for a moment there i was wondering what "windows v2.8.6427" was

The issue is that 3dmark is not dynamic, its the exact same sequence of frames. DLSS needs to prove itself in real games with dynamic environments.

@Hilbert Hagedoorn that time when you most likely are done testing a new vga waiting for the nda to lift and new drivers or 3d mark version rolls out .... !

Went from 37.04 to 52.53 with my setup 7820x@4.7 RTX 2080ti FE at times it did hit over 80fps 4k went from 16.96 to 32.28 if anyone cares lol

HardwareCaps:

The issue is that 3dmark is not dynamic, its the exact same sequence of frames. DLSS needs to prove itself in real games with dynamic environments.

It has to be the exact same sequence, so the test results would be comparable between different cards, different setups and different software environments.

mahanddeem:

It has to be the exact same sequence, so the test results would be comparable between different cards, different setups and different software environments.

He knows that - he's saying that DLSS has an advantage in a demo like this because it's always the same. An AI upscaler is going to be easier to train on a video vs a video game with randomized camera movements/etc. This demo is essentially a video.

While it's not entirely similar, but I do wonder if it's really the time for UL to implement DLSS support at this point. Reminds me when Ageia's PhysX was implemented in older 3DMarks. When Nvidia bought up Ageia's technology, PhysX suddenly was exclusive for Nvidia's graphic solutions. The amount of games that even supported GPU accelerated PhysX was ever few but that support in 3DMark, with PhysX even defaulting to enabled by default, did give the wrong impression of how a Nvidia based system performed compared to a system with a Radeon graphics solution and even, for that matter, a Nvidia system where PhysX was disabled. If DLSS does get supported in more games than Final Fantasy XV (and whatever more that's supporting it), it should be enabled. But at this point, it really only reminds me of that 3DMark PhysX thing.

I'm very concerned for both DLSS and RTX. adoption rate is not looking great. Hopefully Anthem will show us how DLSS works in dynamic multiplayer environment

Xtreme1979:

It's not part of the default Port Royal benchmark, thus it isn't scored as such. You have to launch a separate test called DLSS feature test. Then it gives you a before and after comparison of the technology.

I missed that, if it works like that I don't see much problem with it at all. 🙂