Next Generation Hardware Speculation with a Technical Spin [2018]

Status
Not open for further replies.
Has anything changed for the nVidia option regards CPU provision? nVidia GPU either means 1) nVidia GPU + Intel CPU, 2) nVidia GPU + AMD CPU, or 3) nVidia GPU + nVidia ARM CPU.

1) Expensive
2) Would likely get better deal with all AMD
3) Whole issue of ARM's viability as discussed at length

If MS get Windows running as quickly on ARM as on x64, it's a realistic option for MS. However, they want cross-device compatible with x64 PC more than anything, so as long as ARM performance is going to interfere with that, the next XBox has to be x64, which pretty much rules nVidia out.
Nothing yet.
Just thinking back to //Build2015 the big thing was this:
https://bgr.com/2018/10/12/surface-phone-foldable-phone-might-be-real-microsoft-patent/

And I very much am waiting to see the hardware spec there because continuum is a big part of what this device is supposed to be capable of. I know that the product is still moving ahead, at least no reports of the project being stopped yet, but it appears to be delayed yet again, I was expecting it at //Build this year. Perhaps next year might be the time. I think whatever they can pull off here, would let me know if ARM would be an option. But to be honest, I don't even know anything about the specs of this device, the CPU was largely hidden in secrecy back when we were presented to.

The reason I'm looking at this closely is because the requirements are generally at odds with each other. Mobile needs low power, but continuum needs high processing power.
 
Last edited:
Most games don't make use of hardware flexibility. Pretty much all of them use the same technologies and that's even more true now with the move to PBR and the convergence to a handful of engines across the industry. Speed benefits the vast majority of games versus flexibility for just a few.
Nonsense. Most "standard" features today were anything but standard when the hardware of PS4/ONE were designed. Modern engines do use a lot of novel techniques that are only possible today thanks to the flexibility provided by these machines.
Dreams is just the most extreme example, where the whole rendering pipeline is compute based and foregoes even the most basic primitives typically used.
But there are other commercial games TODAY using SDF ray-tracing for parts of their rendering, like for example: Fortnite. Their large scale AO raytraces against an SDF representation of the scene constructed out of smaller SDF volumes. It is in fact a standard feature of UE4, and can be used for variable penumbra shadows too.
Similarly, all modern engines today are moving towards variable resolution deferred renders with temporal injection and reconstruction for upscaling and AA. None of that is a standard feature of GPUs, nor were they envisioned by GPU designers, they were clever software tricks developers were able to pull off within the constraints of DX9 level hardware initially, and really started flourishing now on the DX11/12 level machines of today.
For geometry processing, most big AAA engines are already doing things similar to mesh-shaders and NGG on consoles using compute because the API is less restrictive there (our own sebbbi was a pioneer in that front together with the devs of AC Unity). GPUs are actually catching up to an idea they probably wouldn't have had themselves were it not for devs experimentation.
Texture Space shading, and before that, Tiled Resources are the same thing. GPU designers took a look at what carmak (and our sebbbi again) had done with virtual texturing by software, and adapted their hardware design to empower such tech and make more viable and efficient.
Man, I could go on and on about software solutions no gpu engineer would have ever thought of if it were not for their designs having become flexible enough to make those ideas possible. Deferred rendering itself is one. POM. Virtually all Screen Space effects such as AO, SSS, Reflections, Specula Occlusion (very important component of RDR2's look), soft z-rested particles, gpu particles...
 
After seeing the results of vega on 7nm, I'm more inclined to believe it's 2020 now. I know navi isn't vega, but 2019 just seems to soon to me now. Consoles generally don't go with cutting edge manufacturing, and 7nm for gpu is very new.

TSMC have producing ~15 million 7nm chips a month (since the summer) to satisfy Apple's 2018 iPhone XS, XR and new 2018 iPad Pro as well as Huawei's 7nm chip needs - apparently without any difficulties and certainly at much higher qualification tolerances that any chip in a console demands.

