TMPLee@DOCKMASTER.ARPA (02/12/89)
I apologize and stand corrected -- didn't think about it closely enough. I now do see that accelerating the region functions would be an essential part of an accelerated graphics unit; without that accelerating the rest (e.g., draw line, copy bits, etc.) wouldn't buy you much. What a wonderful opportunity for a non-clonable product! (I like to pretend I know something about computer graphics since that's what I did for a living 20 years ago, but things have changed a lot since then.) I guess it would be a real toss-up to decide whether you could get more bang for the buck by just accelerating the graphics functions, say, as one enhancement, arithmetic as another, or by doing the Transwarp GS kind of thing that accelerates everything, but nothing super fast. (re SANE inteface to FPU's: sure its a waste of the fast FPU to have the toolbox call interface (surprises me its that large) interposed in between, but it still would be a lot lot faster than the 5 milliseconds they take now. (even if the FPU were infinitely fast, a speed up to 100 microseconds that would result from going through the toolbox dispatcher would still be a 50-1 improvement.) Pushing it to the limit would of course make you want to interface as tightly to teh FPU card as you could, but if you were willing to settle for a little less preserving object-code compatibility would be nice.