AMD expected to talk about their Raytracing Strategy during E3

thats a half-truth perpetuated by angry amd fanboys.

vulkan adopted some of mantles feature set when it was provided but it began as OpenGL Next and was already very low level before mantle was even a concept.

https://www.phoronix.com/scan.php?page=news_item&px=MTc2Mzc

 

Took me 2 weeks to design GPU architecture. Or is that just what you think it took? Navi was done long time before even Raja left. And he was not even working on it.
(Which may be one of reasons why he left. RDNA Navi development started moment Polaris was done which is exactly same time as start of GCN Vega development and Raja was to work on Vega... left behind.)

So, can you compile all those AMD's people who left for intel? And then describe how essential they were for Navi?

I remember talk of Khronos Group tech people praising AMD for enabling them to drastically speedup development.

But hey, take written word of Phoronix instead of spoken word of actual person doing the job.
Statistically, Phoronix is more often wrong in their statements than Adored TV in his speculations. And that's rather tragic. Because Adored makes his claims based on information from random people who may just manipulate him. But Phoronix makes conclusions based on actual data which they misrepresent.

 

When it was presented, I figured that RTX was a software approach based on IA and tensor cores, that multiplicates few rays and denoises them after, and their hardware being a way to capitalize their Volta tensor cores.

I don't know, maybe is the best way to do it, or at least the more elegant, to ditch now-classic shaders and use a mix with many more tensor cores in future cards.

Fully Ray tracing a complete escene is demanding on compute, and doing it brute force and expecting it to be playable it's not possible today.

RTRT is the future, but RTX approach is dependant on IA propietary software that won't be freed by its owner and it could became a relic sooner than later.

Talking about AMD, if Samsung will use RDNA on their mobiles, there is a possibility for it to use IA. New consoles will do it too, but keeping compatibility with actual games will be easier if its hardware compatible too.

Is difficult to predict how it will be before knowing more about, but I will always bet on open source ways, and I hope at least one contender will do it (Intel will join and they are talking about IA too)


Maybe?

PS: remember that today's RTX can only be done in few reflections or few light sources, but not together, so it will be needed a much powerfull technology to do so, and I can't imagine about render a complete game.

 

I can also make you say whatever i want to make up rhetorical question, doesn't make the conversation move forward at all:

Let's see, you're saying the people who left for Intel had no significance at all on the outcome of the project because this kind of technological achievement requires time and engineers.
But, i can arguee those people were hired by Intel, and agreed to leave AMD for reasons, be it money, management decisions, job opportunities.
They could have leave the ship because they disagreed with what was happening for all we know.

Secondly, i'm using plural, saying 'they' as your answered focused on Raja leaving, while it has been clearly reported that it was merely the tip of the iceberg as several engineers to product manager left AMD for Intel, which was actually the core of my concern.