There´s some realm called reality, and it that realm xbox series has better architecture than ps5. There´s other realm called hive mind of 40 year old playstation fanatics, destined to be crushed by reality.
-
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.
PS5 Die Shot has been revealed
- Thread starter LivingD3AD
- Start date
There´s some realm called reality, and it that realm xbox series has better architecture than ps5. There´s other realm called hive mind of 40 year old playstation fanatics, destined to be crushed by reality.
John Wick
Member
So explain to me the architecture in between RDNA1 and 2? Did you also read the rest of his quotes? Or are you just gonna quote the one that fits your warrior agenda?
He also mentioned PS5 has more features than RDNA2 but minus one. Can you explain that?
John Wick
Member
Is that why the PS sytems outsell Xbox?
The PS5 will outsell the SX and XSS. Wake up to that reality
Garani
Member
itsjustJEFF
Member
Based on what I'm seeing, and my extensive, EXTENSIVE, knowledge and experience of looking at die shots... I believe I see some RDNA3 in there, I think I also see so me UE5 enhancements. Pack it up boys, this die shot is the final nail in the coffin...
Die shots LOL
Die shots LOL
Last edited:
SenjutsuSage
Banned
PS5 is RDNA 2 Custom utilizing features Sony wanted/were able to get in the time window available they felt allowed them to launch in time. Clearly Ray Tracing was ready sooner, so that's in. I'm sure there's other customizations specific to third party engines and Sony first party needs, and those will birth all kinds of tricks and innovations. For example, VRS in software worked better on Playstation than Xbox One X for a reason on last gen.
Remember how Sony waited longer on PS4 to get 8GB of GDDR5? Microsoft more or less confirmed they waited longer to get a full RDNA 2 GPU from AMD, and then they went further by tossing in machine learning acceleration hardware.
Imtjnotu
Member
Gaming Discussion
Video game news, industry analysis, sales figures, deals, impressions, reviews, and discussions of everything in the medium, covering all platforms, genres, and territories.
Last edited:
thicc_girls_are_teh_best
Member
Agreed. There needs to be some form of glue / Infinity Fabric. After taking that into account, the area available for substantial amounts of LLC for the GPU would be almost null.
The space between the GDDR6 PHYs seems thicker than usual, but it could be just Infinity Fabric that AMD seems to put "wherever there's some space left".
As for the space between the CPU and GPU, I wonder if there's enough space for AMD/Sony to have implemented some sort of ring bus between the CCXs and that would justify the performance advantage at high framerates.
If they did and it has a similar behavior to what Intel has on their Low-Core-Count server CPUs (and most recently Comet Lake 8-core client), then it wouldn't be as good as Zen3's truly unified cache but it would be substantially better than Zen2.
Here are the access times on Zen2 with separate L3 between CCXs:
On Zen3 with unified L3:
On the 8-core Comet Lake with separate L3 on each core, connected by a ring bus:
This isn't a change that is very important if the goal is to make games that run great at 30 to 60FPS, but it makes a substantial difference when/if the CPU becomes a bottleneck at >100 FPS.
Microsoft wouldn't be very interested in this optimization but Sony could be, due to PSVR2.
It would also give some credence to the leaksters claiming "unified L3" on the PS5 CPU, since developers could assume the L3 is unified after looking at Comet Lake-like access times.
That's a good point-out, hadn't even thought about the L3$ for CPU in this manner. I'm looking at a Zen 3 diagram and comparing it with the PS5 Zen 2 diagram based on the x-rays; I don't think Sony would be able to have a ringbus around both CCX clusters to simulate a unified L3$, but they could maybe have a buffer between the two 2 MB L3$ clusters on a given CCX...though I'm not sure if that's how it already works with Zen 2 or not (I'm assuming it works that way by default).
So, to explain away performance differences that may be favoring PS5 in most instances atm, I'd chalk it up to either the faster GPU clock speed, lack of issues to sort out due to not having segmented memory spaces, and/or Series X's CPU not being fast enough for keeping the GPU fed with instructions in some sort of way. If there's anything outside of that which could deal with customizations by Sony on the CPU, it would probably be some type of alteration on say maybe the ALUs of the CPU, or some change to the FPU that were able to add in by sacrificing support for lower-latency FADD units. That would be about the extent of it IMO.
