CUDA, nice in theory. I'll wait until more real-world apps utilize it before I care though. I'm sure AMD/ATI can come up with something similar given time. Who knows, maybe MS and/or Linux geeks will be able to work out a unified set that will leverage both without being a pain. Right now, however, it's like writing machine language. Great for compactness and / or speed to be sure. The problem is that you have to rewrite your code for every vendor. Maybe do it like BOINC does? Hmmm...

As far as the XDE 500, PFFFTTT! Anyone with a decent (read 40" or larger) HDTV should mock it. Way overpriced for a simple up-converter that's getting very mixed reviews, it's not worth it in my book. I can buy a 40$ up-converter that performs nearly as well (especially through a good receiver using a REON chip, or a good TV) as this thing. The only market it might threaten is the 200$-1000$ DVD player market, but most of those people are moving to Blu anyway. Almost every HD disc player, from the defunct HD-DVD to the still running Blu-ray players does up-conversion competently. All Toshiba is doing here is trying to milk their DVD patents.

All in all, of these two ideas, CUDA shows promise and the XDE showcases Toshiba's failure.