Battlefield V: GeForce RTX 2080 Ti Rotterdam Gameplay

Published by

Click here to post a comment for Battlefield V: GeForce RTX 2080 Ti Rotterdam Gameplay on our message forum
https://forums.guru3d.com/data/avatars/m/252/252732.jpg
People on these forums amaze me sometimes, constantly complaining that all we get is console ports and that developers don’t push our pc’s hardware, then new tech comes along that does push the hardware and still people complain that the resolution isn’t high enough, or the frame rate isn’t high enough. You’ve got to start somewhere with new ways of rendering, I see is as the start of a revolution in how games are created, not a backwards step.
https://forums.guru3d.com/data/avatars/m/79/79740.jpg
Denial:

... What exactly is the issue here other then once again people fighting against progress?
No objection to progress, just that since RT is a feature Nvidia is highly touting, one would think they would encourage wider adoption of it with far more friendly pricing. Especially since its a feature that many with higher res monitors are likely to turn off. But whats $1199 for a good cause, eh (Nv's pockets).
https://forums.guru3d.com/data/avatars/m/115/115710.jpg
They could easily make the ray tracing run at half the resolution and it should improve the performance a lot. I'm surprised how well it runs even at this stage. It's only going to get better and better once they know how to optimise it better. With proper optimisations it will likely run quite well even at 1440p. In the future this is going to help a lot since doing reflections with rasterisation is just pure hack. Dynamic light and shadows too. It needs awfully lot of work from artists and tech artists to get it look right. Pre-baking the lights may look nice at first glance (like Mirror's Edge for example) but that approach just would not work once you add dynamic lights, day and night cycles or dynamic environments. This is a rather impressive first step but the next get ray tracing capable card built with 7nm should make this practical in all situations.
data/avatar/default/avatar07.webp
alanm:

No objection to progress, just that since RT is a feature Nvidia is highly touting, one would think they would encourage wider adoption of it with far more friendly pricing. Especially since its a feature that many with higher res monitors are likely to turn off. But whats $1199 for a good cause, eh (Nv's pockets).
TU102 is a 750mm2+ die. And yet how come we dont hear those voices who used to complain about cheap&easy to produce 300ish mm2, or so called mid-range Gx104. I don't hear them using the same mm2 metric and defending this horrible price of 1300 euro because super-high-end, because sh1tload of features and because it's expensive/hard to produce.
Anarion:

They could easily make the ray tracing run at half the resolution and it should improve the performance a lot. I'm surprised how well it runs even at this stage. It's only going to get better and better once they know how to optimise it better. With proper optimisations it will likely run quite well even at 1440p.
Yes. It's mind boggling that both shadows and reflections are done at full resolution. Like wtf... cut them in 1/3 and they'll still be great, and more importantly - accurate.
data/avatar/default/avatar11.webp
What are we complaining about? Ray tracing is a new tech. We still don't know how it will do at 4k(unless I missed something?) without raytracing. There is an option to disable it right? I'd say keep raytracing disabled for 4k for at least a generation or two.
https://forums.guru3d.com/data/avatars/m/79/79740.jpg
Noisiv:

TU102 is a 750mm2+ die. And yet how come we dont hear those voices who used to complain about cheap&easy to produce 300ish mm2, or so called mid-range Gx104. I don't hear them using the same mm2 metric and defending this horrible price of 1300 euro because super-high-end, because sh1tload of features and because it's expensive/hard to produce.
Yes, its more costly to produce... yet in the end they are laughing their way to the bank with the 63% margins they made from their last quarter. p.s. I presume they want us to buy these to go with our shiny new RTX cards... 🙄 [spoiler]https://cnet3.cbsistatic.com/img/iSk50dhT06wV4QKUVoTwLrXn8a0=/936x527/2018/01/09/c2e68b96-325d-4033-9667-71fa1338264c/004-nvidia-bfgd-big-format-gaming-display-ces-2018.jpg [/spoiler]
https://forums.guru3d.com/data/avatars/m/258/258664.jpg
alanm:

p.s. I presume they want us to buy these to go with our shiny new RTX cards... 🙄 [spoiler]https://cnet3.cbsistatic.com/img/iSk50dhT06wV4QKUVoTwLrXn8a0=/936x527/2018/01/09/c2e68b96-325d-4033-9667-71fa1338264c/004-nvidia-bfgd-big-format-gaming-display-ces-2018.jpg [/spoiler]
hehe yeah I know what you mean 😀 Only that I wouldn't even have the space for that thing tbh, that's why for me they're a fail, and I'm more interested in Samsung's normal TVs that get Freesync. At least I can get them in a size I can put into my living room without having to move my furniture...
https://forums.guru3d.com/data/avatars/m/80/80129.jpg
Fox2232:

