How did Nvidia infringe 3Dfx texturing patents?

[darkblu]

ok, i take Dave's hint and i'll bite.

the thing that pissed me off at the time i was reading that 3dfx patent was that the sucker covered all hardware aparata which had anything to do with "combining signals in a cascading manner through the means of a arithmetic/logic operatior at each junction", i.e. anyhting which effectively could be described as

A   B
|   |
-op-|   C
    |   |
    -op-|
      |
     ...
      |
     out

got covered under this patent's claim. needles to say, there had been devices long befor 3dfx used that in their sst1, even in the PC, which somehow "combined signals cascadingly, applying a binary operator at the junctions", e.g, your PC interrupt controller is one.

 

[radar1200gs]

Yes, the patent would likely have been ruled unenforcable due to prior art by the courts if challenged.

Perhaps nVidia decided such a challenge would prove too expensive, but, you have to consider whether 3dfx could have survived long enough for such a challenge to mean anything.

Given the prior art and 3dfx's financial standings I would definitely have challenged.

 

[Dave Baumann]

The problem was that it looked like it was going to be enforced. IIRC there is usually a pre-trial which looks at the patents to see if they have merit, before going to trail over the damages - I believe 3dfx won the pre trial.

 

[radar1200gs]

true enough, but, decisions can be appealled and overturned and you have to wonder how thorough the pre-trial was given the appalling state of patent process in the USA at the moment.

The judge may have simply seen that 3dfx appeared to have a patent and never checked on whether due process was followed when granting the patent or not.

Given the prior art etc, it would appear 3fdx probably never had their patent application checked as thoroughly as it should have been before approval.

 

[davefb]

was there proper prior art ? looking back i thought that the 3dfx patent was 'new' in that prior solutions were much more "hard-coded" in the silicon. the 3dfx version was more flexible . maybe it was 'obvious' but if it was so obvious , how come nobody had done this before ?


-dave-

err we are talking the 343 patent aren't we ? ?

btw, gary t is named in a jan2004 nvidia patent have we discussed that one ?

 

[darkblu]

was there proper prior art ? looking back i thought that the 3dfx patent was 'new' in that prior solutions were much more "hard-coded" in the silicon. the 3dfx version was more flexible . maybe it was 'obvious' but if it was so obvious , how come nobody had done this before ?

don't know abot prior art from the graphics industry. but i do know that such cascading interconnections of ASICs for the sake of handling greater number of inputs than a single of those ASICs would have been able to handle (at a time), through the means of a "combiner control signal" at each junction have existed long before any one of the 3dfx' sellers-smith-tarolli trio took over the task.

err we are talking the 343 patent aren't we ? ?

we are. as about the 102 patent, nv may have really infringed that (i.e. the dithered trilinear).

 

[davefb]

i guess thats the thing isnt it . solve something different by taking what people are doing elsewhere and if it's a better idea apply it to something knew

 

[darkblu]

i guess thats the thing isnt it . solve something different by taking what people are doing elsewhere and if it's a better idea apply it to something knew

allow me to disagree. you can't take the wheel and patent it in industry 'x' just because a wheel hasn't have any direct use in that undustry so far.

and in the 343 case it even isn't a different industry - ASICs have been know to be used that way for many a decade. know what, the really ridiculous thing is that if you, for instance, decide to take the numer of an IRQ received from the PIC in your very own PC and display that number onscreen as some color - you'd immediately violate the 343 patent. funny, no?

 

[davefb]

? doing what with this wheel ? i bet if i found a 'revolutionary' way of making semiconductors by spinning a wheel , then i could , couldnt I ?

-dave-

gah , wonder if "application of grammer checker on bulletin boards' is worth a patent .. knew<>new , grrr

hmm, what about TV using wheels, baird style

 

[darkblu]

? doing what with this wheel ? i bet if i found a 'revolutionary' way of making semiconductors by spinning a wheel , then i could , couldnt I ?

what i've been trying to hint at is that there's nothing revolutionary in cascaded interconnection of identical ICs. in the wheel-in-semicon-industry analogy it'd be analogous to patent the use of wheels on the wafers transportation wheel-tables.

gah , wonder if "application of grammer checker on bulletin boards' is worth a patent .. knew<>new , grrr

not only that it's worth (as you'd be making quite some money from it), but it's also doable - it'd would go through the USPO like a breeze.

hmm, what about TV using wheels, baird style

*me slaps forehed and rushes straight ahead to the USPO..*

 

[Simon F]

? doing what with this wheel ? i bet if i found a 'revolutionary' way of making semiconductors by spinning a wheel , then i could , couldnt I ?

Quite possibly. (That's in answer to "could I patent it".)