ghetto dual cpu?

i was just thinking, and now i'm in a war with myself over what would happen.

the pins of the cpu are soldered and exposed on the back of a motherboard (removing any structure panel). what if i put a cpu and correctly aligned it to the pins there, effectivly splitting the cpu socket to accomodate two cpus? assuming i had the juice to power both, what do you think would happen? would instructions come back half finished as only one cpu did half the work? would only one cpu operate, that one being the one along the path of least electrical resistance? would niether cpu operate as some pins are least resistant on one side versus the other? would both cpus operate effectively, send different instructions to both, and them both processing it and returning it (if sent back in the correct order?) would teh motherboard burn out with the stress of both cpu's on its back? what if i had a capable motherboard?

what do you think?
crazy, but hey, i've got two pentium 4's and two athlon xp 2800+'s, i think i'll try it later this week!


so, what do you guys think will happen?
 
You'd have a burnt out motherboard, a mouthful of blue smoke and a couple of dead CPU's. Do me a favor and just save up the money, and buy a dual core CPU.

But if you are willing to do this, video tape it. Please.
 
my processor stash includes:
1 AMD k62
3 AMD athlon classics, 500, 600, and 900MHz
3 Pentium 1's
1 486
2 Pentium 4 2.8Ghz w/HT
2 AMD athlon xp 2800+
1 Pentium III, 700mhz
2 Pentium II 400MHz
1 Pentium 4 1.5GHz

its a wonder what people throw away, eh?

i would use the p1's, but i don't have a motherboard for them.
 

darkguset

Distinguished
Aug 17, 2006
1,140
0
19,460
Nothing will happen. The caps on the mobo will probably not have enough juice to drive two CPUs, and even if they did, all the errors created by the current mis-distribution would render the system inoperable from the very start.

Fun to play with though, and a cool video for YouTube, not to mention the originality in the practical way! hehe
 

chuckshissle

Splendid
Feb 2, 2006
4,579
0
22,780
:lol: You can't because it will be backwards unless you're talking soldering it with wires. Yeah, videotape it when you do this and post to youtube. I wanna watch this one for sure.
 

sirrobin4ever

Distinguished
Dec 8, 2006
596
0
18,980
:lol: You can't because it will be backwards unless you're talking soldering it with wires. Yeah, videotape it when you do this and post to youtube. I wanna watch this one for sure.

Agreed, and glad that someone finally pointed out that its impossible to do, as the pins on the back will be backward.
 

ajfink

Distinguished
Dec 3, 2006
1,150
0
19,280
:lol: You can't because it will be backwards unless you're talking soldering it with wires. Yeah, videotape it when you do this and post to youtube. I wanna watch this one for sure.

Yeah, looots of soldering :)
 

Major_Spittle

Distinguished
Nov 17, 2006
459
0
18,780
LOL, a better use of your time would be reading about electronics and how a computer works.

It would be cool just to attach a processor on the back of the socket with nonconducting adhesive and tell people on the internet that it works. :lol:
 

ImmortalZ

Distinguished
Oct 24, 2003
27
0
18,530
If the CPUs are perfectly matched, they will both operate, provided the power is there. The problem being that the branch prediction and such logic might create different ways of doing stuff for each of the CPUs. So in the end, you will just create major collision on the buses.

I'd venture to say that it might boot up, depending on the architecture. But you will not be seeing any performance gain at all. Then again, they might not, as the BIOS POST checking the CPUs will immediately get a scrambled answer because of the CPUs reporting things simultaneously.

If you get the pinouts right (they will be mirrored on the underside - thats a lot of wires you'll need to solder), they won't burn like people here say.
 

mrmez

Splendid
LOL thats hell gangsta
Obviously it has NO chance of working... Even if u soldered wires correctly etc.

The mobo has NO capacity to deal with the extra CPU... end of story.

A better use for ur time? Have a wank :lol: :lol: :lol: :lol:
 

JMecc

Distinguished
Oct 26, 2006
382
0
18,780
Each CPU would be getting the same signals from the mobo and would exec the same instructions. They would then output the same results on the bus, netting no gain. This is assuming that the acceptance of a small current 2 ways instead of 1 does not drop below logic level, yielding a 0 in 1 or both of the cpu's where they should see a 1. No chance of seeing any benefits from this.

Jo
 
i was just thinking, and now i'm in a war with myself over what would happen.

the pins of the cpu are soldered and exposed on the back of a motherboard (removing any structure panel). what if i put a cpu and correctly aligned it to the pins there, effectivly splitting the cpu socket to accomodate two cpus? assuming i had the juice to power both, what do you think would happen? would instructions come back half finished as only one cpu did half the work? would only one cpu operate, that one being the one along the path of least electrical resistance? would niether cpu operate as some pins are least resistant on one side versus the other? would both cpus operate effectively, send different instructions to both, and them both processing it and returning it (if sent back in the correct order?) would teh motherboard burn out with the stress of both cpu's on its back? what if i had a capable motherboard?

what do you think?
crazy, but hey, i've got two pentium 4's and two athlon xp 2800+'s, i think i'll try it later this week!


so, what do you guys think will happen?

wouldn't work, you cant "stack" them like that and you would also (if anything) need bios support etc
 

Scougs

Distinguished
Mar 10, 2006
398
0
18,780
I think if you did this, you would ruin all the hardware invovled.
A single socket motherboard is simply not designed to operate two CPUs. The chipset would probably not talk with either one and the board probably couldn't supply enough power for two CPUs.

If anything did work, both CPUs would have to be the same speed. There would be no advantage to this if it worked because both CPUs would perfrom exactly the same operations at exactly the same speed and the at the same time. Thus, no performance change.

It takes a group of electrical engineers a lot of time using complicated software to design a microprocessor system. You can't simply jerry rig things and get them to work the way you want.
 

darkguset

Distinguished
Aug 17, 2006
1,140
0
19,460
To answer it simply, if it did work, don't you think AMD and Intel would have done it about 20 years ago? If it was that simple we would have had 50 core CPU's simply by stacking one chip on the other. Obviously that is not the case.
 

CaptRobertApril

Distinguished
Dec 5, 2006
2,205
0
19,780
It takes a group of electrical engineers a lot of time using complicated software to design a microprocessor system. You can't simply jerry rig things and get them to work the way you want.

Aw, why not? It worked fine when I was putting a dual carb setup on my 1967 Honda CD175 Benly! :roll:
 

darious00777

Distinguished
Dec 15, 2006
687
0
18,990
The microintrusions into the electron stream would scramble all signals beyond recognition to all operations.

If, however, you were willing to manually add voltage to certain parts of the motherboard, custom write the BIOS, and can find a way to mount a heatsink fan to the opposite side of the motherboard, then you might have some luck.

To maintain signal cohesion, assuming a 266 front side bus on the CPU, you would need to drop the signal to 133. In doing so you may be able to get away with dropping the voltage on the CPUs enough to maintain a level of stability with the CPU to allow calculations to be done, and having enough juice in the capacitors already installed to handle the power requirements generated by two CPUs.

Beyond all this, besides the BIOS, you would also need to rewrite the Northbridge code to allow for the dual CPU usage.



In short, have fun, and follow your dreams. :lol: :p



To everyone else, don't destroy the guy's idea.
 

darious00777

Distinguished
Dec 15, 2006
687
0
18,990
Fair enough. :p

This whole thing reminds me of when I wanted to build a robot that could help me take over the world. I gave up with I realized I didn't know how to make a pulley. :lol: