Tesla

[pixym]

héhé, Yoon,
Your way to argue… is not bad at all!

[YoonKyung]

héhé, Yoon,
Your way to argue… is not bad at all!

In program languages the word, argument has different meaning, a variable which is passing values to a fuction.

Any way, I did very interesting calculation. And the result was,
If we can use 400 PCs which has a GTX480 or 800 PCs which has a GTX280,
we can do real time HD, 256 samples/pixel quality rendering.

[Krisonrik]

héhé, Yoon,
Your way to argue… is not bad at all!

In program languages the word, argument has different meaning, a variable which is passing values to a fuction.

Any way, I did very interesting calculation. And the result was,
If we can use 400 PCs which has a GTX480 or 800 PCs which has a GTX280,
we can do real time HD, 256 samples/pixel quality rendering.

If what you say is correct and by Moore's law, we will be able to do such thing in few years. That is if the computing power does double every 18 month.

[YoonKyung]

If what you say is correct and by Moore's law, we will be able to do such thing in few years. That is if the computing power does double every 18 month.

If you reduce the argument, 'render target' to 640x480, then we can do it with a rack of GPU cluster which has 133 GTX480s, now.
Since 640 x 480 scene has 0.3 M pixels, and we needed 400 GTX480s to render 1M pixel scene,
we can render 0.3M pixel scene in real time with 133 GTX480s.

[Feniks]

Hi YoonKyung,

tnx for the roundup

Since Octane isnt an animation package, I wasnt really referring to 25fps animation,
realtime/delayed more like in 1-4 seconds.

That should allready put the pricetage a bit lower no ?

[YoonKyung]

Yes, you can do it now, with one GTX2XX(1,5~2,0M samples/second),
if you reduce render target to small one, for example 640x480(0.3M pixels), and reduce max samples to 4,
then you can get delayed real time animation rendering.