check if they match
ok=1 ; REM If they do then OK
else
ok=0 ; REM If they don't then not ok
end if
If ok=0 ; REM If ok equals not ok(0)
end ; REM quit program
end if
next tell
goto loop ; REM Repeat loop if all is okdokey.


I've not tried the code. But this would in theory check the answer 10.000.000 times before changing values. This program would in not put any strain on the memory but mainly the cpu. (theoretically)

Please tell me if I'm way off.....

[deathstar13]

you guys are talking way over my head here.

but a few things we need to keep in mind in doing this.

it needs to be able to report errors as it does now,what would be better would be error reporting and count and doesnt stop the run.

i also feel keeping how the program works and its internals remain un molested due to its rather accurate ability.
i feel in the end just making it loop is gonna be the best route,even if that mean not getting error counts.long as it reports as it does now.

i dont want to make this complicated on anyone attempting to make this happen

btw im still searching for the source code.
ive also looked into useing the other programs i linked but most dont work and if they do there is no error reporting.
i feel if the program itself is molested too much it will end up being just like prime imo.
anyhow i think you guys get what im saying and i think doing this is simple as possible is the best route.
GL and i cant wait for any developments!

[deathstar13]

btw here is kanada's homepage the guy who wrote super pi:http://pi2.cc.u-tokyo.ac.jp/

even if i found the source i dont think i would be able to recognize it

also i know who does have the source code this guys is who wrote the mod1.4 version.
http://board.win32asmcommunity.net/i...profile;u=3686
http://board.win32asmcommunity.net/i...profile;u=3686

if you look at his profile there is an icq number so anyone may wanna try and contact him as ive never used icq

n111co@hotmail.com -his email address. im mailing him about this code.he is also a member over at xs.org so i think he will have it or be able to do this himself.

[DaWiper]

If your up for beta testing I can try making a a ultra beta version tomorrow... With the abilities to stop/not stop if error. I can give you a pm when the first alpha is ready. However I'm not a hardcore coder, I just know some "stuff". Give me a pm if your interested....