• 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

MonarchJT

Banned
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?
and you are again there to defend the impossible warrior.
i respect you lol
 
Last edited:
I dont get why every other post here are crying about missing infinitycache? Do you know what happens when any data that resides in that cache is invalid? The whole cache gets flushed, resulting in a latency "tank" when the data needs to be read from ram instead, affecting all code paths elsewhere as well. Having a bigger unified cache wont help you, it will just make it worse.

If only RDNA2 had cache scrubbers this scenario would be void, since only the outdated data gets scrubbed from the cache, leaving the other data intact ready to be consumed. Its a night and day difference.

This is such a clutch feature that AMD would be nuts not incorporating it into RDNA3.
I hope this post is satire.
 

HoofHearted

Member
If the "PS5 is not RDNA 2", neither is the Series X.

They just don't get it, it's hilarious.
how about...


They're both REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY....

REALLY REALLY REALLY...


.... REALLY close to RDNA2?
 

DaGwaphics

Member
how about...


They're both REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY REALLY....

REALLY REALLY REALLY...


.... REALLY close to RDNA2?

You saying you think PS5 is RDNA2 just because AMD said so....

season 1 starz GIF by Outlander


Okay, maybe it is, I guess those AMD folks might know a little about it, maybe...
 
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.
You know now that you bring it up, he did say that quote about MS 'bragging' of RDNA 2 in alluding to PS5 being RDNA 3 (or having features that'd be in RDNA 3) :LOL: . Just from me watching their content a lot, I get a tinge of smugness with him in the way he presents his content, where at least in the console side of things there's tonal usage and phrasing seemingly in favor of Sony and against favor of Microsoft, like changes in vocal inflections and the such. I think that's something they could improve on.

And just having finished watching the part of their latest podcast where they go into the PS5 SOC, there's still stubbornness from him in wanting to admit anything shown of the x-rays is even legit because it hasn't been officiated, when the truth is the window where Sony would've officiated much or any of this stuff has seemingly passed. And the reasoning he gave near the end for why he alluded to IC or unified L3$ for example, to explain the performance results between the consoles in 3P games, well honestly that was whimsical reaching. Figure people into following tech to the degree they do would not jump to fantasy customizations when more logical things based on what Sony and Microsoft already have told us officially (or could be concretely documented) would've served better even if less "exciting".

All that said tho it was worth watching just to see the proverbial bodies pile up under the bus and to MLID's credit there were/have been others pushing ahead with the crazier RDNA 3 custom tech angle, he even named some of them. He's still insisting that Sony've done something with the caches but at least they seem to have settled on that probably just being the cache scrubbers (which, in all likelihood, those could be a reason for some of the results we're seeing, just depends on how easily Sony have made the use of those scrubbers, what type of scrubbing they're using (I'm guessing patrol scrubbing would be easier for devs and more "automated"?), etc.).

Save that ass boy. LOL

All of these channels will be singing the same tune over the next week, it is what it is.

This one's been my favorite for the entertainment value but MGB's was the best for being genuinely apologetic for leading his audience on perhaps too strongly. And to MGB's credit he has not claimed to be an insider or connected to insiders in the industry.
 
Last edited:

HoofHearted

Member
You saying you think PS5 is RDNA2 just because AMD said so....

season 1 starz GIF by Outlander


Okay, maybe it is, I guess those AMD folks might know a little about it, maybe...
Perhaps.. they did invent the damn marketing term ... ;)

That being said - I'm more interested in the actual technical breakdown of the die shot more than marketing speak of "iz dis RDNA2?" ... give me the same level of detail on versions like locuza did in his XSX video ... THAT's something actually interesting to discuss..

....not whether or not Infinity Cache is included or not..
 

MonarchJT

Banned
Perhaps.. they did invent the damn marketing term ... ;)

That being said - I'm more interested in the actual technical breakdown of the die shot more than marketing speak of "iz dis RDNA2?" ... give me the same level of detail on versions like locuza did in his XSX video ... THAT's something actually interesting to discuss..

....not whether or not Infinity Cache is included or not..
I'm so damn waiting for it
 

Zannrebel

Member
Who lied?

And what exactly was the lie?

I ask this because remember ppl were asking the dev about that game.

Regardless, what does that have to do with the games that are PS5 exclusive? Because last time I checked, there are at least 2 we know of right now, one at launch and one yet to come out. Both showcased in July last year.

3, I forgot about Destruction All Stars. 4, forgot about Astro's Playroom
Sony's marketing said SM MM was exclusive to PS5 until the fall showcased they backtracked and said it was coming to PS5. Same with Horizon FW. Jim Ryan lied about that being PS5 exclusive before they back tracked 3 months later about. Disingenuous at best, lying at worse.
 

