walsufnir said:
Again, a tf number is telling you next to nothing in terms of real world applications as no application will ever compute that much to reach these numbers. I went through this in the beginning of this gen and still do. Computers are complex systems with many components relying on each other and they have to work together to get something on a screen. Add to that that different games have different demand on the computing systems so there isn't even a remote rule of thumb. Additionally expect a new generation of the gpu meaning a different uarch underneath, being more efficient at computing. |
I'm aware of the variability with tflops, hence i ended off with "The Scorpio's tflops are going to have to be worth considerably more than the X1's to pull it off.". I'm also aware it can vary by game, hence the relative qualifiers i used were vague (in general, given SweetTalia seemed surprised by just the OT, i felt it best to keep my explanation simple).
It's not outside the realm of possibility that the Scorpio will could hit the mark (nor do i say otherwise), but based on what we know about the Scorpio, and what we can reasonably assume, i personally think it's more likely it won't more often than it does in these situations.
Remember, we aren't working in a vacuum here. While predicting specifics is hard, we can at least make good guesses at the upper end of what reality can allow the Scorpio to achieve. We know whose making its GPU, and how their current GPUs perform. We knows there's a limit to how large an efficiency jump AMD could make between now and the end of the Scorpio's design windows, brand new architecture or not. We can also make reasonable assumptions as to the limit of how much MS want the overall machine to cost, and we can confidently say they want the system to easily scale with the X1. You get the idea.
So again, i could be entirely wrong, but my personal expectation is the Scorpio will at least struggle to hit 4k in instances like this.
