I've heard of %'age figures being given to the general public as yields, but I've never before heard of yields being measured in %'age internally within the industry. Measuring by %'age would be a highly dubious proposition given how little information it presents about the level at which at given silicon is yielding.
One %'age figure can't tell us much in terms of operational silicon and speeds attainable at any each individual voltage level, along with potential modifications via FIB tools to further improve speeds after silicon has been fabbed. Has any form of adaptive bias been applied to the process to skew yields?
What are the results and how has that impacted operation at each voltage level?
If one was to measure yields in %'age you would need a chart of a few thousand different figures to even give a basic impression of how the silicon is yielding. A highly inefficient manner of presenting data to say the least.
Of course it will impact pofits, but nVidia would be well aware that they wouldnt come even reotely close to seeing any profits let alone even breaking even regardless of how well initial NV30 silicon yielded.
The NV30 is initially targeted for the extremely low volume $300+ DIY graphics card purchasers.
The profits from such a market won't be enough to even make a dent in the R&D costs to develop the NV30.
Re-couping development and fabrication costs will come when the NV30 architecture has been transferred to the mainstream in successive revisions.
Facebook
Twitter