Forbes: AMD Created Navi for Sony PS5, Vega Suffered [2018-06] *spawn*

Well it is highly likely Cerny told Sony what they need for PS5 and then went to AMD and said "thats what we need", therefore AMD implemented any solutions they developed and any input or feedback the received in their product line in general. Which of course will find its way in a lot of other products
I see nothing special in particular.
The same probably counted for Microsoft unless MS made sure nothing would go out until the next XBOX is released.
 
AMD press release for RDNA made a statement about the GCN being in 450 million devices. Roughly 150 million of these devices are consoles across only 6 skus. While the other 300 million gpus span 5 different GCN gens across dozens of chips in 100s of different products. Console manufacturers have an order of magnitude more influence than any other outside buyer.
You've got that back to front. It makes little sense to have Sony design a console GPU architecture to then use as your basis for PC parts when those PC parts are going to massively outsell the console parts. Note GCN sold to 2x as many units as consoles even though it wasn't a great, popular architecture. RDNA is hoping to sell many, many more units. It can't do that if it's hampered from being a great versatile part because it's bespoke for a console.

Sony : We want RDNA to have this.
AMD: But that'll be really poor efficiency for PC.
Sony: Don't care, it's going in a closed box. Also, we want that.
AMD: But that's the very opposite of what HPC is going to need for efficient compute.
Sony: Don't care, we just want it for a console.
AMD: Right, everyone. Here's our bespoke console GPU. Now let's try and sell these to PC gamers and HPC...

Any PC part can be made into a console, so obviously (so obvious, AMD wouldn't do anything else unless they were effectively bankrupted and needed to be bailed out) AMD are going to design a GPU uarch that'll serve their wider industry goals of a billion units first and foremost. That's their priority. Anyone wanting customisation is going to have to take what AMD is doing for its own goals and tweak it.
 
Last edited:
The only case of were a company arguably "reacted" that I can think of because of a power difference was when Microsoft upclocked the Xbox One a few months before launch and that was 53MHz....and it was probably only possible at that point because of the overly conservative design of Xbox One due to the RROD issues with 360.

It's possible the 1.8GHz -> 2.0 GHz was a "reaction" but a completely new chip I doubt.
A reaction to what, exactly? There was no news of XsX at the time.

Has anyone actually disproven the possibility of Microsoft just moneyhatting extra Tflops from Sony. Like they paid them $1B or something for them to stay at less than 10 Tflops so they could have double-digit Tflop exclusivity for this gen?

Did Phil Spencer deny this yet?
Guess Nintendo got the shit end of the stick and nerfed the Switch for free?
 
PPE was just part of the project of CELL, it wasn't created for Sony.
Yes, the PPE CPU core was created for Sony and Toshiba, to use in Cell. Sony was one of the two of the technology takers in the consortium created to develop the Cell microprocessor.

I advise you to read up on consortia focused on R&D and technology transfer, which is exactly what happened with the STI consortium, and what happens when Sony or Microsoft partners with AMD.

Why are you still on this? It just shows how pedantic you are if you can't drop a point like this.
You mistook the entity who developed the technology (research & development & technology maker) for the entity who ordered the technology (technology taker). They are two different entities in a consortium.
And when you get called out to this very obvious mistake, instead of learning from your mistakes you have the audacity of resorting to gratuitous name-calling (though not before trying to move goalposts).

I'm done feeding you.


The idea of Sony funding Navi and everyone getting it is very unrealistic. Sony knew AMD were creating a GCN successor. Why would they want to fit the bill and fund their rivals? Why not just let AMD get on with it and buy into whatever they produced? There'd have to be something exceptional to the deal, like some tech only Sony could use for a period, or which was refused on other platforms (maybe allowed on PC but not other consoles).
There's a number of reasons why Sony could do it and you named some of them.

