ID buffer and DR FP16

Hi everyone I have a question

as you know the new "xbox one x" is coming and its a very powerful console with 6 TFLOPs in comparison PS4 Pro only have 4.2 TFLOPs and less memory BW

but theres is something curious about both consoles, X.B.O.X. is "The world's most powerful console" ™ but despite all the "True 4k" logos and mumbo jumbo by Phil Spencer, it seems high profile games are using reconstruction techniques like checkerboard rendering to get to 4K much in the same way as PS4 Pro does

but the PS4 Pro aparently has some extra hardware to help in reconstruction techniques such as ID Buffer and double rate FP16

as far as I understand with FP16 PS4 Pro can double TFLOPs at the the cost of precision but this is not useful everywhere but it has its use and hardware ID buffer helps to know were are the triangles of each object and that helps a lot for checkerboard rendering and to other technique where the game can render the geometry in 4k from a 1080p framebuffer but using the pixels from the textures of the original 1080p resulting in blurry textures(for a 4k image)

all this is explained by Mark Cerny in an article from digital foundry called
Inside PlayStation 4 Pro: How Sony made the first 4K games console


I remember a user here sebbi, posted a link to a very interesting DICE presentation about how all this makes a 30% boost in the frostbite engine

sorry I cant post the links because there is a red banner that says I cant until I post 10 messages :S

both ID buffer and DR FP 16 are AFAIK not present in X.B.O.X. and ID buffer in particular cannot be implemented in software "without a huge impact on performance"

I assume X.B.O.X. is powerful enough to implement all this and still have an advantage but I am interested in knowing how close PS4 Pro can get using all this very clever techniques and maybe someone here can comment about it(without breaking NDA )
 
ubisoft says the new Assasin's Kreed runs equal on both X and PRO... I guess they need to use this FP16 but only PRO has it double rate per clock cicle... Big mistake MS did.
 
Hi everyone I have a question

as you know the new "xbox one x" is coming and its a very powerful console with 6 TFLOPs in comparison PS4 Pro only have 4.2 TFLOPs and less memory BW

but theres is something curious about both consoles, X.B.O.X. is "The world's most powerful console" ™ but despite all the "True 4k" logos and mumbo jumbo by Phil Spencer, it seems high profile games are using reconstruction techniques like checkerboard rendering to get to 4K much in the same way as PS4 Pro does

but the PS4 Pro aparently has some extra hardware to help in reconstruction techniques such as ID Buffer and double rate FP16

as far as I understand with FP16 PS4 Pro can double TFLOPs at the the cost of precision but this is not useful everywhere but it has its use and hardware ID buffer helps to know were are the triangles of each object and that helps a lot for checkerboard rendering and to other technique where the game can render the geometry in 4k from a 1080p framebuffer but using the pixels from the textures of the original 1080p resulting in blurry textures(for a 4k image)

all this is explained by Mark Cerny in an article from digital foundry called
Inside PlayStation 4 Pro: How Sony made the first 4K games console


I remember a user here sebbi, posted a link to a very interesting DICE presentation about how all this makes a 30% boost in the frostbite engine

sorry I cant post the links because there is a red banner that says I cant until I post 10 messages :S

both ID buffer and DR FP 16 are AFAIK not present in X.B.O.X. and ID buffer in particular cannot be implemented in software "without a huge impact on performance"

I assume X.B.O.X. is powerful enough to implement all this and still have an advantage but I am interested in knowing how close PS4 Pro can get using all this very clever techniques and maybe someone here can comment about it(without breaking NDA )
I'd imagine a completely optimized game for 4Pro could come pretty close to what Scorpio can do.
The realities are fairly straight forward though.
a) Checkerboard offers significantly good returns on image quality for it's performance impact
b) FP16 will assist with a few areas of the rendering pipeline, but if we're talking about achieving the same level of image quality between the two, then this feature becomes has less impact
c) Checkerboarding can be accomplished without ID buffer, as shown by Ubisoft when they released Rainbow Six Siege. So the huge impact on performance statement needs some more context as to 'what'.

Where it gets interesting is that because checkerboard offers very good return on image quality for it's performance boost, there's no reason to not use it on 1X in this case, which favourably scales in X1X favour. As for example an arbitrary 50% of a larger number would result in benefiting the hardware with the larger number.

The issue with checkerboarding in general is that it is _not_ a hardware feature. Each company can implement checkerboarding differently, with or without ID buffer. There is no consistency between the performance or quality of each companies algorithm. At least, I don't think there has been a standard developed for it yet.

In situations where a company does not have the capacity to build checkerboard rendering, Native 4K can be achieved on X1X with much higher probability than 4Pro. Other factors like texture resolutions require bandwidth in which X1X has the memory to support 4K assets, where 4Pro is less likely to find success here.

Rounding it out, I'd prefer not to debate about marketing talk, I think a lot of us agree that the messaging is not exactly fair, or clear. But at the end of the day the Xbox One X, will have fewer asterisks in achieving 4K Frame Buffer games than 4Pro, and it will likely keep this trend. You will see a great deal of many CBR titles up to 1800p, and then a final upscale to 2160p on 4Pro. In these situations, X1X will probably clear CBR straight through to 2160p. I expect this to be the case for a lot of graphically challenging titles.
 
i would need to go look at the dice presentation again.
but from what i remember the fp16 mostly helped to reduce register pressure, which will be just as effective on 1X which also has fp16, just not RPM.

