Phenom II 6 GHz+ OVERCLOCK

[chew*]

X2 cards weren't designed to just save space over 4......

I'm not to keen on the technical portion of it but lets put it in laman terms, the shortest distance betweeen 2 points is a straight line.

If the distance between both GPU chips is shorter on X2 versus 2 cards there will be a performance boost, we see this in benchmarks already.

Shorter paths, less resistance = gain.

They implement the same philosophy on reference boards. Grouper, manta and the 790gx amd reference boards all had the PCI X at the very top slot almost touching the NB and the NB was oriented very close to the CPU. It's done intentionally.

[demonkevy666]

X2 cards weren't designed to just save space over 4......

I'm not to keen on the technical portion of it but lets put it in laman terms, the shortest distance betweeen 2 points is a straight line.

If the distance between both GPU chips is shorter on X2 versus 2 cards there will be a performance boost, we see this in benchmarks already.

Shorter paths, less resistance = gain.

They implement the same philosophy on reference boards. Grouper, manta and the 790gx amd reference boards all had the PCI X at the very top slot almost touching the NB and the NB was oriented very close to the CPU. It's done intentionally.

I didn't even think of that. well also isn't it cheaper to make since it's one PCB,

[Ghostbuster]

Shorter paths, less resistance = gain.

Resistance is hardly the effect.

Shorter paths = less delay = possibly higher frequencies

[chew*]

Resistance is hardly the effect.

Shorter paths = less delay = possibly higher frequencies

thx was under the impression that that the longer a single path is however there is more resistance. Just as with anything a .001mm copper strand has less resistance than a .002mm copper strand just the same longer strands work on the same principle or maybe i have it backwards?

Like i said I'm not to keen on it, but i understand why it works.

Yes its cheaper also.

[skiman]

If the distance between both GPU chips is shorter on X2 versus 2 cards there will be a performance boost, we see this in benchmarks already.

Shorter paths, less resistance = gain.

To the best of my understanding, the issue is not electrical resistance, the issue is capacitive reactance between data lines. Crosstalk between data lines on a pcb, is a function of distance between data lines, length of data lines, and the frequency of the data lines. While keeping the data line trace at a fixed width, cross talk will then be a function of length and frequency. In decreasing the length of the data run by co-locating two GPUs on a single card, one can increase the data rate (frequency) between both GPUs before crosstalk becomes a factor, thus failure. In all serious, at the very high frequencies of data transmission in todays PCs, mere millimeters make a difference.