1 - Timed exclusivity for the whole architecture. If Sony really entertained the idea of a 2019 launch, they could have demanded e.g. exclusivity to use RDNA in a console until mid-2020 (which then fell on its face when they decided to postpone to late 2020).
2 - Timed (or not) exclusivity for certain IP modules, e.g. adopting AMD's preferred approach for real-time RT hardware implementation while forcing Microsoft to use other IP like IMGTec's.
3 - Timed exclusivity for specific low-level electrical optimizations, such as techniques that allow the GPU core to run at higher frequencies with lower power.
4 - Timed (or not) exclusivity to an ISA evolution (e.g. RDNA 2.0 instruction set) that allows higher real-world performance at similar FP32 throughput levels (it's perfectly possible Sony did this for Neo's exclusive use of Vega-like CUs on the mid-gens)
5 - Securing long-term discounts after an initial cash influx that fund additional resources available at the start of the development. After all, it's AMD who orders the wafers from TSMC and then sells the chips to Microsoft and Sony.
6 - A combination of the above, and others I couldn't think of right away.



It's far more realistic that Sony just had some involvement, perhaps regards supporting BC.
I disagree.
Sony subsidizing Navi from the start to get exclusivity of some kind, or Sony just purchasing a finished GPU architecture made by AMD, or anything in-between are all perfectly realistic.
You may argue that the former option is less probable, and on that I'll agree with you. Though that doesn't make Jason Evangelho's suggestion any less realistic.
The PPE on the X360 happened, and that's just one very important example (for how closely related to the console industry it is) on how a company can adopt technology that was partially funded and kickstarted by a rival.

The trouble with paying any attention to "anonymous sources say" journalism is that it could be anything.
He never claimed his sources were anonymous.
When Jason Schreier comes up with exclusive info, he doesn't disclose his source either. It doesn't mean his sources aren't vetted before he posts anything.

In the end you either trust the journalist or you don't. Jason Evangelho isn't just your regular youtuber, he's a journalist with a good amount of technical knowledge. He worked at AMD in technical marketing (a position that AFAIK is usual held by engineers) when both the PS5 and SeriesX were well under development. I'm yet to find out an article from him that is just respewing unsubstantiated rumors, or some gratuitous clickbait.

B3D in general was way too quick to dismiss and dogpile on Jason Evangelho and his article, based on either plain ignorance over possible technology transfer deals and/or ignorance on not even reading the article. To the point of believing the article is suggesting Sony developed Navi.
 
The PPE on the X360 happened, and that's just one very important example (for how closely related to the console industry it is) on how a company can adopt technology that was partially funded and kickstarted by a rival.
It's a very different example though, not at all comparable. IBM created Cell for Sony. In the process of creating Cell, that no-one else was going to use, a part of Cell, a smallish percent, helped created a new Power ISA processor. In essence, IBM didn't need to create a new from-scratch CPU for MS because they already had the work done in creating the PPU. What MS got out of Sony's investment wasn't Sony's investment - they didn't get Cell. MS didn't look at Cell and say, "Hey, IBM, can we have a triple-PPU processor, please." What MS got was an IBM Power processor, and what IBM got was a nice deal because they didn't have to spend a lot to create this processor as they'd already done the ground work in researching a small Power processor for use as the PPU.

The Cell comparison would be Sony investing in Cell for use in PS3 and then MS and Apple and PC getting it from IBM without any of the R&D costs. That's not happening here. AMD were going to make RDNA whether Sony were involved or not (I think you can agree on that!). Ergo, Navi was not created for Sony and Sony wouldn't be bankrolling AMD's products for use with AMD's other clients. Heck the investment needed in creating a new GPU architecture would be more than its worth - how many billions has had to be spent to create RDNA? You think Sony paid for that to get a console GPU only to let the rest of the world benefit?? Why not just take whatever AMD provides anyway at much reduced cost?

Also, if Sony were entering such a major partnership, don't you think they'd need to tell their investors? "Yes, we're bankrolling AMD's R&D for the next few years to create our next console SOC. Yes, they'll be able to use the IP after us for other clients but we get 40% of all sales in return." If it comes out that Sony has financed their rival platforms - both MS and PC - without letting anyone know and without clear financial gains at the end, that's not going to go down well.

The greatest involvement Sony could have would be without Navi being created for Sony, but with Navi being created by AMD for AMD and Sony wanting to be involved so investing to bring some preferred modifications to the mix.
 