Just compare the setup to Navi 21 (also a 40 CU GPU); on Navi 21 you can clearly see IFCC runs along the other parts of the GPU and a central link on both sides connects to both L2$ pools. The L3$ (IC) itself runs parallel around the IFCC; that's the only way you can get the bandwidth, by having the cache (which I'd assume is smaller blocks of cache set parallel around the IFCC, not too dissimilar from the way the L2$ is built from smaller blocks of 256 KB clustered together) each getting a parallel interface along the IFCC.
PS5, from the shots I've seen of the GPU, has no IFCC; even assuming IFCC were in the unified memory controller and whatever sliver of L3$ they could fit between the PHYs connected through it, that would still be a very narrow connection and we already know from the way IFCC and L3$ are arranged on Navi 21 (and surely the other RDNA 2 cards like 6900X) that it's a wide, maybe almost ringbus-style parallel connection for many smaller cache blocks interfaced in parallel. So trying to apply that to PS5, would be small chunks of cache blocks sequentially connected together so only the block immediately interfaced to the controller (with assumed IFCC built in) would actually have a direct link. In other words, data in the cache blocks further down would have to travel sequentially, and there's a latency penalty associated with that.
The wide space you are referring to, is probably just general Infinity Fabric to let the CPU and GPU communicate with each other. If you look at the Series X APU schematic, in fact, you will see something similar:
There's similar space between Series X's CPU cores and the GPU, and between the GDDR6 PHYs, yet we already know that system doesn't have IC because, again, there's no presence of an IFCC or pipeline to feed hypothetical IC through to GPU L2$. So we can more or less ascertain that it's exactly similar with Sony's design, otherwise we'd see something more mimicking something similar to 6800 and how its GPU is arranged for IFCC and IC.
K Kerlurk I wouldn't necessarily use "blast processing" as an analogy here; MegaDrive actually did have hardware support for direct DMA framebuffer access (something that wasn't standardized with consoles until the Atari Jaguar and then the 5th-gen systems; 3DO might've had it but I'm not sure). The problem was always in figuring the timings, particularly for commercial software, without screwing up game logic.
Hackers and demoscene programmers have managed to figure out the timings though, and there's a lot of demos online showing MegaDrive using "blast processing" in practice; it's actually quite awesome. Wish more commercial games did it though.
Last edited:
alabtrosMyster
Banned
ethomaz
Banned
You know silicon and driver analysis showed Series X is a mix of GCN, RDNA, RDNA2 and new things?
Full thread:
Specific tweets:
Basically the same for PS5... a mix.
Last edited:
FranXico
Member
Which delusional reality are we in? How did you conjure those falsehoods into tweets? BS all of it!
alabtrosMyster
Banned
An architecture is not "better" until it has proven its merit compared to another.
SenjutsuSage
Banned
The reason why I'm confident there's always more to these systems than meets the eye even if it appears features aren't present is due to what I remember reading about Xbox One X and all the existing game and engine (even for upcoming titles) simulation tests Microsoft ran to find and fix the various bottlenecks. Cerny talks to a shit ton of developers, as I'm certain Microsoft does also. These developers make requests, tell them what specifically they need or want at a deeper system architecture level. I'm confident PS5 is decked out head to toe to with all kinds of tricks. Tricks already known, tricks that will eventually be invented due to the intricacies of the hardware.
In built hardware features just help make things more straightforward, but you will have times when developers just find their own method or may not be satisfied with what the hardware is giving them, and then innovation is born. It has happened every single generation on major console, especially Playstation. The die shot here is fun for us to discuss and dissect, but none of us really know what the fuck is going on here, though we try our level best lol. The PS5 could have features we aren't aware of, and I extend that belief to the Series X side of things also, but often times they may not have a name for them yet.