Sony have been using (exclusively I believe) TSMC for both original PS4, Slim and Pro at 28nm then 14nm and AMD have moved their 7nm to TSMC as Global Foundries gave up. It's difficult to imagine TSMC having any difficulty with what will be, relative to higher-end PC parts, a low-power low-to-midrange APU. TSMC must be on their third or fourth production iteration at 7nm already, this is a mature process for TSMC. The jumps from 14nm to 10nm to 7nm are not complex in manufacturing terms. TSMC rolled a 10nm process mostly just for Apple - I can't think of any other customers for TSMC's 10nm process :runaway:

Has anything changed for the nVidia option regards CPU provision? nVidia GPU either means 1) nVidia GPU + Intel CPU, 2) nVidia GPU + AMD CPU, or 3) nVidia GPU + nVidia ARM CPU.

Nvidia have been getting real cordial towards ARM solutions this past year, an all-in-one APU doesn't sound as crazy as it would have eighteen months ago. :nope:
 
Gamesworks doesn't move the industry forward. Hardly. Gamesworks has been around forever, and only when consoles moved to DX11.2+ hardware did the rest of the industry move as well.
Hopefully it was clear that I never said it moved the industry forward in any particular way. I was highlighting the fact that NVidia doesn't need a console play to push and have their tech used in the way AMD does.

TSMC have producing ~15 million 7nm chips a month (since the summer) to satisfy Apple's 2018 iPhone XS, XR and new 2018 iPad Pro as well as Huawei's 7nm chip needs - apparently without any difficulties and certainly at much higher qualification tolerances that any chip in a console demands.

Sony have been using (exclusively I believe) TSMC for both original PS4, Slim and Pro at 28nm then 14nm and AMD have moved their 7nm to TSMC as Global Foundries gave up. It's difficult to imagine TSMC having any difficulty with what will be, relative to higher-end PC parts, a low-power low-to-midrange APU. TSMC must be on their third or fourth production iteration at 7nm already, this is a mature process for TSMC. The jumps from 14nm to 10nm to 7nm are not complex in manufacturing terms. TSMC rolled a 10nm process mostly just for Apple - I can't think of any other customers for TSMC's 10nm process :runaway:
Manufacturing phone soc's maybe not, gpu design may be harder. That's why I said gpu's, i don't really count phones soc as part of the discussion as if that was any indication they would have consumer gpu cards out by now

So in the context of apu/gpu's they would be early adopters to that process for a 2019 release, for chips of the size that consoles would be using.
 
Manufacturing phone soc's maybe not, gpu design may be harder. That's why I said gpu's, i don't really count phones soc as part of the discussion as if that was any indication they would have consumer gpu cards out by now

Why do you think this? Have you seen the int/fp performance of the A12X chip in the fanless iPad Pro, or the GPU performance? If you want further reassurance, TSMC are manufacturing all of AMD's 7nm GPUs this year.
 
Why do you think this? Have you seen the int/fp performance of the A12X chip in the fanless iPad Pro, or the GPU performance? If you want further reassurance, TSMC are manufacturing all of AMD's 7nm GPUs this year.
I'm basing it on the current situation.

The first 7nm gpu is vega 20 launching end of 2018, (I consider now end of 2018)

I don't expect any mainstream 7nm gpu to launch this year. Unless you know different, when do you think they may come?

The console would be one of the first gen consumer class 7nm apu/gpu.
I personally don't see MS or Sony doing that, but that is just based on my views of them wanting to go on a proven manufacture for type of chip.

Is there any reason why NVidia and AMD haven't already launched consumer 7nm yet, what is their timescales.
NVidia could've used it for their latest 20xx cards, AMD starting with the 590 polaris refresh. Serious questions.
 
But there are other commercial games TODAY using SDF ray-tracing for parts of their rendering, like for example: Fortnite. Their large scale AO raytraces against an SDF representation of the scene constructed out of smaller SDF volumes. It is in fact a standard feature of UE4, and can be used for variable penumbra shadows too.
I didn't know this. This is one of the possibilities we were discussing when suggesting flexibility trumps RT specific hardware.

