Yes and it wasnt great.
-
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: Neo GPU are point-for-point a match for RX 480
- Thread starter Shtof
- Start date
AMD has a set of IP blocks that can go into a chip and sony tells them what features and performance characteristics they want
they're not getting to customize the compute units or anything like that
ethomaz
Banned
They did with PS4.
And the customization was even used by AMD in GCN.
MS did customized Xbone GPU too... ESRAM and DMAs were added by MS.
Yes and it is bad.
geordiemp
Member
Yes, we know Sony uses TSMC for Ps4 at 28 nm, but the RX480 is made by GloFo on their 14 nm process.
There is no evidence yet that Sony will use the 16 nm TSMC, they may go Samsung 14 nm - do we even know if the 14 to 16 mm for the GPU part is an easy switch for AMD ?
I agree that its possible the Nvidia 16 nm GPU TDP are helped by TSMC process, and maybe if Sony go TSMC 16 nm they might get less heat......but who knows. Its still a big unknown.
Maybe the 480 design just gives off allot of heat no matter if its 14/16 nm or Glofo / TSMC ? (In which case we are getting 4.2 for Neo and low clocks lol)
Well, it's what they're using for ps2 emulation, so I think it might be different. It seems to work quite well.
We will see a lot of image reconstruction tech with neo and Scorpio.
ethomaz
Banned
There is nothing like that in PS2 emulation on PS4...
- 2D artworks gets upscaled (there is no other way unless you recreate the artwork)
- 3D elements are rendered in higher resolution (1080 lines)
Temporal reconstruction is what QB does... it get the last three/four 720p frames and try to use them to guess what needs to complete an actual 1080p image from 720p render... that is why moving generates weird artifacts.
The NEO push is generally just to get the box compatible with 4K media. They aren't mandating 4K for games, so it really doesn't matter, devs will likely stay with the same res as PS4 with higher quality AA and rendering samples.
I'm talking about that patent. Wasn't it pretty much accepted at this point that it's how they're handling higher resolution for ps2 games?
I mean this guy seems to think so, and I'm pretty sure he knows what he's talking about when it comes to the ps2 and such.
Neo is a PS4 & it will be playing PS4 games but now it has over 2X the processing power to make that PS4 game prettier however they do it.
My guess is that it will just be PS4 games but they will use the extra processing power to up-render the games. & it's not going to be something that's going to take up too much of the devs time.
The last thread where you brought this up turned into a mess please just end it at that.
ethomaz
Banned
He said it is useful... not that is being used.
PS2 emulation do what I explained... native render to 3D objects with upscaled 2D artworks/textures.
We're not getting 4K native on any console coming in the next few years at the same quality we're getting 1080p on PS4. I'm expecting/hoping for some nice reconstruction techniques or something and 60fps where CPU allows.
Would that not be obvious? If you have twice the GPU power, they can uprender the games to a much higher fidelity. The problem is still that they are mandated to use 4:3 aspect ratios for games that do not support wide screen as opposed to PC emulation, where you can 16:9 any game you want, even those that don't look right in widescreen
Can and will are two different things though. They might just settle for it being locked at 60/30fps all the time and call it a day
serversurfer
Member
Exactly. To run UC4 or Spider-Man at 4k, you'd need 7.5TF or better, which is 25% faster than what they announced for Scorpio. So yeah, just double the frame rates, then spend the rest on eye candy, as you say.
How do you know they're not using it? If it's useful why wouldn't they?
Isn't that super troublesome for ps2 emulation? Look at how long it's taken pcsx2 to get where it is doing that. And how many hacks required...
http://www.neogaf.com/forum/showpost.php?p=199410564&postcount=8294
ethomaz
Banned
Because it is how it works on PS4.
http://www.eurogamer.net/articles/digitalfoundry-2015-hands-on-with-ps4-playstation2-emulation
BTW Sony has all the technical data from PS2 and PS4... there are not doing reverse engineering emulation like PCSX2... they do emulation with actual technical data.
Emulation knowing how it works will always be faster and better than reverse engineering.
