Pretty much the conference.....
Money post.
In their chart they compare a GTX1080 against a RTX2080, the 1080 has 320Gb/s whilst the 2080 has 448Gb/s similar to a 1080ti. So what do they do, they compare at the rez which requires more bandwidth (4k), so they could make GTX1080 look shitty. Was 1080 ever a 4k card? no it wasn't....Yet, I have no bones about 4k tests, because this is where we're heading into with traditional rendering.......but they didn't pitch traditional rendering at 4k 60fps did they?. They pitched raytracing for 2hrs, not at 4k rez, but at 1080 and sub 60fps....
So, I've mainly been asking, why not show the settings for the games tested on the graph. Is it ultra, is it high, does it have AA? All I saw was 4k rez wiith HDR, but why HDR? because everybody knows pascal blows with HDR ON against the competition or even against NV's newest cards. So at 4k rez, GTX 1080's limited bandwidth will take the tumble, with HDR on, even more so, because everybody is playing modern games at 4k with HDR on their 1080's
Two areas where the 1080 can suffer heavy perf loss...
I've heard a youtuber say recently that NV is going to be the best card, it's a forgone conclusion, so they could innovate and nothing could shake them. People said the same thing about CPU's, yet AMD was much further behind intel in the CPU space, even moreso, than they are from NV in the GPU space.
Vega was not fully realized in-terms of enginnering due to a manpowershift, but that won't be the case with NAVI as they are getting good R&D dollars for NAVI collaborating with Sony on that project. They now have Vega at 7nm and giving Vega the love it needed on first light. So they can very well launch a 20.9TF Vega this year, open up their own RT solution and offer folk a 4k 60fps solution with HDR, CB, FP16 and even expand on Vega features.
At this point, AMD is not necessarily in a bad spot with their GPU's, mining helped them make some bank with Vega, their Rx580's outperform the 1060 on many games, has more ram and is now priced even cheaper than a 6GB 1060. Their 4Gb 580's destroy 3GB 1060's and the 570 runs all over the 1050ti. The Vega 56 is a very good card at it's current price in the Vega lineup, and if you can get a sapphire based 56 or even 64, they are the best AIB vega cards for performance, bar none, all the others are jokes in comparison.
Yet I think, AMD now has a chance to topple NV in perf and rez and even have an RT solution whilst at it. NV's best card was 11.3TF and now their best next gen card is at 13.4TF, that's not a huge boost. Clearly their focus was on RT cores and tensor cores, but everything looks a bit underwhelming with low perf and low rez there (raytracing push).......A re-engineered Vega will eliminate all issues they had on first release; unfinished engineering, high wattage, low clocks......Clearly, they will sort that out with Vega 7nm, and even provide a better cooling solution out the gate...and there's no way NV will beat them on the rez and fps front with turing if they do.....Keeping in mind that the archaic DX11 is going the way of the dodo and everybody is on the vulkan train now, which is where AMD arch excels.....Intel and AMD are fully commited to vulkan, even NV has no choice and are on the vulkan train too....
I give everything a chance, so we will see how it pans out, but I think AMD is in a good position to make a move and gain lots of marketshare here.........They don't have to follow NV, they can follow their own vision and right now it's looking gelled with the high rez, high fps transition. Not forgetting games development will be mostly led on AMD hardware due to consoles in the coming years. It's also indicative of AMD's plan to push their hardware to developers even on the PC side to ease the transition (perhaps they knew about NV's focus on turing).....They have recently shipped many Vega and Ryzen kits to PC developers just the same, to get these devs more familiarized and tuned to AMD hardware and arch with the new api's taking shape and gaining momentum. PS5 development will only make this easier and extend this. I think they realized they had a chance here for more than a minute, and are putting everything in place......They are already at 7nm, whilst the other teams struggle to get their nodes on or try to do things which are not necessarily ready for primetime. It will be an interesting couple of months, that's for sure....