From: marcov@stack.nl   
      
   On 2005-08-03, Skybuck Flying  wrote:   
   >> > So far all holes plugged... bring on them holes ! :)   
   >>   
   >> IMHO you showed that the performance impact can be less than 100%. That's   
   > all,   
   >> and was not really new to me.   
   >   
   > Well it's clear from this post you don't believe in this idea and you are   
   > trying to find any little damn excuse to diss it.   
      
   Well, there are two things clear to me.   
   - your initial proposal has not much details, and you don't examine possible   
   problems.   
   - If sb tries to put holes in your defense, you reply vague and when sb insists   
      you act defensive.   
      
   Note that the first will already be a problem for the patent application,   
   _even_ in the USA. And while USA rubberstamps a lot of patents, the so   
   called trivial patents are often really unenforcable. You will need millions   
   to prosecute people who don't want to pay you. (can you imagine what sueing   
   Intel will cost? Or even AMD?).   
      
   Without enforcing, a patent is something that only costs you money, so   
   either your patent application has to be very specific so that challenging   
   it will be hard (and even then you risk that you must litigate to enforce),   
   or, if it is broad and vague, you essentially become a litigation company.   
      
   > That's not how innovation works.   
      
   Yes it is, and even in science at large. It is called "peer review" and having   
   a defense to annoying questions about your findings is at   
   the heart of nearly every academic graduation.   
      
   > There are many things unknown which need to be figured out.   
      
   And if I tease you a bit you are unable to provide a reasonable defense. What   
   do   
   you think Intels lawyers will do? :-) (if you get out under the Mount Everest   
   on   
   paper of prior art that they will drop on you to process at all to make it to   
   court,   
   and find sb who fill finance that)   
      
   > But I am confident that I can build a CPU with this encoding like it truely   
   > hasn't been done before.   
      
   Of course. Please announce it in the proper CPU design NGs if you are done.   
   I wish you a lot of luck.   
      
   P.s. As a _serious_ side note I think you might be able to make money from this   
   easier in specialised CPUs (DSPs?) than in General Purpose CPUs. Moreover,   
   making a competing specialised CPU in some niche market is way easier than a   
   competing general purpose CPU (AMD is trying for years, and still only has a   
   5% piece of the pie, most others are dead). Both because of costs, and   
   market/scale mechanics)   
      
   So try to think of all kinds of analysis that must have adjustable precision,   
   initially limit   
   your design to some high bound (e.g. 256-bit) and see if you can make a certain   
   specialised analysis CPU on a tighter transistor count than a competitor.   
      
   > But that's a risk I am willing to take :)   
      
   Make sure you have access to deep pockets. The risks are huge.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)