B3D in general was way too quick to dismiss and dogpile on Jason Evangelho and his article, based on either plain ignorance over possible technology transfer deals and/or ignorance on not even reading the article. To the point of believing the article is suggesting Sony developed Navi.
It was and continues to be the correct thing to do here at B3D. It is on Jason or anyone else who makes the claim that Navi was designed for Sony to adhere to the burden of proof of which none exists.

Despite showcasing that there is some prior evidence in the past of which some technology may have been developed for Cell but found it's way to Xbox; you do not have full details or understanding of that deal - your evidence is submitted but I'd ask yourself is it bullet proof. There is still a high degree of skeptism on how the actual licensing of technology, the responsibilities of development and who owns what parts of what technologies and how much of it was already designed prior to CELL.

But more importantly what happened in the past is has no evidence of a relationship to what happened with Navi. We do not have any significant "data points" that suggest that Navi was designed for Sony. That isn't to say the claim that "Navi was designed for Sony" is false. But it is correct to reject anyone's (Jason's) claim of it until evidence is reviewed and brought forward for discussion and debate.
 
It was probably more business related than technology related with AMD just prioritising their biggest semi-custom partner. I can't imagine AMD's desktop/laptop GPU sales are greater than PS4/XBO sales....

Also around this time the Switch was in development/launching. Seeing a console maker going with Nvidia may have concerned the AMD executives.
 
That isn't to say the claim that "Navi was designed for Sony" is false. But it is correct to reject anyone's (Jason's) claim of it until evidence is reviewed and brought forward for discussion and debate.
And discussion will focus on most probable cases, including common-sense reasoning, and an acute awareness that an expressed interpretation could very well be wide of the mark. It's not difficult to get to "Navi was designed for Sony" from "Sony is involved in Navi development and has ensured features for their purposes", but it is hard to get from "Navi was designed for Sony" to "AMD based its entire next-gen hardware on a bespoke console part basis funded by Sony to be used by everyone else."

An all-round awareness should see everyone preferring to side with the former possibility, and certainly no-one should be believing the latter without more evidence beyond an unqualified remark, no matter who the source is (well, unless one of those sources is a chief exec at AMD or Sony. ;))
 
In essence, IBM didn't need to create a new from-scratch CPU for MS because they already had the work done in creating the PPU. What MS got out of Sony's investment wasn't Sony's investment - they didn't get Cell. MS didn't look at Cell and say, "Hey, IBM, can we have a triple-PPU processor, please." What MS got was an IBM Power processor, and what IBM got was a nice deal because they didn't have to spend a lot to create this processor as they'd already done the ground work in researching a small Power processor for use as the PPU.

There's a bit of an assumption here. You're assuming Sony and Toshiba didn't subsidize the development of PPE. That would imply the PPE was something that IBM was going to develop regardless because it would somehow fit their core offerings, yet it only saw the light of day in Cell and Xenon.
IBM would take upon themselves the costs of developing the PPE if they were to become the technology taker role for themselves and sell it for their own HPC market, which never happened. No R&D company develops a new technology for the sake of technology push alone.


The Cell comparison would be Sony investing in Cell for use in PS3 and then MS and Apple and PC getting it from IBM without any of the R&D costs.
Let's assume Cell as being PPE + SPEs + Rambus MCU. Sony invested in Cell, not just the SPE and/or Rambus MC.
MS got the PPE for their console. Part of Cell went into the X360. Part of a product of investment of Sony went into Microsoft's console.
This doesn't mean Microsoft didn't have to pay a premium for the PPE (at least over what Sony would have to pay), or that Sony didn't get a substantial discount for being an early subsidizer of the PPE. We don't even know that Microsoft didn't have pay Sony royalties for using the PPE.

Now let's assume PS5's GPU as being RDNA1 architecture + RDNA2 instructions + Raytracing Engine + other-proprietary-IP.
If Sony subsidized PS5's GPU and Microsoft used RDNA1, then part of a product of early(/ier) investment of Sony would end up in Microsoft's console. I don't see why this seems so impossible.


