By: Exophase (exophase.delete@this.gmail.com), January 10, 2015 8:51 pm
Room: Moderated Discussions
juanrga (nospam.delete@this.juanrga.com) on January 10, 2015 8:21 pm wrote:
> Traditional GPUs from Nvidia/AMD differ in several ways from parallella,
> but TBR (Tile-Based Rendering) GPUs must be a bit more close.
What could cause you to think something like that? Because Parallela "tiles" its array of cores? No, tiling GPUs today are similar to the non-tiling ones outside of the tiling part. They've got a lot of vector execution width. Nothing like the little scalar cores in Parallela.
> Traditional GPUs from Nvidia/AMD differ in several ways from parallella,
> but TBR (Tile-Based Rendering) GPUs must be a bit more close.
What could cause you to think something like that? Because Parallela "tiles" its array of cores? No, tiling GPUs today are similar to the non-tiling ones outside of the tiling part. They've got a lot of vector execution width. Nothing like the little scalar cores in Parallela.
