The way the design and manufacturing process works is this:
1) R+D continues at a roughly uniform pace.
2) At some stage, *based on the current market situation* the technical and marketing managers will discuss formation of viable product-lines. This include both the new technology itself, and the two or three "revisions" they can extract from it (die shrink, clock speed bump, memory size increases etc).
3) From this plan a product is designed, "on paper" and eventually taped-out in prototypes. The marketing plan is further refined here, based on the viability of the prototypes (yield, performance, heat output, production costs etc). If the initial design was too ambitious then this stage can be costly and lead to delays (a la ATI in recent cycles).
4) The products are then produced en-masse, and distributed to suppliers. Future revisions, as planned, will start directly from stage 3, and the prototyping process will generally be much faster and cheaper as the raw technology is already mature.
So, yes - of course ATI and/or nvidia could release a 'monster' which was say 50% faster but economically un-viable (low yields and hence expensive, high heat-output etc). However, they strike a balance as with all things. From process 2 to seeing your card in the shops you're generally looking at 6 to 12 months, which is a quite impressively short period of time to my mind (much faster than with CPU design). R+D continues along many different threads, and can often be looking 3 years or more into the future, at the front-end.
The point I was making earlier is that the *driving force* behind this all is stage 1 - the R+D. This determines the technological advancements they have in their arsenal when planning products (stage 2). So, even if a 'new' card is not released for 18months due to the nature of the market (ie lack of competition), the technological advancements are still progressing.
Last edited:
Here's a question - are there any games around atm bar Crysis, that an 8800gtx won't run amazingly well?