I've said it several times before but even with the rumored 4.2tf GPU in Neo, it can handle 1080/60 with some solid visual enhancements over the base PS4. If the CPU is up to the task, it will make for a pretty substantial visual upgrade.
That article doesn't say anything about how they're achieving higher resolutions...
And if you read the post I linked you'd see why knowledge of the hardware or documentation would not really solve those problems.
serversurfer
Member
Sure, but the more, the merrier. Make it as powerful as you can for $499, I say. It'll give you that much more eye candy and help with games that currently struggle to reach 30 and it will make Neo that much more future-proof, meaning it won't drag so much on whatever the next PlayStation is. 4.2TF isn't terrible by any means, but 5.5TF would still be a lot better, especially with a 6TF XBox on the horizon.
Speaking of games that struggle to hit 30, I wonder if the mandated support for PS4 will create pressure for devs to improve performance minimum performance on PS4 games. Or perhaps more liberate than pressure. First, if the PS4 version dips below 30 while the Neo version holds a comfortable 60, that may be seen as a "shoddy port," both by gamers and by Sony. Also, I think part of the reason devs are so willing to drop frames is the drive to put extra stuff in the game. So perhaps the tiered hardware will help to establish 1080p30 and 1080p60 as the respective performance expectations, and Neo will be the place to go for enhanced performance and eye candy, whether you're a dev or a gamer. What say you, GAF?
Dreamwriter
Member
Nintendo kinda did that - they announced the existence of the 3DS when they knew rumors were about to start flying about it.
Totally agree with the first paragraph. I'm certainly not going to argue against more GPU power.
I was thinking the same for performance targets as well. Aiming for 60 with some visual refinements on the higher end model will allow for the standard PS4 to remain a nice looking and very playable1080/30 and there's less added work to improve effects and assets to compensate for the same framerate target with hugely improved visuals.
For first party games they may be limited to framerate increases (if they were unlocked at 40-45fps they might be able to get to 60); res increase/downsampling eg to 1440p, or better AA. Their options are limited without significant extra work as they wont have higher quality assets or effects that multi platform games that are also on PC will have. Depending on the game they could increase draw distance or LoD
Multi platform games will have access to ultra PC settings so they will have a broader base from which to choose the best match for neo. I'd argue here that you won't see any big push for increased framerates - existing 60fps games will still be 60, but 30fps games will stay 30. Partly because that means you can spend the increased power to get higher PC equivalent settings so your games look much nicer in stills, but a little bit to ensure things like multiplayer modes have a level playing field between neo and PS4. Mostly the pretty graphics though.
ethomaz
Banned
Did you read?
BTW did you play any PS2 emulated game on PS4? I'm playing right now Wild Arms 3 and 3D render is indeed 1080 horizontal lines... 2D textures and artworks are upscaled... I don't know how you can think other way.
And yes technally knows how the game uses a hardware is what do emulation better than reverse engineering... Sony emulation is already better in performance than PCSX2.
And your link was based on assumptions over reverse engineering and not direct emulation.
So what? I don't see how that conflicts with anything I said.
Again? If you really think that you either did not read the post, or you didn't understand it.
And how is this based on reverse engineering? Faf has worked on actual ps2 games, so that kinda seems like a joke to me. Like, are you serious dude? Seems like you're just making stuff up.
Anyway..., it was more about how resizing the render target is just inherently incompatible with a unified memory system.
As you can see, it's clear that knowing all the details of the hardware is irrelevant, because you're not respecting those details when using this method anyway, and also because the issues are not even caused by incorrect implementation(or lack of knowledge), but general incompatibility of the technique and the system in general.
ethomaz
Banned
Believe what you want... the fact PS2 games runs like I said in PS4 won't change.
Faf working or not in PS2 games didn't change PCSX2 do reverse engineering and for that it has performance issues while Sony emulation is actual emulation done right respecting all the technical aspects of the hardware emulated.
That is why you don't have unplayable mess on PS4 with PS2 emulation... the games are rendered fine in 1080 horizontal lines for 3D and upscaling the 2D parts.