GPUs are actually catching up to an idea they probably wouldn't have had themselves were it not for devs experimentation.
Very much so. It's the software devs who pioneer and push ideas, and then the best are incorporated into hardware. RTRT isn't nVidia pushing the industry forwards, but a decades old rendering solution pioneered on flexible CPUs finding its way finally to GPU acceleration off the back off flexible compute-based raytracing, compute itself being an adaptation of hardware to accommodate what software engineers were achieving with GPGPU fitting non-graphics tasks to the graphics pipeline to get massive acceleration.

As you say, pretty much everything in GPUs is driven by software.
 
I don't expect any mainstream 7nm gpu to launch this year. Unless you know different, when do you think they may come?

You misunderstand, I only query why you think a 7nm GPU may be harder to fabricate that other chips. Modern GPU designs are far more tolerant of fabrication issues given their massive parallel designs - certainly compared to a quad/octa-core CPU. As for why they're not out there, I presume because they aren't ready yet.

I think that's more a Nvidia/AMD thing than a TSMC thing.
 
Last edited by a moderator:
You misunderstand, I only query you think a 7nm GPU may be harder to fabricate that other chips. Modern GPU designs are far more tolerant of fabrication issued given their massive parallel designs - certainly compared to a quad/octa-core CPU. As for why they're not our there, I presume because they aren't ready yet.

I think that's more a Nvidia/AMD thing than a TSMC thing.
I believe they have their reasons, and that ms and Sony wouldn't want to go first or be part of the first wave.
Given the size etc of 20XX, and where AMD are at in the consumer space, I can see why both would've released on 7nm if their designs were ready for it and there was no risk involved.

As I said before, my heart would like one of them to be next year, so I wouldn't mind being wrong.
 
I didn't know this. This is one of the possibilities we were discussing when suggesting flexibility trumps RT specific hardware.
The feature in question:
https://docs.unrealengine.com/en-us...htingAndShadows/DistanceFieldAmbientOcclusion

This dynamic AO solution will not work for all projects, as it has some invasive limitations which enable it to support dynamic scene changes. In particular, only slight-non-uniform scaling (squishing) is supported. Also, large Static Meshes produce inferior quality results because a small volume texture is mapped to every object.
 
If AMD was making Navi or at least making lots of changes to it for Sony like that tweet suggested about two thirds of AMD graphics engineers working on Navi for Sony was true. I don't see how AMD releases Navi next year and Sony only using it at the end of 2020.
 
If AMD was making Navi or at least making lots of changes to it for Sony like that tweet suggested about two thirds of AMD graphics engineers working on Navi for Sony was true. I don't see how AMD releases Navi next year and Sony only using it at the end of 2020.
Does anyone know how the teams are usually set up there?
I wouldn't be surprised to find that they have teams working on Navi and the following uarch in parallel.

I would have thought that Sony would have been making modifications to the base navi uarch, but requiring 2 thirds of the team sounds pretty big, guess also depends on how long that amount of staff was needed.

Would also be interesting if it affected navis pc release, and if that pushed it (to) close to amd next gen uarch release
 
I like the way you're pro-RT stance highlights the negatives. ;) To me, the results look great and the performance runs fine on existing GPUs. That's a significant win, a great piece of tech, and a great early adoption of SDF which is far newer tech than raytracing (only 20 years old?).
I dabbled with raytracing on the Amiga in 1987. Oh yeah, calculating those rays on a 7.14mhz 68000 processor. :runaway:
 
If AMD was making Navi or at least making lots of changes to it for Sony like that tweet suggested about two thirds of AMD graphics engineers working on Navi for Sony was true. I don't see how AMD releases Navi next year and Sony only using it at the end of 2020.
It was never true. 2/3 assigned to Navi might have been. Semi-custom SoCs are made from completed IP-blocks. Sony, MS and other similar big customers of course give some input on what they would like to see, but first and foremost AMD designs the IP-blocks for their own needs and goals and those can be then used to craft the semi-custom solutions (with possible 3rd party IP-blocks added of course)
 
Status
Not open for further replies.
Back
Top