ID Buffer - wasn't it something like 30% improvement on a task that was 10% of the overall rendering time? So far from 30% overall improvement.

assume this is the presentation your talking about here's the link
http://www.frostbite.com/2017/03/4k-checkerboard-in-battlefield-1-and-mass-effect-andromeda/

worth noting, we don't know what huge impact actually means. loosing say 20% performance on a task that was 10% of the render budget when you have a TF advantage which is magnified by the performance gains of using CB may not amount to much. Apart from being good benefit for 4Pro.
 
Last edited:
I think I'd like to wrap up my commentary that most people are interested in looking at RAW specs and ignore the finer details (easy to generalize discussion, not the fault of the participants). In this case, the customized features of 4Pro are well marketed, and the console has been out for 1 year now, with some additional time before release.

Games must take advantage of those features for 4Pro to make greater gains over it's base spec. The best companies will get the best results.

There's no particularly interesting or marketable feature of Xbox One X. But the whole SoC has been profiled against a great deal of many titles that were made in the past. Profiling allows them to make changes to the design and simulate the results. They know, for a great deal of many titles, how those titles would perform on Xbox One X before the silicon was pressed.

The difference between the two approaches is that with 4Pro you have _potential_, with Xbox One X, you have a confidence interval.

In layman terms, Native 4K/60 fps for Forza 7 on Xbox One X was not an accident, it wasn't an 'wow you made a great piece of hardware look how good our code is working. They already knew with reasonable certainty how Forza 7 would perform on the final hardware.
 
i would need to go look at the dice presentation again.
but from what i remember the fp16 mostly helped to reduce register pressure, which will be just as effective on 1X which also has fp16, just not RPM.

ID Buffer - wasn't it something like 30% improvement on a task that was 10% of the overall rendering time? So far from 30% overall improvement.

assume this is the presentation your talking about here's the link
http://www.frostbite.com/2017/03/4k-checkerboard-in-battlefield-1-and-mass-effect-andromeda/

yes that is, and yes the improvement was in part of the rendering no the whole framerate
 
I have no problems with cerny saying huge performance impact when not using buffer id.
as that may be true in the context of that small part of the render time. But how much it really impacts the overall render time once alternative methods are used we don't know. From what I remember they said they had reasonable solutions for all the other platforms.

may have to wait for dice gdc presentation whenever they do a follow up to know in their case.
be nice if DF gets to do a deep dive with a dev and get a rough idea.
 
Didn't AC: Syndicate run at same res on both PS4 and Xbox. Hard to use AC: Orgins as a good comparison if they may be forcing parity.
AC:S
wasn't it cpu limited, swear also made a comment about not wanting to upset anyone or something crazy like that. But don't quote me on that.
we'll have to wait and see what they actually mean by "same experience" as that may not actually mean actual identical image i.e 2160cb v 1800cb, other updates but you get the same experience.... No idea.
But would be funny if they forced 1080p but no downsampling on 1X (sorry slight of topic)
 
There's no parity. There's a dynamic resolution on both consoles.
I'm not going to derail this thread, but why does everyone think resolution is the only thing that makes up the image? not specifically you, just seems to be the only thing people are concerned with, could understand with XO and PS4 as 90% off difference would be only res and framerate
this would need to go into a different thread.

editing my post as not going to continue this of topic discussion by posting again
Because Ubi only mentioned resolution : http://64.91.255.7/forum/showpost.php?s=90f84cfefb5e8b72e63846b800c0679b&p=240809424&postcount=1
Now, it is quite possible that Origins has better assets on X (more ram).
just proves my point, only thing was resolution, no disrespect to jez but he's far from technical.
he's asked about resolution, and whole article was about that.
 
Last edited:
Yes according to Cerny spatial and temporal anti-aliasing can both (and at the same time) use the ID buffer that track the triangles IDs and the objects IDs.

With the help of the triangles IDs and objects IDs, "some excellent spatial anti-aliasing" reconstructs the image (without the previous frames) and temporal anti-aliasing with the help of the ID buffer uses "an appropriate part of the previous frames" for some "pretty darn good temporal anti-aliasing", even in motion.

It's going to be very interesting to compare the image quality and AA (particularly in motion) of Battlefront 2 on XBX and Pro...
 
What does it matter when every 4K TV anyone is playing the game(s) on can't resolve the resolution in motion? I dont think its going to be interesting at all in the comparisons. The only thing the comparisons will show are the difference in 1 static nonmoving frame.

You'd have to drop down to 1080p Plasma display to see the differences in motion. The trouble with that is you lose out on HDR which has a higher impact on the image quality than resolution increases.
 
What does it matter when every 4K TV anyone is playing the game(s) on can't resolve the resolution in motion? I dont think its going to be interesting at all in the comparisons. The only thing the comparisons will show are the difference in 1 static nonmoving frame.

You'd have to drop down to 1080p Plasma display to see the differences in motion. The trouble with that is you lose out on HDR which has a higher impact on the image quality than resolution increases.
is that true of all 4k tv technology or not?
 
is that true of all 4k tv technology or not?
Motion resolution is going to depend on the display panel itself. I don't think there is an association to all 4K. If it is it may get better over time
 
Back
Top