On Mon, 2004-11-29 at 00:15 -0800, Erik Steffl wrote: > > When I start to understand the registers that a piece of hardware > > makes visible I start to understand the architecture of the product > > itself. If NVidia believes has an advantage over ATI in some area then > > possibly ATI would copy that and improve their performance. Neither > > company probably knows the intimate details of the inner workings of > > the other company's chips and hence thinks they have the advantage. > > is it really that hard to reverse engineer the binary only driver? > Does it make any real world difference for competitors? (my guess it > that it makes no difference and that companies are either fooling > themselves or just don't want to bother with legalities of releasing > info parts of which might be owned by other companies etc.) For a sound card, it's easy. James Courtier-Dutton reverse engineered the Audigy LS in a few days. But a 3D driver is literally like rocket science. It's _hard_. This was discussed on the "open source graphics" thread on LKML. The consensus was that it's essentially impossible. Lee
