Originally Posted by sugarhell
How can you estimate the performance of a new architecture?
Especially if it has a tiled based rasterizer
Oh the demo of a really early sample...
Just wait for benchmarks
By new do you mean GCN 5th Generation?
So they gain 18% increase in performance from a Tiled Based Rasterizer? Sure it was early, but not that early. Vega is coming, what, sometime in May? Had to be pretty close to final results. Of course they could be sandbagging, but that seems like something you wouldn't want to do when demonstrating a card's performance from a PR standpoint. Negative PR is not good PR. These technologies (and assuming the engineering sample wasn't at finalized clockspeeds) like the HBCC and the Draw Stream Binning Rasterizer can surely bring it closer to a 1080 Ti or Titan Xp, but it still would fall short by about 12%. Does that add up to the statements, kind of. However, it has to be priced much better than a 1080 Ti in order for it to seem like a better deal and with it carrying HBM2 it will be difficult for it to be much cheaper than those cards.
Of course I will wait for benchmarks, I just don't want to see people getting upset again like the 1600MHz RX 480 rumors. That whole ordeal was ridiculous, because multiple websites were promoting the concept and I see numerous websites along with Youtubers pushing this story of how it will compete with a 1080 Ti / Titan Xp.
I want it to compete with a 1080 Ti as much as the next guy, and for a good price because that means that I get a 1070 competitor for a good price, but I do not want to be disappointed again like others were with the Polaris launch.Edited by BiG StroOnZ - 4/26/17 at 9:40pm