John Wick

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.
Bigger offender of what? Post on here what he said was hard confirmed fact.
 

jroc74

Phone reception is more important to me than human rights
Sony's marketing said SM MM was exclusive to PS5 until the fall showcased they backtracked and said it was coming to PS5. Same with Horizon FW. Jim Ryan lied about that being PS5 exclusive before they back tracked 3 months later about. Disingenuous at best, lying at worse.
Ok, I'll take your word on that.

Now, what about the 4 games I mentioned? Because it seems some ppl get hung up on Miles and HFW, and just dismiss the others that are PS5 exclusive.

Some ppl are acting like every PS5 game announced for PS5 only has a PS4 version. They dont. And that could still change. But for now.... 2 games dont equal a full blown lie about believing in generations.
 

DaGwaphics

Member
Bigger offender of what? Post on here what he said was hard confirmed fact.

Relax, it's a figure of speech in a video game forum not a legal dispute.

All these types include that disclaimer, but they still trade in the idea that the sources they have are accurate. Espousing the same ideas many times over is still misleading despite the disclaimer, especially when the ideas are really being sold. That's why the content creators dance when they need to in this context.

I feel bad for this thread, I've really leaned in on helping this be a trainwreck. My bad. Everything is off topic. LOL

Also, I'm not advocating for these YT personalities to be burned at the stake, I'll probably keep watching myself. I just find the dancing amusing, that is all.
 
Last edited:

Loxus

Member
jqRbjev.jpg


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.
Where in the world did you see me say the PS5 and XBSX isn't RDNA 2?
I said it's laid out similar, learn to read.
 

Clear

CliffyB's Cock Holster
The thing that's bugging me about this whole topic is the way its actively avoiding all the actually interesting stuff.

The reality is that PS5 APU gives over more die space to "other" function than the entire cpu part! And yet, because there's no easy analog with PC parts we're supposed not to be interested in why its there and what its doing in conjunction with the rest of the chip?

Remember this isn't a co-processor soldered onto the board, its positioned just as centrally as the cpu meaning it likely has just as much access. Its a pretty big deal, especially as you cannot conflate transistor usage with functional impact. MS in their documentation repeatedly referenced that a lot of the new featureset occupies relatively little room on the die.
 

ethomaz

Banned
The thing that's bugging me about this whole topic is the way its actively avoiding all the actually interesting stuff.

The reality is that PS5 APU gives over more die space to "other" function than the entire cpu part! And yet, because there's no easy analog with PC parts we're supposed not to be interested in why its there and what its doing in conjunction with the rest of the chip?

Remember this isn't a co-processor soldered onto the board, its positioned just as centrally as the cpu meaning it likely has just as much access. Its a pretty big deal, especially as you cannot conflate transistor usage with functional impact. MS in their documentation repeatedly referenced that a lot of the new featureset occupies relatively little room on the die.
Actually it is pretty hard to know what it is if you have no example in another chip.
 

Alder

Neo Member
Yes... it is half the size.
The point is what are the differences? Nobody has a answer.
It reduces the efficiency and precision of the core in general, besides reducing the physical floating point registers.

But I believe that you have to be satisfied, it seems to me that this cut was to enable energy and thermal space for GPU to reach high clock.
 

ethomaz

Banned
It reduces the efficiency and precision of the core in general, besides reducing the physical floating point registers.

But I believe that you have to be satisfied, it seems to me that this cut was to enable energy and thermal space for GPU to reach high clock.
That is not true.

You can have all the precision even with less silicon... in fact the lower precision theory was the first to be removed from the list.
Looks more like they cut things like FPADD from the unit.

From what the experts believes the native 256bits precision is still there.
 
Last edited:

LiquidRex

Member
With regards to Nintendo, I just can't help but feel the legendary game designers are dying out, and being replaced by a generation of Mobile gamers. 😔

Edit: damn it, wrong freaking thread. 😜
 
Last edited:

SlimySnake

Flashless at the Golden Globes
Ok, I'll take your word on that.

Now, what about the 4 games I mentioned? Because it seems some ppl get hung up on Miles and HFW, and just dismiss the others that are PS5 exclusive.

Some ppl are acting like every PS5 game announced for PS5 only has a PS4 version. They dont. And that could still change. But for now.... 2 games dont equal a full blown lie about believing in generations.
especially when their cross gen game looks so good we all thought it was next gen.

yPWXmMK.gif


Now if they downgrade the shit out of this game come launch then yes we can start to criticize Sony on the same level as Microsoft.
 

Alder

Neo Member
That is not true.

You can have all the precision even with less silicon... in fact the lower precision theory was the first to be removed from the list.
Looks more like they cut things like FPADD from the unit.

From what the experts believes the native 256bits precision is still there.

