Super Resolution can't come soon enough. Hopefully with it the RX 6700X would actually be better placed in terms of price outside of RT.

#Kickstarted

Aaand download started, gonna tell you tomorrow more about it.

@JEMC Thanks!

Ouch 1080P at 60FPS max settings for RE8 with my GTX 1070. Hopefully will be able to pull 1440P with medium settings. Really need that 3070Ti!

if i wouldn't totally suck at bullethell i would have taken a try

Weekly WAN Show Recap brought to you by Linus Tech Tips 

*** NEW CONTEST ***

Win one of three Noctua NH-U12S redux bundles
https://hexus.net/tech/features/cooling/147589-win-one-three-noctua-nh-u12s-redux-bundles/

The three winners will get an NH-U12S redux CPU cooler, an NA-FK1 second fan kit, two NF-S12B redux-1200 PWM case fans, and a tube of NT-H1 3.5G thermal paste.

This contest is GLOBAL.

All this talk about AMD and an DLSS competitor is weird. Nvidia has already pulled far ahead and they're gonna continually improve DLSS until AMD has "something". AMD won't have an alternative to DLSS until they do exactly do DLSS. So until AMD announces to include hardware for ML we can safely ignore any comparisons between them and Nvidia. I think it's actually pretty insulting to Nvidia to call anything AMD has proclaimed so far as "competitor" or "alternative" to DLSS. So we should really stop those comparisons until AMD actually has something remotely resembling DLSS.

Seriously guys, whatever edge smoothing technology AMD has in development cannot even come close to real AI image construction.

I wouldn't shill for Jensen just yet. It's been shown that AMD can catchup with Nvidia if they want to, the problem is doing this without breaking the bank. Remember AMD is the only company that has to compete in two fronts with under half the marketshare of their competitors: Intel is filthy rich and Jensen's circus has been the main attraction since its foundation. Fighting a two front war isn't exactly easy. 

But the fact they went from being the slowpoke to running over Intel shows it's perfectly possible for AMD to come up with their own AI upscaler. 

And we've seen time and time again that the best tech not always wins. Ask Sony with the Betamax or Nvidia with G-Sync.

In the end, the solution that's cheaper and easier to use and implement on as many devices as possible will win. In this regard, coming up with that UE plugin could be the winning factor for DLSS. In any case, first we have to see what AMD comes up with, and then we'll see what happens next.

Open source technology usually always wins out due to lack of licensing fee that comes with it. Nvidia tech usually sticks to its own platforms with propriety tech. However with the new Switch and Nvidia R&Ding into CPU/GPU's.. DLSS could blow up a lot more in the games industry. But yeah we all remember what happened to PhysX, 3D Vision and less-so GSync so keep expectations in check. DLSS is bloody good tech though!

And its not to say Super Resolution will be actually good as in directly as good as DLSS, but maybe we should wait and see The more people can can have access to higher IQ from GPU, the better. The higher the adoption rate the more developers will be inclined to use it. I just hope it won't be the case of devs choosing one thing over the other.

I'll add a few more things to the conversation. See the thing with DLSS is that you don't need Tensor Cores to do it per say. Tensor Cores help greatly but both DLSS and DirectML can be done without Tensor Cores and we have real proof of concept with DLSS and Control. Here's a brief history on how it went.

Target Resolution: 4K

DLSS 1.9:

Done Through:
Shader Cores

Pros:
Looks quite similar to Native 4k Image
Performs Very similary to the Rendering Resolution such as 1440p
Can be done in any GPUs that has INT-4 and INT-8 support including RDNA 2 if Nvidia wanted to

Cons:
Needs Per Game AI Training
Going below 1440p Render Resolution starts to look bad
Noticeable Artifices and Ghosting when in Motion and wasn't very good at dealing with Transparency

DLSS 2.0:

Done Through:
Tensor Cores

Pros:
Looks very close to Native 4k image, sometimes even better than Native
Motion and transparency issues are mostly gone
1080p Render Resolution looks better than DLSS 1.9's 1440p Render Resolution
Doesn't Need per Game AI Training as the AI is now Generalized. Just needs in-engine support

Cons:
Requires Tensor Cores

Results:
Fap worthy performance and image quality

Now according to Nvidia, they needed Tensor Cores because the Shader Cores are too slow. Technically speaking, they are correct. The AI acceleration in a 2060 is twice as fast as a Series X thanks to the Tensor Cores. And because of that, Tensor Cores I believe are certainly needed to achieve what we have. But that doesn't mean that something isn't do-able with just Compute Units as DLSS 1.9 shows.

The problem with AMD is that their vision on how to approach this is so behind that it's going to take time. They haven't even figured out on which method to even go let alone starting to develop for it. Most likely, we won't see anything until 2022. And thanks to DLSS 2.0's massive improvements and it's ease of implementation, there will be a ton of games that will get it while everyone else waits for AMD's solution. The kicker of course is, once AMD does come out with their Super Resolution, it will work on Nvidia's GPUs as well and won't be anywhere close to DLSS in terms of performance and image quality for a long time.

https://www.techspot.com/article/1992-nvidia-dlss-2020/