Prior to the PS4 gen we had no idea the Killzone developer had pulled a trick on everybody where they checkerboarded resolution to get to their target. That shit caught everybody off guard. I believe they were the first to do it to my knowledge, and then it became like this big industry standard and I think Cerny even had tweaks and changes made to PS4 Pro to better serve the technique, which we should expect PS5 likely takes to another level in some fashion. Look at what the Metro dev said. There are techniques that will be found based on the flexibility of AMD's ray tracing hardware that haven't even been discovered yet. This will happen on Playstation 5 as well as on Xbox Series X, so at least for me when I discuss these things and talk about what feature one has that another doesn't, I'm not foolish enough to think that any of these things mean anything if the games aren't backing it up.
In built hardware features just help make things more straightforward, but you will have times when developers just find their own method or may not be satisfied with what the hardware is giving them, and then innovation is born. It has happened every single generation on major console, especially Playstation. The die shot here is fun for us to discuss and dissect, but none of us really know what the fuck is going on here, though we try our level best lol. The PS5 could have features we aren't aware of, and I extend that belief to the Series X side of things also, but often times they may not have a name for them yet.
Prior to the PS4 gen we had no idea the Killzone developer had pulled a trick on everybody where they checkerboarded resolution to get to their target. That shit caught everybody off guard. I believe they were the first to do it to my knowledge, and then it became like this big industry standard and I think Cerny even had tweaks and changes made to PS4 Pro to better serve the technique, which we should expect PS5 likely takes to another level in some fashion. Look at what the Metro dev said. There are techniques that will be found based on the flexibility of AMD's ray tracing hardware that haven't even been discovered yet. This will happen on Playstation 5 as well as on Xbox Series X, so at least for me when I discuss these things and talk about what feature one has that another doesn't, I'm not foolish enough to think that any of these things mean anything if the games aren't backing it up.
kyliethicc
Member
just a quick note - Navi 21 is 80 CUs, not 40.That's a good point-out, hadn't even thought about the L3$ for CPU in this manner. I'm looking at a Zen 3 diagram and comparing it with the PS5 Zen 2 diagram based on the x-rays; I don't think Sony would be able to have a ringbus around both CCX clusters to simulate a unified L3$, but they could maybe have a buffer between the two 2 MB L3$ clusters on a given CCX...though I'm not sure if that's how it already works with Zen 2 or not (I'm assuming it works that way by default).
So, to explain away performance differences that may be favoring PS5 in most instances atm, I'd chalk it up to either the faster GPU clock speed, lack of issues to sort out due to not having segmented memory spaces, and/or Series X's CPU not being fast enough for keeping the GPU fed with instructions in some sort of way. If there's anything outside of that which could deal with customizations by Sony on the CPU, it would probably be some type of alteration on say maybe the ALUs of the CPU, or some change to the FPU that were able to add in by sacrificing support for lower-latency FADD units. That would be about the extent of it IMO.
Just compare the setup to Navi 21 (also a 40 CU GPU); on Navi 21 you can clearly see IFCC runs along the other parts of the GPU and a central link on both sides connects to both L2$ pools. The L3$ (IC) itself runs parallel around the IFCC; that's the only way you can get the bandwidth, by having the cache (which I'd assume is smaller blocks of cache set parallel around the IFCC, not too dissimilar from the way the L2$ is built from smaller blocks of 256 KB clustered together) each getting a parallel interface along the IFCC.
PS5, from the shots I've seen of the GPU, has no IFCC; even assuming IFCC were in the unified memory controller and whatever sliver of L3$ they could fit between the PHYs connected through it, that would still be a very narrow connection and we already know from the way IFCC and L3$ are arranged on Navi 21 (and surely the other RDNA 2 cards like 6900X) that it's a wide, maybe almost ringbus-style parallel connection for many smaller cache blocks interfaced in parallel. So trying to apply that to PS5, would be small chunks of cache blocks sequentially connected together so only the block immediately interfaced to the controller (with assumed IFCC built in) would actually have a direct link. In other words, data in the cache blocks further down would have to travel sequentially, and there's a latency penalty associated with that.
The wide space you are referring to, is probably just general Infinity Fabric to let the CPU and GPU communicate with each other. If you look at the Series X APU schematic, in fact, you will see something similar:
There's similar space between Series X's CPU cores and the GPU, and between the GDDR6 PHYs, yet we already know that system doesn't have IC because, again, there's no presence of an IFCC or pipeline to feed hypothetical IC through to GPU L2$. So we can more or less ascertain that it's exactly similar with Sony's design, otherwise we'd see something more mimicking something similar to 6800 and how its GPU is arranged for IFCC and IC.
K Kerlurk I wouldn't necessarily use "blast processing" as an analogy here; MegaDrive actually did have hardware support for direct DMA framebuffer access (something that wasn't standardized with consoles until the Atari Jaguar and then the 5th-gen systems; 3DO might've had it but I'm not sure). The problem was always in figuring the timings, particularly for commercial software, without screwing up game logic.
Hackers and demoscene programmers have managed to figure out the timings though, and there's a lot of demos online showing MegaDrive using "blast processing" in practice; it's actually quite awesome. Wish more commercial games did it though.
Navi 22 is gonna be 40, that'll ship sometime this year. And of course Navi10 was 40 CUs.
And ya that area around the GPU and memory interfaces on both the XSX and PS5 SoCs are just connective "fabric" used for linking everything. Not L3 cache.
Last edited:
FranXico
Member
So the FPU customization still is an unknown at the end of the day?
I think it's better to give it a rest for now.
assurdum
Banned
Damage control for what?Oof lol
Sony fanboys are gonna try to damage control, Reeeeee doesnt even have a thread discussing it lmao
prinz_valium
Member
I would call Xbox Series RDNA 1.8 and Ps5 RDNA 1.6
Guess that's fair.
Custom chips are custom.
Guess that's fair.
Custom chips are custom.
assurdum
Banned
You already know there isn't anything of special inside the series X hardware unachievable for the ps5. A developers has spent much of his time to explain to you how mesh shaders or VRS work in RDNA2 architecture. But nope. Still we use the MS propaganda as evidence of something around the full RDNA2 buzz word. I really don't understand why persist in such useless argumentations.
Last edited:
thicc_girls_are_teh_best
Member
Watching MLID's take on the SOC and...damn....it is kinda hilarious ngl. Bodies are under the bus, don't drive too fast
I get it tho, I'd be mad too if I felt people were misquoting me left, right and center. But that kind of comes with the territory if you're a tech channel that appears to have some reputability. Still though, this segment is quite funny considering everything that's happened x3.
I get it tho, I'd be mad too if I felt people were misquoting me left, right and center. But that kind of comes with the territory if you're a tech channel that appears to have some reputability. Still though, this segment is quite funny considering everything that's happened x3.
SenjutsuSage
Banned
I'm going with the facts as they've been provided by Microsoft, AMD and Sony. You don't like it, go take it up with Sony and AMD for not challenging any of the assertions confidently made by Microsoft on Series X's featureset compared to PS5, or go ask Sony directly about not answering questions posed multiple times by Digital Foundry. Propaganda you say. Go talk to redgamingtech and moores law is dead about propaganda. One is even now saying he never said any such thing even as there's a video showing him saying it.
SenjutsuSage
Banned
Watching MLID's take on the SOC and...damn....it is kinda hilarious ngl. Bodies are under the bus, don't drive too fast
I get it tho, I'd be mad too if I felt people were misquoting me left, right and center. But that kind of comes with the territory if you're a tech channel that appears to have some reputability. Still though, this segment is quite funny considering everything that's happened x3.
He wasn't misquoted. He lied. The man said I wouldn't be bragging about being RDNA 2 if I were Microsoft because Sony is RDNA 3. He said it with confidence like he knew it was. The other statement he keeps pushing as full context is still hot garbage because he again says Microsoft is lying about being full RDNA 2. They've met every criteria to prove that they are by having 100% of the GPU graphics featureset as laid out by AMD at their RX 6000 reveal event. Infinity Cache is something separate from the core graphics featureset introduced in RDNA 2. Claiming Series X isn't full RDNA 2 for not having Infinity Cache is akin to disqualifying a console part for not having RAM as fast or power draw equaling the desktop parts. Infinity Cache was an addition to avoid having to spend money on more expensive, faster and more power hungry RAM, or avoid going with a wider mem bus. Due to the clever nature of the addition, and where it's located they got expected IPC gains from it also.
John Wick
Member
Some very good points. The Cerny/Sony patents filed point to the PS5 having some of these features. I think Sony's bet on SSD and IO is gonna pay off. The PS5 is going to be an efficiency machine. Just that UE5 demo is an eye opener of whats possible.
Garani
Member
Exactly. But we are still here wasting time on the RDNA1 FUD. Like people actually built the damn chip
John Wick
Member
I don't know about MLID but RGT has been fair on both consoles. He's never once stated it as fact. But as speculation from his contacts friends etc.
SenjutsuSage
Banned
I remember, even I've made the SSD jokes, but I KNOW there will be clever visual tricks that will be accomplished through the combination of the PS5 hardware and high end use of the PS5 decompression unit/SSD. So while not a graphics card per se, it will help games look better in a lot of ways. I fully expect as much. What we're doing here is only "interesting" because we have nothing else to talk about lol. We would sooner love to see some badass gameplay footage drop from one of these consoles major exclusives, and instead talk about that, but this is what we have. A die shot.
SenjutsuSage
Banned
In fairness, I don't really watch them much, but a large amount of discussion I've heard appears to be centered around the belief that they were pushing that stuff fairly hard. But since I don't really watch them much, I can't really confirm or deny. But I've seen some video clips that are.. well rough.
John Wick
Member
It's like us discussing the die shot. We are all speculating. Now RGT could just say I don't know and wait till Sony or AMD confirm it which would be boring. So they ask around, research etc and speculate the reason why such and such occured. It's when people take one thing out of context and make it into a headline.
Loxus
Member
While still looking for RDNA 2, I found RDNA 1.
New infrared (NIR) die-shot of Navi 10
Processor Diagram - Navi 10
PS5 looks to be laid out the same,
while XBSX looks like half Navi 10 with the two CU clusters laid out next to each other the same way as PS5 Zen 2 Core clusters.
New infrared (NIR) die-shot of Navi 10
Processor Diagram - Navi 10
PS5 looks to be laid out the same,
while XBSX looks like half Navi 10 with the two CU clusters laid out next to each other the same way as PS5 Zen 2 Core clusters.
SenjutsuSage
Banned
I suppose I understand speculating and talking to people you think are giving you good info and doing your own research to arrive at possible conclusions as to what a piece of hardware may be doing to produce content people will want to see, but my take sometimes is that they may go much further than that, knowingly pushing information they know isn't the case for views. But alas, won't knock someone for making their money and doing their hobby they enjoy.
demigod
Member
Yo
Mod of War
EviLore
how is this dude not banned yet for ban evasion? I know who you are, New York’s Finest.
DaGwaphics
Member
Watching MLID's take on the SOC and...damn....it is kinda hilarious ngl. Bodies are under the bus, don't drive too fast
I get it tho, I'd be mad too if I felt people were misquoting me left, right and center. But that kind of comes with the territory if you're a tech channel that appears to have some reputability. Still though, this segment is quite funny considering everything that's happened x3.
Save that ass boy. LOL
All of these channels will be singing the same tune over the next week, it is what it is.
S0ULZB0URNE
Member
PS5 RDNA 2.2
XSX RDNA 1.8
Kholinar
Banned
Because they're totally intimidated by forum nobodies holding them accountable for others taking their mere speculation as gospel.
DaGwaphics
Member
More covering their six because the ego won't allow them to ever be wrong. But I guess it's up to personal interpretation.
Makes for good entertainment though, regardless.
bentanchorbolt
Banned
This thread is ridiculous.
Kholinar
Banned
No, I'm serious. You do realize these techtubers pull tens to sometimes hundreds of thousands of views with most I've seen circulated around here averaging from about 50-100k subs?More covering their six because the ego won't allow them to ever be wrong. But I guess it's up to personal interpretation.
Makes for good entertainment though, regardless.
