Expreview has many pictures of a naked 55nm Zotac GTX 260. The PCB has been redesigned with all the memory chips on the front side and the power module has been upgraded to a 4+2 phase design. There are still 216 stream processors and the clock speeds are the same as the current 65nm version... for now...






Senior Member
The ram move to the front should help RAM overclocking (which is already pretty awesome)
Member
Yea that chip is still on the same real estate.
n00b
Member
n00b
I have no doubt that this is a step in the right direction - slightly cheaper,cooler?, and better overclocking. However, what about when the GTX295 (which uses this core revision)comes out? Sure it will most likely destroy ATI's card, but at what cost, not just to purchase the card, but to keep it running! I'm sure the electricity to bill would start to climb or go through the roof. Although most people that have money to spend on these cards should have enough to pay any bills.
ATI always seem more power efficient and cheaper than NV, but they do need more than 800 SP's to compete with NV in the performance department(Yes I know they are a different architecture and use different programming methods) . Now imagine an NV GPU with 800 SP's, energy efficient, great cooling, with a respectable price (Yes, it is hard to imagine, especially the last part, but who cares).
Senior Member
n00b
danke for the clearing.
Member
n00b
It's a difference in how ATI uses their shaders over nVidia. nVidia uses them strictly for shader ops (and now PhysX support) but not much beyond that. They still stay old-school with most bump-mapping and other effects done in the ROPs (which is why they have such huge numbers still). ATI does most of their image quality stuff with the shader processors. All Anti-Aliasing and Anisotropic filtering runs through the saders as well as bump-mapping and normal shader code. This is what makes it very difficult to compare these cards by raw specs. Theoretically speaking nVidia should stomp an ugly mud-hole in ATI but due to a more efficient way of handling things ATI can hold it's own. Until you actually use an ATI driver then you're pretty much screwed.
I knew ATI used different process to render things etc, but not to that level of detail. Thanks for the insight.
So to summarize what you just said, your saying ATI's stream processors are not 'crapper' or less powerful than NV's, but since they process a lot more data, ATI needs to more than triple the number of SP's than what NV use, to make up the difference in performance.
Sorry if what I just said is completely the opposite to what your were trying to explain. However, if that is what you were saying, I believe that it's a very good way to render and process data etc.
"Until you actually use an ATI driver then you're pretty much screwed." I'm not really sure what you meant by this, but isn't it the drivers that usually let ATI down. My previous experience of ATI's were hampered by their drivers (and a very unstable X1950Pro, although that doesn't really count in this discussion, because it was produced before the move to Unified Shaders).
Maybe ATI can improve their performance by adding some eDRAM on the same package as the GPU. It'll make them a little more expensive, but Nvidia isn't exactly cheap either.
Has ATI removed Vertex Pipelines from their R700 generation GPU's?
http://www.gpureview.com/ati-rv770-chip-151.html