informed or not, his comparisons don't really pass muster. you don't know what kind of specs their video codec is capable of. and a server case full of GTX 280's running in parallel is a ridiculously powerful machine. the only real limitation i see is lag. hardware is easy, and when you start talking the numbers you have been (or the eurogamer estimates), investors will give them money to buy more hardware. hardware isn't an issue.I.S.T. wrote:That guy is informed as hell, actually. Look up his posts at www.beyond3d.com 's forum. His name is grandmaster.sweener2001 wrote:the article seems a little uninformed, or the opinion was formed too fast.creaothceann wrote:http://www.eurogamer.net/articles/gdc-w ... rk-articlePanzer88 wrote:they say they can accomplish it over the internet via their amazing new video compression.
there are no specs on onlive's video encoding, so using a current and publicly available codec can't even necessarily give a ballpark estimate. eurogamer's servers probably don't compare to what onlive will have.
does anyone know what one of those tesla super computers is actually capable of? i actually don't.
i understand each of the concerns, and gizmodo has a more in depth playthough experience posted. but making these comparisons to things that are probably nothing like onlive's setup doesn't fly.
i actually hope the service works. whether i jump right in is another thing completely.
And a Tesla is basically a GeForce GTX 280. Yes, I'm not kidding.There's been a lot of research into using graphics cards/chips for normal computations lately(The reason why is the raw computational power of a graphics chip is usually higher than any consumer CPU. The problem is they are too specialized.). A tesla supercomputer basically just has a fuckload of them.
it's just way too early to make any actual educated speculations. if this project were so obviously going to fail, i don't think it would have come this far. the ONLY thing i can think of that had hype like this and failed was the Phantom.