Wow, that's a heavyweight there! How much?

Interesting acquisition indeed. I can see some clear advantages to having mental image in-house; however, historically, I'd still have said it might not be worth the money. But NVIDIA is going to end the year with $2B+ in cash, so I'm sure they figure they might as well use it for strategic acquisitions like that one.

I also ponder whether more minor and/or major acquisitions are going to follow. There's only so much NVIDIA can spend its cash on ($300M+ cash flow/quarter; that'd be overkill for buyback imo). At least they're still hiring super-aggressively:
- January 2005: 2101 employees.
- January 2006: 2737 employees.
- January 2007: 4083 employees.
- November 2007: 4609 employees.

Are we sure NVIDIA didn't misunderstand Moore's Law, and presumed it to refer to numbers of employees?
_EDIT_: As for Mental Images, a bit of gooling around tells me they had 100-250 employees (presumably much nearer to 100) with $10-50M revenue. Still, not small.

Hi think, their most important motivation is to gain access to some world-class raytracing expertise in face of the challenges that will come with Intels Larrabee.

So does this mean that ray tracing might not be a fools errand? :)

I think it's more about catering the professional market in general, not specifically raytracing.

Mental Ray only uses it selectively, like all good renderers, and what makes sense for offline rendering has very little to do with what makes sense for games. So no.NVIDIA probably wants to sell GPU accelerated rendering clusters plus mental ray and support contracts ... could be a profitable business, but seems small potatoes to me for NVIDIA.

Not in the light of this (http://www.tgdaily.com/content/view/35272/118/).

I would have expected Intel to do this rather then Nvidia. Anyway- Competition is good thing.

Wow, this is news! Further, combined with "Chances of NVIDIA acquiring AMD" (http://www.tgdaily.com/content/view/35289/128/) , this could make NVIDIA become a real multi purpose cpu/gpu platform competor to Intel in the long run.. It all goes multi-purpose anyway, but at least this is perfect news for all raytracing enthusiasts!realtime final gather vs. cube map arrays, anyone?