Of course, depending on how well the Navi core works, it could just be a software hack, not dedicated cores. Given how poorly (relatively speaking) the RTX does with ray tracing features enabled, I'm not sure it's really at much of an advantage.

I do hope that the AMD cards are competitive and they've been a lot better than NVidia at open drivers. Here's hoping they make an impact. Planning to upgrade to a Zen 2 processor when they come out, and if AMD is competitive at least on price, may switch or keep my 1080 and pass my old computer off with a different gpu.

The article seems pretty speculative. AMD might introduce ray tracing features, they might be faster than nVidia's 20x series cards. Hold off on your purchase cowboy! The included graphs of nVidia gobbling up market share over the past year aren't helping either.

Back on the team liquid forums, we used to call this an announcement of a possible future announcement.

Will read the article when it is an actual one.

I think ray tracing will be handled well enough by generic GPU hardware simply using APIs like Vulkan. Nvidia probably oversold their dedicated components, and AMD will try to go the more straightforward way to compete. Which is good, if same thing can be achieved with portable cross-GPU approaches. GPU lock-in is bad and only wastes developers' time.

To somone familiar with ray tracing algorithms: How much efficiency can be gained in using different (invented in the future) algorithms n' tricks (e.g. The kind of stuff that modern game engines use that didn't exist when CG was in it's infancy) or is real time raytracing purely limited by the hardware (By which I mean in general rather than current hardware)?

I'm waiting for GPU's that can output a holographic interference pattern. It looks like random noise, but makes a 3d image if you shine a laser through it.

I think ray tracing actually has advantage here as a precursor to the information needed to calculate the interference pattern.

Please do, Ray Tracing is not something that is really hard to implement and has pretty good impact for gamedevs. Unifying how light is handled can make lots of prebaked light data disappear, making iterations of models / levels more consistent.

Nobody cares about the RTX feature. Gamers always prefer higher framerates over improved indirect lighting which is in practice rather subtle. Envrionment mapping was the king so far for a reason. Psycovisually accurate reflections are the least important thing to look for.

AMD should ignore the RTX move by Nvidia completely and focus on reintroducing their old VLIW architecture for shaders wich saved almost an order of magnitude in power consumption.