Of course, actually paying for the 2nd chip at the time of purchase of the main card (350$? I think I would prefer a HD5870 at this price, thanks), and having the 2nd chip forever glued to the first is just dumb, IMHO. Even in that case I would think about it because of the extra power consumption and what it would give me on the end, and probably it would end up in the cardboard box. The only way I can think of having a 2nd videocard dedicated to PhysX as a possibility is if you previously had that card, you have bought a new one to upgrade your PC, you can't sell the old one (or you prefer to keep it for back up, or something) and you feel sad about keeping it completely in disuse inside a cardboard box. But is it worth buying an extra card or the extra power consumption? You could get some little fps more on those 2-3 games, maybe. And those 2-3 games can be run on a G200 chip when doing both graphics and PhysX. Having a card dedicated in an exclusive way to PhysX has no sense for starters. and there's always uninformed people and fanboys that could bite.
It's a good way to get rid of their stock of obsolete chips, if they sell the cards to someone. they are packaging a G92 (only for PhysX) and a G200 on a single PCB. That's all there is to it really! It's a very very niche product but there will be people interested in it because they get the benefits of rendering GPU + dedicated Physx GPU without the cost of 2 pci-e slots! There are a lot of people out there using this kind of setup but with 2 cards instead, so they'd probably be interested in this or otherwise those who wanted to go this route but don't have a spare PCI-E slot. The card can have 2 sli connectors because you need to consider the GTX275 as a GTX275, so you can run tri-sli on top of having a dedicated GTS250 chip for Physx processing. So I'd imagine that communications is probably a little faster than having two separate cards doing the same task, but VRAM maybe shared?. So they aren't linked via SLI or anything fancy like that it's just two gpu chips on a single card over the same link. Try and imagine this single card as having both a GTX275 and a GTS250 installed on a motherboard, where the 275 acts as the primary GPU and the GTS 250 acts as a Physx PPU. Well I'd say considering the purpose of the card, both chips simply share the same PCI-E 2.0 bandwidth. How are these two separate chips linked or is it that they dont need to be and that both share noting similar except the PCIe bandwidth? I am very interested i knowing how did nvidia manage to put two sli connectors for Tri SLi, because i recall that the GTX 295 had a single sli connector.
0 kommentar(er)