You think Sony paid for that to get a console GPU only to let the rest of the world benefit?? Why not just take whatever AMD provides anyway at much reduced cost?
Technology gets developed everywhere, all the time and sooner or later most of the world benefits (or suffers) from it.
Sony subsidizing the early development of an architecture and their direct rival using it doesn't mean Sony is always getting the short stick of the deal.


Also, if Sony were entering such a major partnership, don't you think they'd need to tell their investors? "Yes, we're bankrolling AMD's R&D for the next few years to create our next console SOC. Yes, they'll be able to use the IP after us for other clients but we get 40% of all sales in return." If it comes out that Sony has financed their rival platforms - both MS and PC - without letting anyone know and without clear financial gains at the end, that's not going to go down well.
I don't think Sony needs to specify the full details of all the deals they make to their investors.


It was and continues to be the correct thing to do here at B3D.
I'll disagree that the correct thing to do at B3D is to dogpile on an article and its author if the ones doing the dogpiling even failed to read as little as the article's title.

The article's title reads, on its first line: AMD created Navi.
This thread's title? Sony Engineering the AMD GPU.

Most of the discussions generated around this article are purely driven by fake readings, fake news, fake interpretations, willful ignorance and probably a bit of dishonesty.

¯\_(ツ)_/¯


But more importantly what happened in the past is has no evidence of a relationship to what happened with Navi. We do not have any significant "data points" that suggest that Navi was designed for Sony. That isn't to say the claim that "Navi was designed for Sony" is false. But it is correct to reject anyone's (Jason's) claim of it until evidence is reviewed and brought forward for discussion and debate.
I completely agree with all of the above. Other than Jason Evangelho's background, there isn't any other data point to support that article's claims. There's also the fact that he might have misinterpreted what was told to him.
But none of it automatically means what he wrote is "dumb" or "idiotic" or many other adjectives that have been used.
 
I'll disagree that the correct thing to do at B3D is to dogpile on an article and its author if the ones doing the dogpiling even failed to read as little as the article's title.

The article's title reads, on its first line: AMD created Navi.
This thread's title? Sony Engineering the AMD GPU.

Most of the discussions generated around this article are purely driven by fake readings, fake news, fake interpretations, willful ignorance and probably a bit of dishonesty.

¯\_(ツ)_/¯
I get where there could have been more maturity.

Looking from the other angle, for those that did read the article thoroughly, it didn't come across as a typical article; it felt definitely less like reporting on something, and more like inserting his own conjecture. For the sake of it, his final paragraph:

The other interesting aspect to all of this is that my sources never mentioned Microsoft in the Navi conversations. This is pure speculation, but maybe Microsoft's next Xbox devices -- code-named "Scarlett" -- won't use Navi at all. Perhaps it will use a separate semi-custom solution incorporating Vega, or something else entirely that we're not privy to. Either way, the conversations I had referred to Navi in the past tense, as if it was already finished.

Perhaps Sony is closer to a PS5 than Microsoft is to a next-generation Xbox?

Or the issue that @Shifty Geezer had with it
Pull on that thread and it sheds more light on Vega's troubled development, due to the majority of Koduri's engineering team being swiped away not just to work on Navi, but to work on Navi for Sony.

On a related note, a new rumor emerged recently about Navi for desktop being merely a midrange part and not competing with Nvidia's high-end GeForce cards. That makes perfect sense if it was developed primarily for a console first.

or this piece here
Taken as a solitary piece of news, this may sound borderline ridiculous. But when zooming out to a wider angle -- as I did today analyzing AMD's history of consumer graphics products and its semi-custom business -- it fits like a glove. The same sources who spoke to me about the PS5 had more to say about AMD's business motivations with Apple, Microsoft and Sony in the link below.

Generally speaking it's the type of article I would expect from a Sony based news site than Forbes. Even WindowsCentral (as completely focused MS news venue) I don't think lately would have published an article with this much of their own conjecture.
 
I love cooking analogy. As I see it, Sony's and MS coming to burger joint (AMD) and orders burger (APU) from the menu. They can add some cheese and jalapeno (chefs ok with that - he's got all kinds of ingridients for the customers to play with), get rid of onion or ketchup and order the meat rare or well done. That doesn't mean that they made the burger. Even if they add some (secret) bbq sauce on top of everything (which is unlikely).
 