You need physical space, there is no free lunch

In the Zen/Zen+ microarchitecture the floating point physical registers, execution units, and data paths are 128 bits wide. For efficiency AVX-256 instructions which perform the same operation on the 128-bit upper and lower half of a YMM register are decoded into two macro-ops which pass through the FPU individually as execution resources become available and retire together. Accordingly the peak throughput is four SSE/AVX-128 instructions or two AVX-256 instructions per cycle.

Zen 2 doubles the width of the physical registers, execution units, and data paths to 256 bits. The L1 data cache bandwidth was doubled to match. The number of micro-ops issued by the FP scheduler remains four, implying most AVX-256 instructions decode to a single macro-op which conserves queue entries and reduces pressure on RCU and scheduling resources. AMD did not disclose how the FPU was restructured. Die shots suggest two execution blocks splitting the PRF and FP ALUs, one operating on the lower 128 bits of a YMM register, executing x87, MMX, SSE, and AVX instructions, the other on the upper 128 bits for AVX-256 instructions. This improvement doubles the peak throughput of AVX-256 instructions to four per cycle, or in other words, up to 32 FLOPs/cycle in single precision or up to 16 FLOPs/cycle in double precision. Another improvement reduces the latency of double-precision vector multiplications from 4 to 3 cycles, equal to the latency of single-precision multiplications. The latency of fused multiply-add (FMA) instructions remains 5 cycles.
 
Go and ask Locuza if PS5 is RDNA1.5? Then ask him the specs for it
I would say it's at best RDNA 1.3 or 1.45, but no way it's 1.5.

I'm an expert in version numbering schemes, I tell you it's not that close to RDNA 2, the xbox is full Direct X 12 Ultimate RDNA 2, this is why it's in a league of it's own in terms of performance. I mean, you can't even compare series X to other consoles, it's not even fair.
 

ToTTenTranz

Banned
the truth is the window where Sony would've officiated much or any of this stuff has seemingly passed.
To be fair, the PS4 Pro's rapid packed math was confirmed by Cerny pretty late in the game IIRC (almost half a year after release?).
Besides, there's still one big peripheral coming for the PS5 that warrants its own set of optimizations, which is PSVR2. It could very well be that features like foveated rendering will only be mentioned when PSVR2 is announced.
Speaking of which:



Displayport over USB!
So VirtualLink got shelved in the Summer of 2020, but a couple of months later AMD launches a line of GPUs supporting it and Sony supports it on the PS5?
Maybe it's coming back in some other yet-unannounced form? I did hear there are a lot of new headsets coming out in 2021.

Regardless, it looks like PSVR2 will be using a single cable for power, comms and display. No more dedicated power bricks, breakout boxes and half a dozen cables running around. One wire between the headset and the console and that's it. Great news!

This could also be the reason why the PS5 is using a 350W PSU despite showing the same power consumption as the SeriesX. The VirtualLink can send up to 27W of power towards the headset. Subtracting the wattage difference needed for powering VirtualLink from the PS5's PSU would put it practically with the same power requirements as the SeriesX (315 vs 323W).
 

PaintTinJr

Member
If PS5 isn't rdna2 how can they get those clocks then?
Someone correct me if I'm wrong, but aren't those RDNA2 PC card clocks a boost clock - meaning not a sustained game clock? And on the PS5, it is a continuous boost clock - IIRC from Road to PS5 wording - so not actually an AMD boost clock or an AMD game clock, no?
 

Kenpachii

Member
Someone correct me if I'm wrong, but aren't those RDNA2 PC card clocks a boost clock - meaning not a sustained game clock? And on the PS5, it is a continuous boost clock - IIRC from Road to PS5 wording - so not actually an AMD boost clock or an AMD game clock, no?

No RDNA1 card can push PS5 clocks let alone for a mass product like the PS5.

RDNA2 easily can hit it.

I am convinced its RDNA2, but things could have been cut for the sake of heat / power control. Still makes it RDNA2 tho.
 
Last edited:

PaintTinJr

Member
The predetermined and pre-emptive clock changing on the PS5, multiple times per frame - every 2ms IIRC - seems like something beyond current AMD products IMHO, and was even noted by Cerny as the reason they had to cap the clock at 2.23Ghz to make sure the thermal control logic was stable - paraphrasing a touch there - so presumably using that control logic, and just beefing up power supply and cooling for edge cases would have allowed them to take the clock higher.

Cerny's self posing Road to Ps5 question, about whether they would minimally use 256bit instructions, or major 256bit instruction usage - that can draw a lot of power - also brings in the continuous boost clocking topic IMO, because if by using pre-emptive variable clocks can keep power usage the same, you would then assume that circuit impedance - for the active circuitry - would remain constant too.

So if block design previously left space between logic/transistors to accommodate a temporal latency for heat build up in a functional block - to keep it stable when resistance rose - then maybe with the new paradigm shift to continuous boosting even nullifies the heavy power draw of 256bit instructions, and has given them latitude to even reduce spacing for transistors, because the heat never comes in the normal way, because clocking pre-emptively avoids the rise in resistance.

