• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

[Digital Foundry] Gears Tactics PC - The DF Tech Review: CRS, Performance, Engine Analysys, new DX12 features?

Andodalf

Banned
was wondering this myself when that XsX games 4chan leak came out and said a lot of games using UE4. It's a great engine, but I can't wait to see what the next level is

I feel like UE4 still has some room to grow, it just got it's first ray tracing elements added in. UE3 had massive growth from a game like ME1 to Arkham Knight, and I hope UE4 gets as well used at the end of it's life as UE3 did.
 

CrustyBritches

Gold Member
You mean like getting it to upscale if you’re above the minimum FPS?
In the benchmark under 'Window Resolution Scale' it states '2560x1440(100%)'. Under 'Frames Rendered' it will state "xx.x% Full Res". I'm confused what that means, so I was wondering if this game has resolution scaling.

On a side note, I'm not seeing much difference in visual quality, if any, between VRS 'Off' and VRS 'Quality'. Basically a free 6-8% performance bump. For what it's worth, I think DLSS 2.0 pretty much nullifies most use cases for VRS. I'm researching UE4 DLSS 2.0 support and it looks like Nvidia created a DLSS 2.0 UE4 branch that currently in beta. Mechwarrior 5 is getting it and it's a UE4 game.

P.S.- Feels good to be on RTX and getting all these features like RT, VRS, and DLSS.
 

ZywyPL

Banned
I've posted this a million times, so I'll do it again, having an SSD feeding video ram directly allows you to render things that are otherwise impossible:


I even commented this video, I don't know if it was you who posted it back then tho - as impressive as it is, those demos run at sub-30FPS, which is simply not enough for games, on a professional market where a single frame can be rendered in minutes or even hours, even 1FPS is considered as real-time, but for games that's simply not enough. And that's a card more powerful than even XBX, with 16GB HBM2, on-board SSD and HBCC, so I highly doubt next-gen consoles will be able to deliver comparable (subpar) results, let alone smooth gameplay.
 
I even commented this video, I don't know if it was you who posted it back then tho - as impressive as it is, those demos run at sub-30FPS, which is simply not enough for games, on a professional market where a single frame can be rendered in minutes or even hours, even 1FPS is considered as real-time, but for games that's simply not enough. And that's a card more powerful than even XBX, with 16GB HBM2, on-board SSD and HBCC, so I highly doubt next-gen consoles will be able to deliver comparable (subpar) results, let alone smooth gameplay.
Hmmm I get it, it's just for fast loading times? Man I have been fooled.

Have you thought that if the data-set is smaller (50 to 100gb let's say)... How would it pan out? Why even bother with developing a whole console and how to make the best use of the available bandwidth to fill memory so fast if there is no benefit to it? They might as well have put in some SATA III SSD and then concentrate on the GPU/CPU/Memory speed. You have to thing that even MS has built APIs around the newly available bandwidth, and they include that SSD of theirs in their PR efforts as well, I suspect there is more to it than you believe.

The example given by Cerny, and MS I believe, is one of having to load the assets of what is supposed to be in view in the next second instead of planning what needs to be loaded minutes in advance (obviously in a scenario of dynamic loading), thus in effect giving you more working memory when compared to a slower SSD, and even more compared to a machine that has a regular HDD. Obviously once that data hits the rendering pipeline the GPU won't have more resources than it did before, but it certainly can get better data to work with.
 
Top Bottom