You can actually play games to see that... so your accusation of "I am making stuffs" just proves false when you have over 20 games running on PS4 like I explained... I can take pictures if you wish... Sony doesn't need to complicate when you can change the 3D render just fine via emulation without issues... maybe you need to study a bit of emulation or in this case work with it to understand.
Ok, I think you just don't understand then. The method described in the patent would result in higher resolution rendering (it's just another method of doing it), while also avoiding the issues involved with a simple upsize.
And I already know a thing or two about emulation, I'm part of the pcsx2 testing team, and I've worked on a couple of my own (admittedly pretty simple) emulators myself(not that it really matters).
ethomaz
Banned
The method used in the patent try to recreate a higher resolution using lower resolution rendering... it is not the same lol... the use of this tech can give the impression of better/sharper IQ than simple upscaling but there is disadvantages too like worst IQ in moving scenes or artifacts... rendering in higher resolution will be always give a better/sharper IQ.Ok, I think you just don't understand then. The method described in the patent would result in higher resolution rendering (it's just another method of doing it), while also avoiding the issues involved with a simple upsize.
And I already know a thing or two about emulation, I'm part of the pcsx2 testing team, and I've worked on a couple of my own (admittedly pretty simple) emulators myself.
What PS4 does to emulate PS2 games is nothing like that.
PS4 runs all the 3D game code in high resolution... there is no upscaling or weird temporal reconstruction tech involved... think like you running a game on PC and changing it resolution... the issue is that 2D artwork/textures are not created to fill higher resolution so they needs to be upscaled... simple image upscaling.
PS2 games render are dynamic like all games and can be set to render at right resolution just fine... the issue are non-rendered objects like 2D texture/artwork that needs to be upscaled.
A better example...
Just think about a PC game... you can change the resolution ingame but if the assets are made in lower resolution they will need to be upscaled... you can actually run old games in PC at 4k with 2D assets upscaled... that is what happens on PS2 emulation.
In terms of IQ you have in better option: render in higher resolution (PS2 emu) > retroconstruction based in lower resolution (patent) > upscaling from lower resolution.
The patent will be most used because it does a better job than upscaling but Neo won't have power enough to render in 4k... that is not the case of PS2 emulation that didn't stress PS4 to do it in higher resolution and better framerate.
serversurfer
Member
Are you talking about the patent Thuway linked a few pages back? Because that doesn't increase rendering resolution at all. It doesn't affect rendering in any way, because it's all applied long after the image was created. If you actually read the patent, it's talking about upscaling "multimedia content," so like, movies and stuff. Sure, you could do the same thing with live, game video, but it still won't have any effect on the rendering itself.
This is just an upscaling technique. Nothing whatsoever to do with rendering; it's all done after the fact.
Ok I'm not sure who to believe then.
This is directly from that PDF
"In some instances, the uprendering is applied to legacy PlayStation® electronic games that often generate frames having a resolution of [640×448]. Accordingly, the uprendering can be applied to increase the resolution of frames as the game is being played back to generate higher resolution images.
...
the resulting uprendered image 220 results in twice the resolution of the source image based on the 2×2 uprendering matrix (i.e., a resolution in this example of [1280×896])."
Right. Not "higher resolution rendering". That was the wrong choice of words. Didn't get enough sleep last night, so my brain isn't working too well right now.
ThoughtsOfSpeaking
Banned
Some serious pixy dust being snorted here. Now there are 2 neos?
People not understanding the difference between up-scaling & up-rendering is not my problem.
The new Xbox One will upscale all games to 4K & your 4K TV can upscale all games to 4K but PS4 Neo & Xbox Scorpio will most likely up-render games to 4K..
Exactly. If people would stop replying to that nonsense interpretation of upscaling, there wouldn't be any issues.
There have been two Neos floated as being in development for a long time, depending on what Sony wants to go with.
This is the initial rumor on that.
Nope. This simply isn't true. You could easily get there or almost there with dynamic resolutions. Staying at 1080p is a horrible waste. At least go to 1440p.
Why? Until 4K adoption becomes widespread there's no need to go higher than 1080p.
I'm sure some games will use downsampling but that's about it.
