? about dual Athlon MP 1800s

Some joker is selling them on eBay for about $210 the the mobo combo w/pair included!

How does this setup rank for gamers or people that want to make digital media, versus something like an nForce2 with XP 3000 (333 FSB, not 266)???

Any input would be appreciated.  I know that this would generally be a server setup, but it makes me wonder.

Also, one other question.  Do the dual MPs = double processing speed?  How fast are they opposed to today's fastest CPUs (not the 64bit ones)???

Think of it this way, you can expand the flow of water in a pipe, but you are not making water flow any faster.

If you HAVE to build a computer now you'd be NUTS not to build a 64bit system. 64bit computing is around the corner and it's the next big jump so I'd either get the amd64 3000+ (which is actually not too expensive) or wait a few weeks and buy then.

No need for dual processors unless you're building a server.

Also, nforce2 and 3000+ will blow it out of the water and you may be getting scammed anyways because MP's are just unlocked xp1800's that allow you to change multiplier in the bios. The nforce2 mobo's automatically unlock the cpu anyways(the old palamino cores anyways)

I am going to build the 64 bit system for sure.  However, I was curious about the dual MPs.  I did understand the subtle difference between the XP and MP, but wanted to know if the dual CPUs make the system run twice as fast.  That is something that I never quite understood.




Dual CPU's would make a part of the system run twice as fast. You'd probably get slowed down somewhere else, unless the rest of your comp was equally fast.

Is this for a home pc? or a server?

When you have dual processors, you are doing parallel processing. This means that you try to split running processes into parts that can be done at the same time on different processors. If a program requires the output of part A before it can start part B, then it can't do A and B at the same time.

So the increased throughput due to adding another processor is dependent on the amount of work that can be done in parallel as opposed to in series. If you can keep all the processors busy all the time, you will have more or less twice the throughput. But that's rare. Usually you will get significantly less than that. And there is also some overhead costs for passing messages between the processors and figuring out how to split the problem into parallel pieces.

Andrew Yao is interprocess communicating with the

It was for a machine to work with digital media.

I am going 3000 instead.