Associate
- Joined
- 30 Sep 2005
- Posts
- 144
64 bit precision requires a floating point format, doesn't it? That's what Carmack indicates in his quote, and that's what engine developers are talking about when they mention 64 bit rendering - additional precision achieved through FP16 render targets. In other words, HDR.Zefan said:It's not actually anything to do with the 3d cards... you can run 64bit textures on a 9800pro or whatever. It's the difference where the cpu comes in to shift textures around that matters. I have no idea why they're boasting on their site about it as all cards will support it.
So the original poster's friend is partly right. Cards below the X1xxx series lack floating point blending support, so they lack the ability to have 64 bit colour precision.
You can see the difference in visual quality, today, because developers can achieve 64 bit precision through the FP16 format. So pick any game that uses floating point HDR.HangTime said:If we do get 64bit 3d accelerators, it will be interesting to see what kind of performance it gives (and difference in visual quality of course).
I'm not sure what the benefits of integrating this precision throughout the pipeline would be, however.
EDIT: BTW, this has nothing to with 64 bit, as it's used in CPU's. 64 bit there indicates the size of the CPU's internal registers, so the maximum amount of memory that can addressed (and the largest integer that can be processed).
Last edited: