Right and Outsiders is a SE game made by People can Fly. So if true this is in Poland for a game and not next gen related.
The SOC's that AMD are building for Microsoft and Sony are *highly* customized specifically for the vendors based on specs, cost, and performance requirements that each platform defines for AMD. There is no off-the-shelf equivalent of anything that goes inside an Xbox or PlayStation, and the idea they are "pre-made" really misunderstands what's happening and sort of diminishes the role of Cerny or the architects at Microsoft in terms of how much custom work goes into the chip.
AMD has IP and Processes. Sony and Microsoft work with AMD's list of available IP and Processes, add in some of their own requirements, and have an entirely new SOC created specifically for them. This process takes years. Everyone is aware that the PS4 Pro GPU had unique HW instructions for temporal reconstruction, as well as support for FP16 that did not exist together in other part of the AMD portfolio. The ability to decode Xbox 360 textures and much of the native DX12 instructions in the Xbox SOC's only appear in the versions that ship on Xbox (the DX stuff I believe ended up in later AMD GPU's.) This is the 'secret sauce' that is so often talked about, and it absolutely can work for (or against) a specific console. I don't think Sony got much mileage out of FP16, and Xbox One's implementation of ESRAM didn't help them as much as expected either. While the Checkerboard techniques helped PS4 Pro a lot, and the ability to decode X360 textures is why you have such a good back compat story on Xbox.
Here's the best way I can think of to describe it. Think of AMD as a Caterer. They have a list of ingredients (e.g. Zen, RDNA, HW RT, etc. etc.) and they also have a set menu based on their ingredients. (RX5000 series, RX Vega series, RX 500 series). Most people think that Xbox and Sony choose from the Menu. But what actually happens is they choose from the ingredients, add in some of their own specific ingredients (DX instructions, Back Compat Code, Checkerboard, etc.) and have an entirely new dish created just for them.
My point is that even though AMD's R&D efforts provide the groundwork, the specific SOC found in PlayStation and Xbox consoles are completely bespoke designs in collaboration with AMD and the platform makers, and are designed years in advance. This is why anyone speculating that wholesale performance changes can happen at any time don't seem to realize how far in advance these decisions are made.
Yes small changes can happen throughout the process and plans evolve from the initial spec to the final production silicon. But you can't grow the performance of the chips this late in the game without restarting some pretty critical, long-pole parts of the process.
I'm only speaking from one side of this, but I'm pretty sure it's similar on the PlayStation side.
EDIT: Upon a re-read, I'm not giving enough credit to AMD's engineers either. It really is a collaboration between the companies. But the performance, cost and business aspects are defined by the consoles. The real silicon engineering is a collaboration.
Amazing post as always. Thank you for your insight. I hope people are paying attention. This makes the news that Sony started changing around the PS5 back in 2017 even more intriguing.
I don't understand how they get away with not reporting Xbox Live subscriber numbers. Wouldn't investors want to know that?
Oh I guess all the info we got was in advance of the conference itself
Oh. Came from work, so can't figure out what's joke and what's not lol
The SOC's that AMD are building for Microsoft and Sony are *highly* customized specifically for the vendors based on specs, cost, and performance requirements that each platform defines for AMD. There is no off-the-shelf equivalent of anything that goes inside an Xbox or PlayStation, and the idea they are "pre-made" really misunderstands what's happening and sort of diminishes the role of Cerny or the architects at Microsoft in terms of how much custom work goes into the chip.
AMD has IP and Processes. Sony and Microsoft work with AMD's list of available IP and Processes, add in some of their own requirements, and have an entirely new SOC created specifically for them. This process takes years. Everyone is aware that the PS4 Pro GPU had unique HW instructions for temporal reconstruction, as well as support for FP16 that did not exist together in other part of the AMD portfolio. The ability to decode Xbox 360 textures and much of the native DX12 instructions in the Xbox SOC's only appear in the versions that ship on Xbox (the DX stuff I believe ended up in later AMD GPU's.) This is the 'secret sauce' that is so often talked about, and it absolutely can work for (or against) a specific console. I don't think Sony got much mileage out of FP16, and Xbox One's implementation of ESRAM didn't help them as much as expected either. While the Checkerboard techniques helped PS4 Pro a lot, and the ability to decode X360 textures is why you have such a good back compat story on Xbox.
Here's the best way I can think of to describe it. Think of AMD as a Caterer. They have a list of ingredients (e.g. Zen, RDNA, HW RT, etc. etc.) and they also have a set menu based on their ingredients. (RX5000 series, RX Vega series, RX 500 series). Most people think that Xbox and Sony choose from the Menu. But what actually happens is they choose from the ingredients, add in some of their own specific ingredients (DX instructions, Back Compat Code, Checkerboard, etc.) and have an entirely new dish created just for them.
My point is that even though AMD's R&D efforts provide the groundwork, the specific SOC found in PlayStation and Xbox consoles are completely bespoke designs in collaboration with AMD and the platform makers, and are designed years in advance. This is why anyone speculating that wholesale performance changes can happen at any time don't seem to realize how far in advance these decisions are made.
Yes small changes can happen throughout the process and plans evolve from the initial spec to the final production silicon. But you can't grow the performance of the chips this late in the game without restarting some pretty critical, long-pole parts of the process.
I'm only speaking from one side of this, but I'm pretty sure it's similar on the PlayStation side.
EDIT: Upon a re-read, I'm not giving enough credit to AMD's engineers either. It really is a collaboration between the companies. But the performance, cost and business aspects are defined by the consoles. The real silicon engineering is a collaboration.
I don't understand how they get away with not reporting Xbox Live subscriber numbers. Wouldn't investors want to know that?
Oh. Came from work, so can't figure out what's joke and what's not lol
Right and Outsiders is a SE game made by People can Fly. So if true this is in Poland for a game and not next gen related.
"Get away with it"? Sounds like you want some ammunition for your console-war rhetoric.
Thanks! 2017 for changes to a 2020 launch are very plausible. Again, I know nothing but that timing would work.
That alone says enough. Thanks!Totally depends. Trying to be specific enough to make my point without disclosing anything I shouldn't. It's a collaboration is the best thing to say.
Here's the best way I can think of to describe it. Think of AMD as a Caterer. They have a list of ingredients (e.g. Zen, RDNA, HW RT, etc. etc.) and they also have a set menu based on their ingredients. (RX5000 series, RX Vega series, RX 500 series). Most people think that Xbox and Sony choose from the Menu. But what actually happens is they choose from the ingredients, add in some of their own specific ingredients (DX instructions, Back Compat Code, Checkerboard, etc.) and have an entirely new dish created just for them.
15tf is that much closer to realityThanks! 2017 for changes to a 2020 launch are very plausible. Again, I know nothing but that timing would work.
haha ... At this point if Sony even farts, it would be 20+ pages of meltdown and new threads.There would have been 20 pages of meltdown and a new thread if this were the case lol
No, I wasn't saying if any of the posts I quoted were insiders, just wondering where the predictions were coming from. You answered it, so thanks. Those that did answer, thanks.Oh goodness no. If I had insider information then like Klee I'd ask for a ban (personally, I lack self restraint). I drew my numbers from a few key (and some utterly bereft of context) rumours:
1. The supposed rumour about PS5 & XsX are relatively quite close in power.
2. The whole GitHub leak of some APU that may or may not have been some rendition of PS5's APU. Plus all that Komachi leak about frequency of base PS4, Pro and an unknown HW.
3. DF tested X's framerate performance against 5700XT which is rated at 9.75TF and was, on average, displaying twice the FPS. So for X, the only metric I have considered is TF figure and how it translates to real world performance. Of course, this post comes to mind.
4. All the probable GPU diagrams thus far are based on existing Navi chip blueprint which is bereft of RT HW. If, like nVidia, AMD are to feature dedicated RT cores then within a finite amount of space, something must give to make space for said HW. So presumably that is one reason why PS5's GPU, for its given area, cannot host more than 36CUs (I am assuming 40CUs total w/4CUs disabled for yield purposes). Of course, that's all speculative and AMD's HW driven RT can easily be something else. Personally, I'd love for Liabe Brave to chime in.
IIRC Oberon was clocked at 1.8GHz for Firestrike test and Komachi's tweet mentioned 2GHz and so between that 9.216TF is the ceiling with 20WGP w/2WGP disabled.
BGs gave me his permission to quote him from the old place, not he says anything new but I think we should let go the TFs measurement as both machines will be close. So the main focus for next-gen will be contents and it'll be good for everyone.
"BGs" said:I will not give any numerical data, and I will explain the reason.
As you will understand, it is not information I can give. Not even any developer with two front fingers will tell you, it would mean being fired automatically and facing possible legal consequences. If a developer gives you that information, it's not very smart. On the other hand, when I registered, I wanted to make it very clear in my presentation message that I did not come here to give information that could commit me or someone.
Likewise, it is true that it doesn't matter what I can tell you, I mean that I would not get you out of doubt, because the information I may have is current. That is, it is not the definitive one (which does not mean that there is no such definitive information, I simply do not know it).
What I can say is that you are making films of biblical proportions and based on data that, if true, were 1 year ago in a given circumstance. But do you really believe that real data is filtered? And I understand that this is because of the lack of information, and if there is no information, then people make it up based on personal interests. And you have to understand that many people are interested in generating interest and debate, for example by publishing something and deleting it at the moment. Guys, we already have an age to know how this works.
First, what TF? The TF of Marketing? Or the effective TF? I understand that people need to know (and I need to explain, but I have to bite my nails) but what you really should care to know is that, in practice, one could have more TF than the other and give less performance. You should think less about TF and think more about machine balance.
This generation will NOT be the generation of the visual difference that makes you decide what to buy. It will be the generation of the games. Yes, one will be faster than another (but you will not notice it in practice). One will load the screen in 10 seconds and the other in 7 seconds. One will have a better sound than the other. One will run the game at 60fps and the other ... too. What will really matter will be the quality that development studies can achieve with their time and budget.
What you will have to do this generation is to think what game you want and what platform it is on. And if it is for multiplatform games, buy the cheapest one or where you have more friends, because you will not notice the difference (yes, there are people who are able to perceive a laser pointer on the moon, and others like to play 30cm from the TV trying to find out if the 4K is native or scaled by switching quickly between different video inputs to find out the difference to justify their purchase, but those people are not happy. Fortunately, those people will have it very difficult in the next generation... in multiplatform games).
(And yes, PC graphics cards will always be better. And also much more expensive and less exploited. What happens is that NOT all games are on PC. For those who have forgotten, this is the key, the games).
The new consoles will not disappoint. Not me, at least.
I'm sorry, I don't have more time for now.
Cheers.
Unless I misheard, Satya said in the conference call that Xbox Series X would be the "fastest console ever".
That's a valid interpretation, but not what I got from it. They next few posts by BGs might be even more intriguing.I hope I haven't fucked up this quoting.
This is really interesting to me. I have no clue who BGs is, but based on what I've seen of the reaction here, I tend to trust that they may just be worth listening to.
Reading this is the first time that I'm thinking that an 8-9 TF PS5, in line with the GitHub leaks may actually be possible.
Just to be clear, I am entirely of the opinion that the actual real-world performance of the two machines will be basically indistinguishable, as this person says here, and as most insiders have been suggesting, we need to look past the raw TFs to see what these new consoles will offer us, and I'm not sure that any credible insider has suggested that there is a large gap between the machines.
I've posited before that maybe Sony have come up with enough secret sauce to make up for less flops, and this seems to go along with it.
I feel like there's a real possibility (not sure how big a possibility, but a possibility) that Sony and Microsoft have ended up with a similar level of performance, but that Sony are going to be managing it at a significantly lower BoM than Microsoft, which would be an interesting situation to say the least.
I've said before that I think it would be a mistake for Microsoft to launch at a higher price than the PS5 if they don't have a significant power advantage. What happens if their design costs them significantly more to build but doesn't give them a clear performance edge? Would they be willing to swallow a bigger loss to stay competitive? I do hope so if this is indeed the case.
The above is a particularly speculative bit of speculation, so do feel free to tell me I'm wrong.
What I can say is that you are making films of biblical proportions and based on data that, if true, were 1 year ago in a given circumstance.
And you think there is something wrong with this plan? Is it not the exact Anaconda/Lockhart playbook?Okay, I was not precise enough: So you suggest Sony ordered tape outs of several chip designs from AMD for testing until late 2019 to throw some of them away after eating the cost?
No it is not. Those are 2 chips for 2 SKUs.And you think there is something wrong with this plan? Is it not the exact Anaconda/Lockhart playbook?
That's a valid interpretation, but not what I got from it. They next few posts by BGs might be even more intriguing.
I got something different because of this:
correctThe channel MBG on YouTube isn't the most trustworthy, is he? Also Tidux, best to ignore his stuff. Correct?
Now? no. Eventually? Pretty much.
There is no practical difference between two SKU that sell 20M + 30M, and two SKU that sell 50M + 0M units. Said another way, MS strat potentially being 2 SKU is completely irrelevant - the fact is they are supposedly making 2 complete chip designs (and may even end up throwing one away and going with a 1 SKU strat!)No it is not. Those are 2 chips for 2 SKUs.
If you go further you will find some new insight given by Albert Penello .
I will link it for you:
Thanks! 2017 for changes to a 2020 launch are very plausible. Again, I know nothing but that timing would work.
I edited my post, after reading your post again...I agree with your interpretation, lol.Excellent point. I definitely read that but it didn't stick in my mind when creating that post.
It is a thing, but don't expect it this year.
"There's no reason for us to coincide it with a new console. From the point of view of the consumer, to be bombarded with many many things -- oh, you have to buy this, you have to buy that -- is a message that we don't want to send. In some ways, it's good to have a little breathing space between those things."
So you are still on the train of 2 chips developed in parallel? That is fine. Let us agree to have a different opinion and point of view here.There is no practical difference between two SKU that sell 20M + 30M, and two SKU that sell 50M + 0M units. Said another way, MS strat potentially being 2 SKU is completely irrelevant - the fact is they are supposedly making 2 complete chip designs (and may even end up throwing one away and going with a 1 SKU strat!)
Nonsense. Their action game mechanics have never been what they're known for, ranging from serviceable to borderline janky, while their character rendering and world-building are first-rate. Neither has anything to do with who owns them, and I hope they keep growing as a studio and honing their craft. If Hellblade 2 is as thematically remarkable as the first entry it just gives me another reason to own an Xbox even without being a top-tier action title. Isn't that the whole point?
That's an interesting question. It may be a SoC made for some other purpose or a simulation out of PC parts. Or maybe Sony did request its tape out and this is what we'll get in PS5 if they won't bite the cost of producing another SoC instead.Thank you. What do you think Oberon is though? Not a tape-out? Still simulation?
Sure, that's what is done after tape out and this is why you generally need anywhere between 3 and 12 months between the first chip coming back from the factory and it being actually ready for mass production.Some things can be fixed/updated with metal mask changes that don't touch the underlying bulk silicon. Those would lead to numbered iterations in the stepping.
This makes zero sense so I'm inclined not to believe that. You don't need the final SoC to build the console either.Consoles go into the tapeout stage long before a year before launch. According to Albert Panelo, a year before launch an almost final version of the console is already built on an assembly line.
Okay.
This is the point where you can produce several chips in parallel. It will cost more of course but it will cost significantly less than doing several tape-outs and also considerably less for a semi-custom part which is built out of parts and blocks which are already in production, in some other product(s).It takes probably a year or more of a locked down major configuration (number of cores, for example). Building blocks are being refined over time. The full chip design is being refined and assembled and timing paths are being slowly eliminated day after day.
Nobody is talking about "change".To have a major configuration change at any time is a major disruption and would result in a schedule slip. The longer you wait to make decisions like that, the worse the schedule delays get.
This is what I've said. A year from a tape out to launch. In case of AMD anyway.A taped-out chip then has 9 months to a year of test to make sure it doesn't have bugs. If it does have bugs, that means you have more steppings, and you try to get as many bug fixes into each stepping as you can.
Which is what? Putting more ACEs into the frontend? It was always a plan for future GCN evolution and putting more units hardly counts as "a huge role in the development of the chips". First GCN iteration didn't really need this simply because of the realities of s/w back when it launched.That's not entirely correct. Sony also plays a huge role in the development of the chips that go inside their consoles - so much so that the custom feature that Sony had worked on for the PS4's GPU actually got incorporated into the future Radeon discrete GPUs for PC in 2012-13.
By "parts" I mean execution units, cache sizes, number of blocks in the frontend and backend, etc., not actual PC parts available on the market.They don't just choose the parts. If they were to just choose the parts, the PS4 GPU would have ended up being exactly a Radeon HD 7870 with 2 CUs disabled with lower clock speed. But it's not.
Last time I've checked Cerny didn't have any skills to do any kind of GPU h/w customizations.The GPU was further customized/enhanced over the standard 7870 by Sony (by Sony - I mean Mark Cerny).
Sony has had a choice of what tech out of the currently available stack they wanted to see inside the chip which they've ordered from AMD. They didn't add any tech above what was already available in Polaris and Vega (ID buffer is a minor and mostly s/w change; should be pretty evident from the fact that CBR solutions are running nicely on any modern h/w, not just on PS4Pro), they didn't have any hand in developing said technologies either - apart from providing high level feature input along the lines of "we want to have h/w RT acceleration in there". People are severely overestimating Sony's and MS's involvement into the development of both these chips and the technologies which they are being built upon. This is all AMD, from zero to finish, and they take this high level input not only from Sony and MS but from their own product teams, from the developers, from competition's roadmaps and from a hundred of other sources in which Sony and MS aren't even a majority - in full accordance to the share of revenue and profits which AMD gets from these two semicustom clients.They did the same for the PS4 Pro. If Sony hadn't requested customization to the PS4 Pro's GPU, it would have ended up being exactly an RX 480 with lower clock speeds. But it's an RX 480 (Polaris) with a number of additional features from Vega architecture. This is because Sony has a hand in the development of these chips.
Thanks. Steppings don't change the chip configuration and rarely affect feature sets (and when they do it's mostly due to disabling stuff which doesn't work as intended).
"The most powerful" and "Our most powerful" is so different. Now I'm going to have to listen!
I am personally not at all sure, my favoured guess right now is along the lines of Liabe Brave, I think we haven't seen the full extent of an Oberon or whatever is the latest. My point is, it makes no sense to question the idea that Sony developed two chips - either in parallel or in sequence - and not also have a problem with MS doing the exact same thing.So you are still on the train of 2 chips developed in parallel? That is fine. Let us agree to have a different opinion here.
Correct btw, Lockhart and Scarlett SoCs would have to be developed in parallel.My point is, it makes no sense to question the idea that Sony developed two chips - either in parallel or in sequence - and not also have a problem with MS doing the exact same thing.
If you are wondering did he have a verified tag no he did notkleegamefan was officially verified, was he not? The links are in the OP.
The SOC's that AMD are building for Microsoft and Sony are *highly* customized specifically for the vendors based on specs, cost, and performance requirements that each platform defines for AMD. There is no off-the-shelf equivalent of anything that goes inside an Xbox or PlayStation, and the idea they are "pre-made" really misunderstands what's happening and sort of diminishes the role of Cerny or the architects at Microsoft in terms of how much custom work goes into the chip.
AMD has IP and Processes. Sony and Microsoft work with AMD's list of available IP and Processes, add in some of their own requirements, and have an entirely new SOC created specifically for them. This process takes years. Everyone is aware that the PS4 Pro GPU had unique HW instructions for temporal reconstruction, as well as support for FP16 that did not exist together in other part of the AMD portfolio. The ability to decode Xbox 360 textures and much of the native DX12 instructions in the Xbox SOC's only appear in the versions that ship on Xbox (the DX stuff I believe ended up in later AMD GPU's.) This is the 'secret sauce' that is so often talked about, and it absolutely can work for (or against) a specific console. I don't think Sony got much mileage out of FP16, and Xbox One's implementation of ESRAM didn't help them as much as expected either. While the Checkerboard techniques helped PS4 Pro a lot, and the ability to decode X360 textures is why you have such a good back compat story on Xbox.
Here's the best way I can think of to describe it. Think of AMD as a Caterer. They have a list of ingredients (e.g. Zen, RDNA, HW RT, etc. etc.) and they also have a set menu based on their ingredients. (RX5000 series, RX Vega series, RX 500 series). Most people think that Xbox and Sony choose from the Menu. But what actually happens is they choose from the ingredients, add in some of their own specific ingredients (DX instructions, Back Compat Code, Checkerboard, etc.) and have an entirely new dish created just for them.
My point is that even though AMD's R&D efforts provide the groundwork, the specific SOC found in PlayStation and Xbox consoles are completely bespoke designs in collaboration with AMD and the platform makers, and are designed years in advance. This is why anyone speculating that wholesale performance changes can happen at any time don't seem to realize how far in advance these decisions are made.
Yes small changes can happen throughout the process and plans evolve from the initial spec to the final production silicon. But you can't grow the performance of the chips this late in the game without restarting some pretty critical, long-pole parts of the process.
I'm only speaking from one side of this, but I'm pretty sure it's similar on the PlayStation side.
EDIT: Upon a re-read, I'm not giving enough credit to AMD's engineers either. It really is a collaboration between the companies. But the performance, cost and business aspects are defined by the consoles. The real silicon engineering is a collaboration.
2020 allowed them to use a full 7nm product stack with significantly less risk to production.See what gets me is that if a Sony was indeed looking at the "let's cheap out and sell PS5 for $399" strategy, what made them decide to switch in 2017 for a 2020 release? This is the part nobody can answer.
Lockhart could just be a die harvested Scarlett.Correct btw, Lockhart and Scarlett SoCs would have to be developed in parallel.