Custom Mobo estimates..

[Schweppes]

Alright.. This WILL sound weird...and I'm not the most..hardware inclined fellow on the planet. But! I'm wondering if anyone here could give me an estimate on the price for a mobo that will support 30 p75's, and no your not going crazy I did just say 30 p75's. IT's a project that I'm sure will cost alot. But if it works out will be awesome. any ideas?

 

[santaclaus]

maybe if you describe what the hell a p75 is, then maybe some people herrre can assist you in your little project

 

[Odeen]

Pentium-1 75mhz. Either a 75mhz FSB, or, most likely, 1.5 multiplier on a 50mhz FSB. Either way, keeping thirty processors happy on a single piece of silicon is not an easy task, if achievable without pouring millions into R&D.

 

[Gerbil333]

Why on earth would you want to put 30 Pentium 75's in a system? I imagine a single Opteron would outperfm that...

 

[Peter]

Yes. One single current CPU on a stock chipset would blow that thing away.

Besides, you're not only asking for a custom mainboard, you'd need a custom made and HUGE chipset as well.
If you have 30 P75 there catching dust, why not collect some more and tile the bathroom in a truly decadent style?

 

[fell8]

Let's not forget the custom OS that can run 30 CPUs.

Methinks 30 brand-new high-end rigs and sufficent networking hardware would cost a fraction of what you're proposing here. Good luck!

 

[Schweppes]

Originally posted by: Gerbil333
Why on earth would you want to put 30 Pentium 75's in a system? I imagine a single Opteron would outperfm that...

A single p4 would easily out perform it yes. This whole thing would only equal up to 2.2ghz. I told you It was a crazy idea. But I sure it's possible. It shouldn't make any difference between a quad p3 and a 30 pentium1 75mhz(for our absendminded friend above) You don't need a custom OS for a quad why would you need one for my 30mobo ?

 

[megamanS]

For multiprocessors, you'd need software that can divy up the work for each of the 30 cpus. Changes in cpu architecture mean that a new cpu is even faster than simply being n times more megahertz. So much so that 1 chip doing 30 tasks one at a time would still outdo 30 older chips working together.

 

[Aenslead]

I am sorry to say this, but unless you are an electritian master... or a person with A LOT of money and resources, your dream might just end up being that: a dream...

The answer is: no estiamtes can be made because it is "impossible" to make... unless ofcourse, you tie them all up with nods, just like supercomputers... which is also quite expensive, and in your case, quite difficult to use and quite unuseful as well.

 

[fell8]

It shouldn't make any difference between a quad p3 and a 30 pentium1 75mhz(for our absendminded friend above) You don't need a custom OS for a quad why would you need one for my 30mobo ?

Name one OS that will run this single MB that doesn't yet have even a logic model, much less a blueprint or silicon.

 

[tcsenter]

A single p4 would easily out perform it yes. This whole thing would only equal up to 2.2ghz. I told you It was a crazy idea. But I sure it's possible. It shouldn't make any difference between a quad p3 and a 30 pentium1 75mhz(for our absendminded friend above) You don't need a custom OS for a quad why would you need one for my 30mobo ?

Apparently, you're not the most software inclined fellow on the planet, either?

What you're asking is next to impossible because nobody is going to actually sit down for a week with a team of engineers and perform the necessary cost projections to answer your question unless you're paying them for their time. If you were offering a few thousand (or more) dollars for an estimate, I'm sure you'd get a better response. If you don't have the money to pay an engineering firm a few thousand dollars for an estimate, you surely can't afford whatever the cost would be.

A more workable solution would be to build a Beowulf cluster with 30 Pentium nodes.

 

[Aenslead]

in few words: Forget it.