I'll just petition the mods here:
Can we get a thread title change to
(2019) Forbes: AMD Created Navi for Sony, Vega Suffered *spawn*
and a link to the article in the first post?
https://www.forbes.com/sites/jasone...nys-playstation-5-vega-suffered/#148088ce24fd

I get multiple points of view here on the topic, in particular the annoyance that some members may feel on the topic, but in defense of the counter argument to not bias/slant forum members (and their accompanying posts) before they even read have a chance to read article; we can accomplish this task of keeping things neutral and mature by lifting Forbes article name and replacing this thread title with it.

It is unlikely we will see real data points that Sony engineered Navi; so I think it's likely going to be less heated this way.
 
Sony Sabotaged GPU scheduled for Microsoft Xbox. :runaway:

Good new thread title?

Anyways, changed thread title yet again. Will see what can be done about first post.
 
I'll just petition the mods here:
Can we get a thread title change to
(2019) Forbes: AMD Created Navi for Sony, Vega Suffered *spawn*
and a link to the article in the first post?
A good counter-argument to mine comes from this linked article in the Forbes one:
https://www.forbes.com/sites/jasone...-amds-graphics-identity-problem/#7b8d88dd32e2

It suggests the AMD doesn't design a GPUs for general graphics etc. and then provides a flavour for specific clients, but actually makes GPUs for clients like Sony and Apple and then sells these as GPUs. That is, the semi-custom concept is actually bespoke hardware made-to-measure that's then mass-produced for the PC space regardless whether it's a good fit or not.

I find that hard to swallow. ;) It's a silly way to be competitive with nVidia.
 
A good counter-argument to mine comes from this linked article in the Forbes one:
https://www.forbes.com/sites/jasone...-amds-graphics-identity-problem/#7b8d88dd32e2

It suggests the AMD doesn't design a GPUs for general graphics etc. and then provides a flavour for specific clients, but actually makes GPUs for clients like Sony and Apple and then sells these as GPUs. That is, the semi-custom concept is actually bespoke hardware made-to-measure that's then mass-produced for the PC space regardless whether it's a good fit or not.

I find that hard to swallow. ;) It's a silly way to be competitive with nVidia.
Yea, or the obvious other choice is that AMD is generating generic designs that would best fit their semi-custom client profiles - in which they could then take and customize for their own needs.
The article nearly insinuates that the Polaris 400 is a design based off Xbox One X because they released around the same time. When it clearly performs at a 580 level.
And that also doesn't explain how Sony was using Polaris and funneling Vega features down. So what were Vega and Polaris then?

I get that it's possible (Entirely) that specific GPUs are based off their semi-custom clients. Like how Bonnaire seemed pretty much like a XBO. But i have a hard time with seeing entire architectures designed in this way.

Generations only come once every 6-7 years. AMD has to have a real strategy in between those time periods.

Where would AMD be if Sony and MS cancelled their Navi contracts. Would AMD cease to have a product to deliver? Seems awfully too dependent for this to be the way Jason writes it to be.
 
Last edited:
It suggests the AMD doesn't design a GPUs for general graphics etc. and then provides a flavour for specific clients, but actually makes GPUs for clients like Sony and Apple and then sells these as GPUs. That is, the semi-custom concept is actually bespoke hardware made-to-measure that's then mass-produced for the PC space regardless whether it's a good fit or not.
The problem is: it kinda fits the timeline, AMD developed and then released GCN, two years before XO and PS4 release, then stuck with it for the whole PS4/XO life cycle, refusing to dramatically change it or break away from the formula, until almost two years before the PS5/XSX release, when they developed RDNA which -coincidentally- is also the basis of PS5/XSX.

Before that AMD had a lot more liberty to change architectures, they went from VLIW5 to VLIW4 to GCN in the span of just 4 years (from 2007 to 2011), but then they stuck with GCN from 2012 to 2019, which is a frigging 8 years period!

It seems that at the very least, AMD stuck themselves into a lockstep rhythm with the console release cycle, only changing architecture when the console cycle is about to end, and seems RDNA is heading toward the same fate too.
 
Back
Top