/probably wishful thinking and completely wrong, but Cerny definitely seems happy about the self posing question, as though the option was available because they had a solution.
 
Last edited:

longdi

Banned
For me it's none of it being confirmed to date by Sony. Mark Cerny is a details heavy guy. If VRS, Mesh Shaders and Sampler Feedback were there he would have said it. The moment he said Primitive Shaders I had a suspicion.

Metro developers confirmed they are doing VRS for PS5 in software aka the shaders like they did for PS4, DF Alex saying PS5 doesn't have VRS among other things said by DF regarding feature differences between the consoles, same for Mesh Shaders and Sampler Feedback (say what you will about DF, but their access to dev sources are legit and second to none)

Then it's Microsoft, backed up by AMD, saying that Xbox Series X|S is the only console with full hardware support for all RDNA 2 features AMD showcased at the RX 6000 reveal event. We've caught Microsoft being cute before, but I don't think this is one of those cases. AMD wouldn't have allowed such a statement were it not accurate. Microsoft has always known what their advantages are over PS5, which is even why they mentioned to DF about their clocks on GPU and CPU being stable or fixed.

Not really making a judgment on what's better between fixed or variable, but clearly Microsoft saw it as an advantage over PS5. I feel each side from the get go were pushing what they felt was their big advantage. Sony pushed their SSD I/O, Microsoft pushed their GPU and CPU power.

yep it's down to sony to tell us more about their custom works done on ps5 soc. until then, cerny did gone into details about his I/O works, api tools and a bit of audio. so we can infer that's where ps5 strengths lies in.

imo we should be seeing the strengths of SX custom works on the more 'traditional' areas of gpu/cpu/ram. proof is in the pudding, soon.

what i expect to see higher fidelity on SX and more dev friendly on ps5 helping it to punch above it's weight for a year or so.

will SX fidelity run away like ps4? of course not, but it will do well enough to question the value of $499+gp+higher fidelity..
 
Last edited:

Kenpachii

Member
Because my definition of RDNA2 says it’s impossible. /s

The rdna2 “based” notion obviously requires the 7nm+ process, at the very least.


Ever since AMD announced the RDNA2 architecture, they have reiterated a singular goal: they wanted to achieve a 50% jump in perf-per-watt over RDNA1. And that they would accomplish it entirely with architectural improvements, not process improvements

Honestly i don't see why people even debate if its rdna1 its not. its 2.

It's probably on 7nm because that what was available at the time. I don't see how u can enhance rdna1 to run at 2,23ghz even remotely without it automatically becoming RDNA2.
 
Last edited:
Honestly i don't see why people even debate if its rdna1 its not. its 2.

It's probably on 7nm because that what was available at the time. I don't see how u can enhance rdna1 to run at 2,23ghz even remotely without it automatically becoming RDNA2.

If there's one specific part that's been confirmed to be RDNA2 it's the GPUs CUs.

20200329152923.jpg


Without that it would be impossible to hit those clocks. They also have a robust cooling system which also helps.
 
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.


He was going on for quite a while, but help me out with what I'm suppose to be seeing there. I understand various sections or functional blocks of the different GPU may have version numbers and such, but often times those version numbers aren't even settled on and can even change. For the record, Locuza himself readily acknowledges he himself is no expert on these things either.

What I do know, however, is this.

Xbox Series X is packing all the same DirectX 12 Ultimate Feature support as all RX 6000 GPUs, has built in hardware support for every new feature AMD highlighted at their reveal. Sampler Feedback Streaming, as built for Series X is actually not a default DX 12 Ultimate feature and has additional customizations on top of it according to a Graphics R&D & Engine Architect at Microsoft. Sampler Feedback is just a core feature of what Microsoft built custom for Xbox Series X. Not only does Xbox Series X cover the full DX 12 Ultimate feature set, it actually exceeds the DX 12 Ultimate spec for Mesh Shaders for thread group size. Max on RX 6000 series is 128. Series X goes up to 256, and 256 on Series X does indeed produce superior results to all other thread sizes. RX 6000's main advantage would be it's a much larger GPU, but Series X actually has a more advanced Mesh Shader implementation.

What else? Xbox Series X has Machine Learning Acceleration hardware support whereas RX 6000 does not. So Series X isn't only RDNA 2. By all documented accounts, it actually exceeds it.

fY8iQ71.png








Xbox Series X goes beyond the standard DX12 Ultimate feature Sampler Feedback, and has custom hardware built into the GPU to make it even better for streaming purposes. Drop this GPU on PC and give it as many compute units as RX 6800XT while freeing it from the power restraints of a console, and it's likely the superior chip in the long run when DX12 Ultimate becomes more prominent. Oh, and as a desktop chip it would also have IC. :p
 
Top Bottom