• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Unreal Engine 5.6 Delivers Up to 35% Performance Improvement Over v5.4

on an RTX 5080

Cared to benchmark on a Radeon RX 9070 XT just to get higher image quality and in the name of fair competition and not advertising the green goblin ?
 
I mean developers could re-compile the game with the newer UE5 sdk if they want and release it as a patch
It’s not as easy as just a recompile. Unless your game useless nothing but UE blueprints the version changes are code breaking.

Cared to benchmark on a Radeon RX 9070 XT just to get higher image quality and in the name of fair competition and not advertising the green goblin ?
No.
 
I wonder how large the performance gain is for the base PS5 because 35% is definitely not enough there.
 
Normally, these would be some seriously impressive numbers.
But if you've followed the benchmarks of UE5-based titles and noticed how punishing they are on the hardware (while showing very, very little differences between low and max settings), you realize UE5 was just rushed out the door in a very rough shape.
IMHO, the differences might be small, but they still have an impact. We are almost at the end of the 9th gen of consoles, and I can barely tell a difference between games that launched at the beginning and games that launched now. Even a few 8th-gen titles are still looking good by today's standards.

It's not just UE5 games, a ton of modern games have impressive graphics at minimal settings, the biggest visual differences being how many assets are loaded, and texture resolution.
 
Does a 50% performance delta(60-90fps) in an empty city with no game logic running impress anyone? To me, it’s at best a questionable benchmark.
The performance gain likely came more from optimizations in RT. True optimization depends on the devs and those working with UE5 usually lack the time or expertise to achieve it.
 
Hardly. This would mean recompiling the whole game on the side of the game dev created this game and it's a long, annoying process. Virtually nobody would care.

So we should only expect this to apply to the games that aren't yet released.
I can imagine it being a costly but worthwile endavour for certain live-service games. If the gains are really that big on their use-case, then again they probably don't depend on what is being optimized here
 
What does this mean for people that don't play benchmarks? Will it only apply to games compiled on 5.6+ or can this be patched into titles already released?
I would love if most studios decided do the effort, but I doubt it will happen, it's a lot of work and it can add new bugs.
On the other hand I think Epic Games want UE6 to be the one where all games can be updated to new versions without having much work to do! But I doubt it will release before 2028 if not 2030 lol
 
You've probably meant "not in development". As it's same, and hardly anyone would be eager to recompile any existing projects, and perhaps breaking the existing code/core integrity, while simultaneously raising the development costs.
It may be only viable for those, who would rather prefer the long term sustainability of their games/projects in long term. Though, knowing how short-lived the current gen gmes are, these news are only referable to the future games.
Mid-development engine upgrades aren't too uncommon. I've seen several recent games release on UE 5.4 (which came out a year ago) despite starting development years earlier. Fantasy Life i, Rune Factory: Guardians of Azuma, Clair Obscur, and MindsEye, for a few examples that came out in just the last month or so. And Splitgate 2 was compiled on UE 5.5, which is even more recent.

An engine upgrade can be a pain in the ass for sure, but it can also be worth it if the benefits are strong enough, and a 35% performance uplift for hardware lumen may be appealing enough to get devs to go through with the process.
 
Does a 50% performance delta(60-90fps) in an empty city with no game logic running impress anyone?
It's an apples to apples test, same rendering load, only change is the engine version. It might not be 100% representative of the uplift in every scenario, but it's in no wya an invalid test.
 
With such a boost it kind of feels like the engine has been in beta-stage for quite some time
 
I feel there’s always something very wrong with Unreal Engine. It pushes graphic fidelity, but at the expense of problematic and lengthy shader compilation resulting in stuttering issues in general. It looks better, but frankly, I am happy to trade the incremental graphical fidelity for better performance. UE5 as you can tell by now is a performance killer, where it is not uncommon to experience low framerates and stuttery gameplay. There are standouts like Expedition 33, but it’s far and few between.
 
LOL, 88fps @ 720p resolution. :))))
 
1750564742150.png

This here is the biggest deal IMO -- these unreal engine 5 stutters are so brutal.
 
oh 5080's good, that's good. Question: What is the improvement on 40 and 50 series 60, 70 and Ti models where it actually matters?
What about 30's?
 
I would like to see this test done on 7900 to make sure no FG trickery was involved.
 
I wish game engines could be easily upgradable for uses like this. This is great for same scenario, but if you apply this new engine to a game 3 years in the future, they'll just throw more visual crap in the game, entirely negating the gains, so no use for it. But if you could improve performance like this in existing game where it's unlikely they'll just add bunch of crap in it, it would be great.
 
Just watched the video. It's so fuckstupid to see what seems to be Paris filled to the brim with American cars and buses.

The sheer level of incompetence.
You’re free to make your own. Please let us know when it’s available.
 
Does it though?:

4:39 is funny devs claiming things that were done routinely pre DX12 era, are not possible to do today, so they admitting things have gone backwards?
 
Yup, I think a lot of people don’t understand this. Every poorly running UE5 title will remai as is. We’ll only see the effect of 5.6 mayyybe in two years or so.
So in essence it means we will never see the difference ;)
 
"Lighting feels more accurate."

Something doesn't feel right with this sentence...

4:39 is funny devs claiming things that were done routinely pre DX12 era, are not possible to do today, so they admitting things have gone backwards?
Two random screenshots of random comments that include neither usernames nor profile pics and aren't even put in the context of whatever is being ranted about are hardly "devs claiming" anything...
 
Last edited:
Back
Top