This is how nVidia playing their cards. And IMHO is the beginning of the death of Radeon.

1) Instead of pushing the limit with usual methods like adding # of cores, die shrink and increasing the memory bandwidth, they play with the graphics quality this round, ray tracing.
2) But this ray tracing thing is not do-able with only brute force method. It has to come with the library and API. And nVidia had already begun the groundwork with game engines. This move, I believe is akin to the nVidia's move in GPGPU. It's like putting the first flag to claim a new land. Look at where AMD is today in GPGPU application, especially in deep learning. The hardware is there but the community support in terms of library is almost non existence. CUDA is so far ahead that no community is willing to start all over again with the OpenCL.
3) nVidia use the Gigarays as the new quantifiable benchmarking standard. Like it or not, soon a lot of demos will show the difference without RTX and the Radeon cards will fall in the "ugly" version of the games. If AMD were to follow suit and play the same game, it is definitely behind by at least one gen. If it does not play the same game, they can only sell to budget gamer.
4) The Radeon is doing great in the past two years due to cryptocurrency mining and also game consoles. All game consoles run on Radeon, except the Switch. While it remains to be seen Sony and MS will switch to RTX in next gen (remember nVidia's per-chip loyalty scheme in Xbox and PS3?), it definitely put pressure on them to switch once the PC master race starts to compare the graphics qualities.

So there you have it. Low margin, low volume (if lost the next gen console contract). How to fight?

There is no proper benchmark because this is the first time it's done real time la. If there is a benchmark, almost all previous gen cards will get less than 5fps. The theme of this gen is graphics quality. Different ball game d.

This post has been edited by davidletterboyz: Aug 21 2018, 11:26 AM

The forex does not seem to add up?! looks like the 2080 in Msia is almost equal to the 2080Ti price in US?!

I thought the Star Wars demo was around ~22fps?

Jen Hsun said it's the first time it's done. So, need a new way to measure and benchmark. Even the metric to measure operation rate is new. RTX-OPS?

Yes. I think it's very hard for AMD from now on.

One thing that caught my attention yesterday is actually the Tensor cores inside RTX. Jen Hsun said it's 10x faster than 1080Ti?