Do you actually think they're pressured or 'covering' from comparative nobodies like you? I don't think MLID even comes on to this site.
FatKingBallman
Banned
This should be interesting to you then
Locuza :
Timestamped 5:08
I think you have way better knowledge than Locuza, so, destroy him. You can do it.
Last edited:
Crimson_Fate
Member
I remember MLID saying the RDNA 3 but it was in the context of being sarcastic since people were saying it was RDNA1. He mentioned that he PS5 had RDNA 3 features and so far as I know that has not been disproven yet. So there is nothing for him to cover.More covering their six because the ego won't allow them to ever be wrong. But I guess it's up to personal interpretation.
Makes for good entertainment though, regardless.
DaGwaphics
Member
Of course they don't care about me per se, I'm sure they don't know I exist. But, as with any type of celebrity (which these YT types kind of are in a limited way) they automatically damage control when something doesn't go as they expected. Channels like MLID primarily get those clicks because viewers assume they have the inside track, compromising that illusion could cost them real $$$. That's why they are so defensive in response to this. It's why I gravitate more towards someone like a Spawnwave type that isn't trying to build a reputation on having secret sources, but is just someone that follows the industry and puts some ideas in one place.
I'll admit the RGT guy was a bigger offender, but the MLID crew leaned on some inaccuracies a lot harder than they are letting on now.
Last edited:
FranXico
Member
Very little one could tell from die shot.
If anything were to used on AMD's roadmap, it would be cache scrubbers or anything special about GE that we're not aware of.
Not likely, and even so, it would not be a major feature either.
Knitted Knight
Banned
We back on the RDNA 3 train? I remember...
It becomes major when AMD starts to market it with fancy marketing buzzwords, and beefs it up x2 on a PC card. It's all make believe salesmanship to sell tech advancements to an enthusiast crowd for a short time span.
Last edited:
Demon2blades
Member
Keep In mind Microsoft also
"confidently" branded Xbox series X the most powerful console ever awhile back but then changed it to most powerful Xbox ever for reasons unknown ... The words of billion/trillion dollar company's around advertising means nothing there's always a play on words somewhere
Crimson_Fate
Member
Just remember.. if anyone ever asks which console has the most power (PS5 or XSX)
It's clearly the PS5.
PS5=350W
XSX=300W
Its all word salad.
DForce
NaughtyDog Defense Force
You guys took his words out of context.
He compared Infinity cache to how cache is handled on the PlayStation 5.
Time stamped.
He also clarified his statement on what he meant by RDNA 3.
How often do you call out Colt, Dealer and Dee Batch for doing this exact same thing?
Mark Cerny did clearly stated that PS5 will have HW RT and you were just about calling him a liar and got yourself banned from Resetera for doing so. Just because Sony doesn't flat out and say which features are in the PS5, that doesn't mean people can automatically write off the features just because MS said it has full RDNA 2. We know MS has a history of using these marketing tactics because the Xbox Fanboys across social media eats up these buzzwords.
Last edited:
alabtrosMyster
Banned
Even if the ps5's GPU was based on ATI 3D rage tech from the 90s, it performs astonishingly well... They made the right choices (so did MS, MS just oversold their goods).Exactly. But we are still here wasting time on the RDNA1 FUD. Like people actually built the damn chip
So on and so forth.
I hope for the likes of those who posted that kind of stuff that xbox get their
in order soon.
Last edited:
Garani
Member
While still looking for RDNA 2, I found RDNA 1.
New infrared (NIR) die-shot of Navi 10
Processor Diagram - Navi 10
PS5 looks to be laid out the same,
while XBSX looks like half Navi 10 with the two CU clusters laid out next to each other the same way as PS5 Zen 2 Core clusters.
1) we do have photos of RDNA2
2) PS5's APU is damn near carbon copy of the RDNA2 Big Navi layout, minus the Infinity Cache
3) XSX APU's layout is practically the same as the one used by the Xbox One (yeah, the first one, the always connected thing that got trolled to the moon by Yoshida with the funny "how to share a game with your friend")
So what? They are both RDNA2 based and have a different layout for different priorities and latencies. You can't compare to other chip's layout.
