Unreal Engine 5 Tech Demo, [UE5 Developer Availability 2022-04-05]

Hey man, UE is a fantastic resource, especially for people like myself who can only dabble as a hobby. What you're all putting out is amazing, and what is in the hands of mere mortals like myself is frankly unbelievable.

I know it's easy for the world to make you cynical, but you're working on something amazing and a lot of us really, really appreciate that.

 

Apologies, I didn't mean it to be specifically directed at you, but I also probably misinterpreted the response. I'll edit my post appropriately.

 

Test of Nanite foliage in 5.1: https://www.artstation.com/artwork/5BbdlO



No idea how it works, it seems like the disjointed cluster LODs would cause a lot of cracks. But one could just disregard that, use standard LODs with clusters and hey cracks disappear. Or maybe some crack filling strategy. Either way visibility buffer with alpha test is doable, as Forbidden West shows.

 

Not shure about reliability, but used MSI Afterburner also, and the results are quite similar.

 

Its a compiled build.

 

I always thought Afterburner pulled the CPU info from what task manager shows. Is this correct?

 

Afterburner should be correct (as opposed to task manager). Afterburner doesnt pull the results from the task manager.

 

Man, having foliage with as much polygons as Nanite allows in games would be a game changer.

Foliage in modern games still looks so fake, like 2D bitmaps (which they probably are).

If Nanite is not doing well with foliage, maybe Mesh Shading could help. Mesh Shading also allows for extremly efficient culling, which could come in very handy when rendering a ton of foliage on screen.

 

Apparently nanite got some form of support for world position offset.

 

Dynamic + nanite = dynanite

 

I have found a way to improve performance in the Matrix demo further.

Disabling any upscaling and running at native 1080p seems to run best on my machine.

Strangely, upscaling be it TSR or DLSS sometimes seems to tank framerate, or not improve it by much.

I wonder what is going! Performance TSR and DLSS at 1440p also run much slower than native 1080p, despite running at a much lower res. Strange things going on indeed.

Nevermind: in the shipped build I used, DLSS Quality is enabled by default. I thought it was defaulting to TSR.

 

Give me sleeping dog sequel please.

 

I hope someone will create the first Unreal's opening level (where you go out of the spaceship) using the new UE5 engine

 

@Dictator I am sure you are hard at work on your Matrix video as we speak. I really wonder to which extent the city sample uses HW-Raytracing.

Maybe you could stack up the 5700 against a 2060 Super again and see how they perform.

I wonder if it really is like you said in the Matrix console demo video, HW-RT offering better speed and better quality at the same time. The 2070 should perform a lot faster if the Matrix demo is using RT acceleration properly. I do not have access to a card without HW-RT so comparing it is difficult, but from what I have been seeing in videos on cards without HW-RT and with HW-RT, the quality is pretty similar and both can use the Epic setting. So non HW-RT cards are running these effects in software which likely does mean you get much better performance on cards with HW acceleration.

Or the second possible outcome is the difference in visual quality is just pretty small and cards with RT acceleration perform actually slower as they use more advanced effects automatically. Which would be very, very bad for RTX users like me.

I am very sure though this comment is entirely irrelevant as you are testing that out right now lol.

 

I was mostly curious to understand why my GPU is never fully saturated in this demo, not necessarily the stutters.

Most likely candidate is of course CPU, I just thought it’s interesting there’s some passes hammering the PCIe bandwidth, like I said though I’ve only spent a small amount of time looking at this demo

 

That is sick as hell.

 

Hey managed to download from the mega links, had to reboot my PC after every file downloaded to play around the IP dl limits.

Is it me or the graphics are not as impressive as hyped? There are plenty of triangles on display a given, but i am still seeing muddy textures, lighting and reflections, spotted many jaggies in the distance, most of the structures are repeated and dull. The metahumans look not much different to today's open world NPC, just a bit more detailed skin textures.

There is also some weird artifacting screen glitches when i run side ways or swing the camera about. Anybody facing this strange visual glitch? It is something like the dejavu effects in the original Matrix! clippy gltichy!

 

Anything that is giving you system-level "core usage" is not going to tell you anything useful in terms of single-threaded CPU bottlenecks. You need something that understands user-mode threads (unrealinsights, vtune, and that sort of thing).

Honestly in a game engine like this if the CPU isn't near 100% loaded it's safe to assume it's likely a CPU bottleneck.

 

Why do you expect it to improve frame rate? Upscaling costs something as a bunch of the pipeline still runs at the upscaled size. The way NVIDIA has marketed it as a performance improvement is by putting "super-sampling" in the title and constantly comparing it to running brute force at the upscaled resolution. Don't get me know, upscaling is great and a better use of resources in most cases than "native 4k" (as much as such a concept even really exists anymore), but the process itself costs a few ms of performance compared to not upscaling. Unless you're saying that upscaling some lower resolution *to* 1080 is slower than rendering at 1080, in which case that's a bit odd but it's possible it's not actually doing that as there are a number of interacting cvars (also note the upscaling you see in editor is driven by the editor controls while in a cooked game it is based on some different cvars).

It also costs vram, as discussed. A bunch of the buffers are higher resolution, and the texture LOD bias will cause more virtual texture usage (although I'm not sure how the pool is sized). If performance is taking a reasonably dive, check vram usage as that might be what is putting it over the top here.

Thus it is not unexpected that upscaling 1080->4k will cost performance vs. just not upscaling the 1080 buffer. It should look notably better though... closer to the 4k image than the 1080 one in most cases.