They -will- keep getting bit in the ass with yield issues is they keep selling 500-600mm^2 dies. Maybe not as bad as this generation with TSMC's own problem adding to the issue, but they could compete on price much better if they could squeeze the same performance out of a smaller die
They need to be 'bit in the ass' by yield issues before it becomes an issue. When it interrupts their ability to sell parts as they normally do then it becomes an issue. Them shipping A3 for launch parts is normal, their being constrained by yields early in the life cycle is normal, when it starts to impact their bottom line then it becomes a bite in the ass. Their is a shocking disparity between enthusiast analysis and business reality.
Using GPGPU PPW to legitimize the power draw of a gaming card is fatuous.
And you have so deemed that it is to be a gaming card explicitly? Did you tell nVidia that you have so decided it so it must be so? It doesn't appear that they got your memo dictating what their products are supposed to be
If NV wants to become big in the GPGPU market that's fine, but they need to keep their GPGPU products and their gaming products separate instead of cramming a compute card and a gaming card into one die, creating a bloated chip that most of GTX480 buyers will never use fully
I can't even figure out what you are trying to say here, are you implying that nVidia should make a vector processor and use that seperately from a GPU to reach their intended goals? Or are you implying that they should nigh double their R&D to have a seperate Quadro and GeForce line? Either option is significantly more costly on the business end then doing it exactly how they are now. As far as not using the chip 'fully' and it being a waste of die space, how many people that purchased a 5xxx part are using DolbyHD output via HDMI? That is a quick generic example, but the general point is that both companies do have die space that could easily be considered 'wasted' as *most* of their users don't put them to use. That doesn't mean it is a bad idea, in any way shape or form, for them to have the additonal functionality. You may think it is, and to that I would say why is it you don't own a multi billion dollar GPU company yourself and show them how it is done
Why can't NV make a good gaming card AND a good compute card?
That would be counterproductive on so many different levels. One of the reasons is that nV is actively pushing to get GPGPU used by more in the mainstream. Chicken and egg. The other is the staggering increase in R&D costs associated with such a strategy. Right now large portions of the costs of developing each part of the GPU are shared, those would have to be doubeld if they were to split the lines up. 3DLabs, Glint and SGI to name a few used to try that more specialized approach, nVidia ended their businesses in those markets using their single monolithic die approach. The R&D warranted by a niche part can't compare to that used for mass market offerings. Going with seperate dies for each task with nigh no performance advantage is a good setup for failure.
It winds up with most of GTX480 buyers having tons of compute power they don't need/want, and most quadro/tesla buyers with a bunch of gaming prowess they'd just assume replace with more compute power.
How many GTX480 owners do you know that don't want the compute power? Quadro is still primarily a 3D workstation card, so yes, most of those users still do want the "gaming" power that it has to offer. As far as Tesla buyers go, I would simply say they can go with the part that offers superior performace/watt that doesn't have GPU functionality on it, oh wait, guess that doesn't exist. I think you will find that people who purchase Tesla parts have no problem with its' price/performance or performance/watt metric as the scales of economy have made GPGPU performance its' own market that nothing effectively competes with. On that front, nV has clearly proven its' utter superiority. The only question at this point is how it will work out for them in the general consumer/gamer market. A whole bunch of people on this board seem convinced that it is a failure. This forum also overwhelmingly thought the 4xxx parts were superior to the last generation nV offerings while nV outsold ATi 2:1. If nV sees a large sales decline and has difficulty in turning a profit using their current strategy then you can be sure they will reconsider their approach. To date, it has worked out very well for them.
It's like trying to build a sedan with a pickup bed and 4WD.
No, it's more like developing a high performance engine and dropping it into your regular cars to spread out the R&D costs. Worse fuel mileage, more weight, higher cost and rarely do you hear anyone complain outside of fans of the competition.
Facebook
Twitter