jrodga2k5 said:
You know that whole Power issue really puts me off a bit! I mean I'll upgrade my power eventually but the fact it requires 1x8pin and 1x6pin or 3x6pin PCI-E connectors is bordering on the unthinkable! Thats just to overclock the card as well
I think the whole power requirement issue with R600 is a bit overblown, to be honest; all of these 240W or 270W figures floating around should be regarded with a great deal of suspicion.
In the video of AMD's recent press conference someone asks if it's true that R600 will use 270W. The reply is that in their demo setup each card is using roughly 200W. The demo setup was an R600 Crossfire rig doing GPGPU-type calculations, i.e. with all of the shaders working absolutely flat out to produce a total of 1 Teraflop.
If the card runs happily with two 6-pin connectors that means that, at stock speeds, the absolute theoretical maximum peak power output even for the tiniest fraction of a second must always be
comfortably below 225W. (225W is the most you can pull through two 6-pin connectors plus the slot). So, all in all, it looks as though the card probably won't ever pull more than 200W unless you're overclocking. If you do overclock then the consumption will get close enough to 225W that they make you use different connectors in order to maintain a sizeable margin for error.
For practical purposes I suspect that a 750W power supply will be enough to run R600 Crossfire at stock speeds: the graphics cards will be pulling about 400W
peak, and I doubt the rest of the system will require another 350W. Remember there are people running 8800GTX SLI on 620W PSUs. If you're a rabid overclocker with a quad-core CPU than 850W would extend the comfort zone.... I'm still uncertain as to who is supposed to derive any tangible benefit from a 1000W power supply.
The lack of 8-pin PCIe 2.0 connectors on most PSUs is very annoying, though.

I'm hoping that the 850W Silverstone Olympia PSU (due out in early May) will have the right connectors and be of reasonably good quality. We'll see.