Because people buying premium sku consoles probably have premium TVs. It's not a stretch. I own a 75" 4K tv and have had 4k for 3 years now. 1440p minimum. 1080p is really ass once you've had 1440 or 4K. I've been gaming on 1440 on PC for 5-6 years now it's great.
1440 on a 4K tv will pull the same results of blurring details as running 720 on a 1080p.
Panajev2001a
GAF's Pleasant Genius
The way Sony is approaching higher resolution rendering is more accurate without requiring hacks by virtue of how it is rendering: you are essentially rendering the same scene 4+ times only changing the camera projection very slightly each time then accumulating the frames once you collect them. They were kind of doing it in hardware on the GSCube and it is like having 4 PS2's rendering in parallel and then using some component to merge the frames they produce.
This way each virtual GS is seeing the same memory size and the same exact layout... it will not know what other people will do with its output
.Panajev2001a
GAF's Pleasant Genius
The patent I saw was talking about essentially generating 4+ frames and the pixels in each were unique (sub pixel camera projection changes would produce Just that, a kind of weird way to do SSAA and sell the output unmerged
), but the point was compatibility with the way games used to treat the GS and its eDRAM that would break many of them if you simply forcefully increased the backbuffer and front buffer resolution. Given the amount of hacks in PS2 emulators on PC necessary to achieve it with a decent degree of compatibility, it is understandable Sony went a different way to minimise compatibility issues.Their approach is clever if you think about it, instead of messing with how PS2 games made direct and exact use of GS eDRAM, depending on its size and various buffers exact locations and dimensions, they are using 4+ parallel virtual PS2 consoles each running the same code except for a very very minor change in how the scene is projected initially.
lightchris
Member
I only glanced over said patent but the base principle seems to be very simple.
Instead of increasing the rendering and frame buffer resolution as it is common, the method indeed just renders at the original native resolution (something like 640×448 for PS2 games as quoted above).
It's not just rendered once though, but multiple times with the only difference being shifted sample positions.* Then in order to create the final larger resolution image, all of the smaller images are combined.
So for example for a 4x increase in resolution (2x along each axis), the same scene is rendered 4 times at native resolution. In the end you have 4 color values per pixel, one from each of the 4 images. These 4 values are then spread out to the final image, so that eg the 4 pixel square in the upper left corner of the final image consists of the 4 top-left pixels of the smaller images.
This method is a bit more expensive than rendering the scene at 4x the resolution right away, but it can help eliminate errors that can occur otherwise if the application was meant to be rendered at a sinle fixed resolution. It should be noted that there are no disadvantages in IQ. The result is a true high resolution image.
* A sample position is a point within each pixel at which the geometry is evaluated, eg. whether a triangle is hit (and at which coordinate). Usually when not using any traditional anti-aliasing the pixel center is taken as the default sample position. The sample pattern examples at Wikipedia may explain it better; in this case the top left grid is used.
The patent isn't about temporal reconstruction (unless we're not talking about the same one).
I think there might be a general misconception about 2D elements. Elements like HUD or text are just textures slapped onto simple polygons (usually a quad) and are rendered the same as everything else. The GPU doesn't know about 2D, there's no explicit upscaling, just the usual texture sampling that is used for every other texture too. Of course the texture resolution doesn't increase when the rendering resolution is increased, so at some point every texture will look blurry (as soon as they are magnified).
edit:
Seems like I took too long to type this post. Panajev2001a already explained.
The way Sony is approaching higher resolution rendering is more accurate without requiring hacks by virtue of how it is rendering: you are essentially rendering the same scene 4+ times only changing the camera projection very slightly each time then accumulating the frames once you collect them. They were kind of doing it in hardware on the GSCube and it like having 4 PS2's rendering and then some component merging the frames they produce.
This way each virtual GS is seeing the same memory size and the same exact layout... it will not know what other people will do with its output
Yeah, definitely a cool trick.
Nah it'd be much better. 720p simoly wasn't very many pixels to scale up, but 1440p is a much better base to work from. If scaled internally on the console it'll look great.
Except the post you quote talked about running it at 4K. To do that you would need 4x PS4 spec. You'd need less power at a lower res, but then you would be running at 4K as stated.