On other hand, they thought that nV is cheating with their numbers too. Can you actually run that raytracer w/o AI cleaning involved? because nVidia's statement about Gigarays/s is based on subjective result image. I wonder what would be result if you actually made benchmark which just counts ray checks performed.
According to Tom Peterson that number is an average from real world testing across multiple applications and not a single benchmark. I'm not sure why the AI cleaning part would matter, rays cast is rays cast. May not be enough for an image without AI denoising but im not sure what that has to do with the number of rays being cast when comparing two architectures.
alanm:

No objection to progress, just that since RT is a feature Nvidia is highly touting, one would think they would encourage wider adoption of it with far more friendly pricing. Especially since its a feature that many with higher res monitors are likely to turn off. But whats $1199 for a good cause, eh (Nv's pockets).
I mean if the argument was simply "pricing on the new cards is too high" i'd agree. But no one is really pushing that as a primary argument in this thread.. everyone is just complaining about the performance, general performs of RTX cards (why is that even relevant in this thread) and glossiness of the image.. then tacking on the RTX card price as if it somehow makes DXR/Raytracing less revolutionary. Their implementation of DXR is platform agnostic. AMD will be able to accelerate the raytraced effects with their own variant of RTX or however they plan on implementing it in their hardware. I don't see how it's controversial. It's a new tech, it's super exciting because as performance scales image quality/bounce/resolution will scale. It will scale to multiple different effects and not just reflections. It basically eliminates the need for all the shitty lighting hacks in place now that devs have to use in order to get viable performance with somewhat believable quality. Even if it's only partial and wonkly implemented.. it's fucking raytracing in real time. Super cool. The pricing thing will fix itself in time and not relevant to the tech, at all, in my opinion.
https://forums.guru3d.com/data/avatars/m/216/216349.jpg
So still too son for ray tracing in games i guess? I´ll wait 3 to 5 years so i can properly enjoy games with RT. It´s cool what Nvidia is trying to do but it seems the hardware is till lacking the raw power for RT.
https://forums.guru3d.com/data/avatars/m/245/245459.jpg
When I was watching that video I was imagining I was there playing the game, and I didn't think about the graphics once - if ray tracing means that with a 2080ti you're gonna be at 60 fps and 1080p then ray tracing is the first thing you're gonna turn off, you want this game running at 1080p 144fps instead, perhaps that was why the guy was playing the game in such a slow & sniper'esque style, not enough fps for faster game play styles. Also, if you're a fan of running at 1440p or 4K then you'd hope a GTX 2080ti would be able to run that resolution at maxed out game details - lol, not 1080p for that card for 60 fps!
data/avatar/default/avatar17.webp
Fox2232:

On other hand, they thought that nV is cheating with their numbers too. Can you actually run that raytracer w/o AI cleaning involved? Take it this way: 1 Gigarays/s 1920x1080 = 2,0736 M pixels on screen 60fps on same screen = 124,416M pixels on 1080p screen per second for 60fps gaming if 1 ray per pixel is done. Raytracing unit capable to do 1 Gigarays/s can perform 8 rays per pixel on 1080p screen and still deliver 60fps. With 8 rays per pixel, you do not even need denoiser. Now take into account what kind of raytracing performance nVidia advertises And fact that in those games, raytracing is applied only on some pixels. Like reflective surfaces, places where shadows are expected, ...
It's not out yet and they're already cheating? Wow.... altaresification1 week ago Why 3.2 billion and not 10 billion? [URL='https://www.youtube.com/channel/UCMKGLHgAfoRikL3lKsz-rHQ']https://yt3.ggpht.com/-eSGJwOqjNrE/AAAAAAAAAAI/AAAAAAAAAAA/YQBP1QIqtFc/s76-c-k-no-mo-rj-c0xffffff/photo.jpg OTOY[/URL]1 week ago Because we are doing full cinematic path tracing on a complex interior scene. With Cornell Box scene it is closer to 10 billion youtube.com/watch?v=_EbPDMaXxiY
https://forums.guru3d.com/data/avatars/m/80/80129.jpg
Robbo9999:

When I was watching that video I was imagining I was there playing the game, and I didn't think about the graphics once - if ray tracing means that with a 2080ti you're gonna be at 60 fps and 1080p then ray tracing is the first thing you're gonna turn off, you want this game running at 1080p 144fps instead, perhaps that was why the guy was playing the game in such a slow & sniper'esque style, not enough fps for faster game play styles. Also, if you're a fan of running at 1440p or 4K then you'd hope a GTX 2080ti would be able to run that resolution at maxed out game details - lol, not 1080p for that card for 60 fps!
People played twitch FPS games, like quake, at pro levels on 60hz monitors for years.. 60fps is not effecting anyones playstyle. Also the entire previous page in this thread is about how the performance here isn't even close to final - that they already know of a way of increasing it by 30% and the other changes they have in store will bring it up even higher - especially at higher resolutions.
https://forums.guru3d.com/data/avatars/m/263/263710.jpg
[youtube=WNS8a60YVyc]
https://forums.guru3d.com/data/avatars/m/245/245459.jpg
Denial:

People played twitch FPS games, like quake, at pro levels on 60hz monitors for years.. 60fps is not effecting anyones playstyle. Also the entire previous page in this thread is about how the performance here isn't even close to final - that they already know of a way of increasing it by 30% and the other changes they have in store will bring it up even higher - especially at higher resolutions.
Not to derail the thread, but for me 60 fps would affect my playstyle, because only 60 fps makes faster gameplay less easy to compete at close quarter combat, so I'd probably find I'd lose more encounters, which would lead me to then switch to a slower more sniper'esque style. If lots of the other players are at over 60fps its a distinct disadvantage - when everyone was on 60 fps in quake then it was an even playing field. When I switched to 144fps up from 75 fps I couldn't believe how easy close quarter combat had become. I can't imagine them optimising ray tracing in BF V to get to 144 fps at 1080p, they're only talking about 60 fps!
https://forums.guru3d.com/data/avatars/m/115/115710.jpg
Fox2232:

On other hand, they thought that nV is cheating with their numbers too. Can you actually run that raytracer w/o AI cleaning involved?
It's not cheating. The thing is, you can use ray tracing to get rough enough image that's good enough so that it will look good after cleaning. It's going to take a long time before you can ray trace in real time without any denoising. However, meanwhile denoising works really well enough for stuff they are now using it and it makes the real time aspect possible. Everything will be physically correct. Also it likely a bit more complicated than simple napkin math like that in practice.
data/avatar/default/avatar26.webp
cryohellinc:

This. I was expecting next release to be capable of 4k 100fps+, instead, we got RTX.
The 2080 Ti is still going to be 40-50% faster then a 1080 Ti in pure shader performance without RTX. So if you don't use the new features, you might reach that. (Extrapolated from the 2080 numbers that leaked, which match/surpass a 1080Ti, and the 2080Ti has almost 50% more shaders then a 2080)
data/avatar/default/avatar13.webp
Freaking 1080p... a small step forward for raytracing, a huge leap backwards for graphics overall...
https://forums.guru3d.com/data/avatars/m/80/80129.jpg
Fox2232:

If you had 10 Billion raytraced intersections done per second, you do not need any kind of cleanup and you do it all, not just one or 2 effects. 10 Billion raytraced intersections gives you 80 rays per every single pixel on 1080p screen at 60fps. That's photorealistic amount which does not need denoise. Take Bryce or anything out there that does raytracing and test how different number of rays/s created different quality. Even 4K which would do 60fps while calculating 20 rays/pixel would need almost no cleanup. And as I wrote, they are using it in BF5 just on reflective surfaces which are apparently covering only small fraction of screen.
I don't think it can do 10B/s while also rasterizing an entire image + you have to factor bounce - I believe in the DF video they say they bounce twice before being calculated into a cubemap.
https://forums.guru3d.com/data/avatars/m/260/260048.jpg
nevcairiel:

The 2080 Ti is still going to be 40-50% faster then a 1080 Ti in pure shader performance without RTX. So if you don't use the new features, you might reach that. (Extrapolated from the 2080 numbers that leaked, which match/surpass a 1080Ti, and the 2080Ti has almost 50% more shaders then a 2080)
That we will only prove after the 20th of September.
data/avatar/default/avatar04.webp
Fox2232:

10 Billion raytraced intersections gives you 80 rays per every single pixel on 1080p screen at 60fps. That's photorealistic amount which does not need denoise. Take Bryce or anything out there that does raytracing and test how different number of rays/s created different quality. Even 4K which would do 60fps while calculating 20 rays/pixel would need almost no cleanup. And as I wrote, they are using it in BF5 just on reflective surfaces which are apparently covering only small fraction of screen.
Except that one ray isn't only one calculation, because rays bounce around, split off into shadows, reflections and refractions, and do all sorts of shenanigans. You could count the number of "pixels" you can render that way if you wanted to, but its not the same as the number of rays.