AMD Talks RDNA2, Raytracing, Roadmaps and a possible teaser of Big NAVI

none of them are our friend, they are all just looking for ways to seperate us from our money - it always seems its just the 'cost ignorant' people who come out with it too.

 

You don't own AMD Graphics card, but You do know everything about them? Am i wrong?
Only Nvidia card i ever had was 1070ti (and 6800gt, but who counts that?), every other card was from AMD or ATI, and i never had any issues with them. More issues i had with that one nvidia than any AMD card. Why do You all Nvidia fanboys always talk bad about AMD, when You never own one? I had 5700XT from Asus, and i don't have any issues with some black screen, or any other problem You fanboys talking on forums, and You know why? Because there is no problem with drivers, there are problems with something else maybe, and non smart guys who can't read! Every problem and solution for problems You have with Your PC, You can find on internet. But You need brain and skill to read. That is my oppinion, no offense to anybody.

 

I think the best thing would be if AMD actually pulls a rabbit out their arse and deliver a 4K gaming card at a decent price Nvidia will have to turn down their prices and all consumers both Nvidia lovers and AMD lovers will be better off

 

i think you're market unaware to continue the rantings that nvidia is doing anything other than adjusting to market costs.

 

What is the difference between 7nm enhanced and 7nm+? none. They are both a improved process of the actual 7nm.

 

R VII:
Stream processors: 3840
TMU: 240
Memory bandwidth: 1,024 GB/s

5700 XT:
Stream processors: 2560
TMU: 160
Memory bandwidth: 448.0 GB/s

I'd say you need to take into account that substantial difference when comparing them. If you get as much torque and HP out of a V6 as from a V8 but with less fuel consumption, it's something to pay attention to. In other words if Navi had had the same number of compute units, it might have eaten as much power, but it would have been a whole lot ahead of R VII.

 

One uses EUV for some layers the other doesn't, it's just improved over the initial TSMC 7nm launch.

As for the 50% comparison they are clearly comparing it to 14nm and not 7nm GCN. An additional 50% over RDNA is what they are "aiming for" according to the actual call. I think it's doable - RDNA still had a lot of left over GCN parts that needed to be updated. Couple that with probably getting most games to utilize VRS and you can definitely improve the perf/w significantly on the same node.

 

N7 is initial DUV version (RDNA)
N7P is second gen N7 using DUV (enhanced)
N7+ is an EUV version

 

Ahh, more AMD propaganda.....

 

No let them keep doing it - they basically highlight themselves as biased, making it easier to figure out whose opinion to ignore.

 

We will see, talk is cheap.

Vega's architecture was also supposed to bring 4X the performance per watt than previous generation.

But I hope this time it’s what AMD says.

 

Yeah I don't recall 4x perf/w for Vega at all and quickly looking at slides for Vega's launch I don't see this claim.

Can you provide a source for that?

 

AMD's latest Ryzen Mobile 4000 APU's are already pulling in some great performance per watt numbers. I expect AMD to hit the projected 50% for RDNA2. I have a feeling they have nailed what was causing the most power inefficiencies. How well will this stack up to Nvidia's 3000 series I'm not sure. Even with the 50% performance per watt I still expect Nvidia to be ahead especially in the high end but would be happy to be wrong.

 

@kings Source for those slides? This looks like something Wccftech would put out from rumors. If its from AMD cool but I doubt they are.

 

Says underneath - under NDA until April 1 2017...

 

First slide is fake:

http://www.pureoverclock.com/2017/04/i-got-hit-hard-by-this-april-fools-joke/

Second slide is from the Ve.ga site AMD used to tease Vega launch - it was referring to RPM. When it leaked even WCCFTECH says this:

So if anything they claimed that Vega is twice as efficient as it's previous architecture. I don't know if that's true or not but they definitely didn't say 4x perf/w or 4x, outside of the context of RPM.

Edit: Changed my post to make it more clear.

 

Yes I can read. I just want to know the source of the slides. If my memory serves me right this was a "leaked" slide deck that AMD never put out to the public.

 

Yeah there we go that is what I was trying to remember. Yes they did get about 4x performance on single precision compared to there own prior generation.

 

Yeah, my mistake guys, I apologize. I didn't know about these Aprils fools slides.