PS5 Die Shot has been revealed

Looking at the pics the biggest difference from the Series X and Series S APUs are the right size.

It is about the size of the Zen 2 APU silicon that is not part of either CPU or GPU... that big part of the silicon is not present on AMD or Series APUs.

The CPU seems to have some FP units removed... it is not a standard Zen 2 CPU.
 
Last edited:
A picture of my crotch is more interesting.
do want i need it GIF
 
One of the "hopes" for this generation was better physics, something that can benefits a lot from a strong CPU.
"But this can be done by the GPU".
Yes, it can, but you want to tax the GPU with extra work instead of letting it concentrate on pushing pixels?
The hell you know if it's a better or a worst thing for the ps5 CPU perfomance? Doubt they removed this part for the worst. It's simple logic. ps5 has maybe a separated RAM/bandwidth as pc? Anyway I remind Cerny say something about removed chip parts compared the pc equivalent because useless in his opinion, somewhere in road to the ps5.
 
Last edited:
Him ....lol
They can't even explain what people that don't know what they are looking or what to look for.
You ask a honest question and then can't explain anything but be condescending to members that aren't very knowledgeable in this field.

You came into the thread with an attitude of "lol nobody cares". I didn't realize you wanted people to take you seriously and explain things. Perhaps you should have asked instead.
 
We can't be sure what it's different on this FPU, but we can at least see that it isn't what we should expect to see on a Zen 2 CPU.

fxRNWcM.jpeg


On image posted by MasterCornholio MasterCornholio from the previous post you can clearly see the expected Zen 2 FPU inside the SeX dieshot (put two Zen 1's FPU one above the other and you have the Zen 2 FPU). The FPU we are seeing on the PS5's dieshot id different, it's long and thin, it's half the size.
[h4][/h4]
Yet series x has struggled in some cpu limited areas, so we know what real world results have shown us.

We need a proper breakdown of this and really some info from Sony before we know anything.

If no unified cache then it only further adds questions as to why ps5 out performs in some areas where cpu is a limit. Youd have to put it down to efficiency of data transfer and all components working in harmony, with minimal bottlenecks. Or there could be many other customisations that this grainy die show doesn't even begin to show or explain.
 
SOLD! I am good with that! :D



Well, RT is something attached to the CU, not part of it. But it's easier to start from an architecture that already has all the bits and pieces for RT than to try to attach is to an old generation architecture.

In the end it doesn't matter: the games do the talking, and as far as we can see, Sony is on par.

Not on par, but actually on top in most games.

Which is what should matter for people. Who cares if it's rdna 1.32 customized or 2.12 stock?

You understand none of this shit, it matters only for devs and those are only tags. You can achieve the same functions with another name... Stop thinking you're nuclear scientists.

The results are here, PS5 rocks and is probably a bit faster overall, who cares how they achieved that...
 
HOLY SHIT

Damn is neogaf even allowing this post? Damn someone is going to get in a lot of troubles :lollipop_fearful:
Just sold all my sony stocks

before the masses understand how fucked up this situation is
 
Last edited:
You came into the thread with an attitude of "lol nobody cares". I didn't realize you wanted people to take you seriously and explain things. Perhaps you should have asked instead.
I did it was a joke .
hqg GIF

Just trying to have fun while learning new things.
 
Yet series x has struggled in some cpu limited areas, so we know what real world results have shown us.

We need a proper breakdown of this and really some info from Sony before we know anything.

If no unified cache then it only further adds questions as to why ps5 out performs in some areas where cpu is a limit. Youd have to put it down to efficiency of data transfer and all components working in harmony, with minimal bottlenecks. Or there could be many other customisations that this grainy die show doesn't even begin to show or explain.
Or because Sony is the better selling platform by a lot developers put 80% of the effort on the Sony side. It only makes sense to makes sense to polish to best selling sku and get the others running. Control perfect example using the Sony custom io and other features series x was a quicky pc port.
 
The hell you know if it's a better or a worst thing for the ps5 CPU perfomance? Doubt they removed this part for the worst. It's simple logic. ps5 has a separated RAM/bandwidth as pc?
exactly, in terms of altering or changing something that was percieved as a "must" in last gen having the ram pool not only split up again but even one fast and one slow one comes across as ,from all partys involved , the compromise with the biggest performance impact.
And People need to understand that Cerny was not realy beaten by Xbox One X. PS4 Pro came out an entire year earlier for 100Dollar less. There was no goal or even need to be competitive with Xbox One X .
So some Xbox Fanboys saw the almost 2 Tflop more on Xbox one X and thought that the MS Team "beat" Cerny and suspected that they will repeat that with Series X.

This DIE Shot will finaly proofe that PS5 APU is more sophisticated than the Series X one. And that this is even true if something from RDNA2 Featureset is not there as it is in Series X. If Cerny gets rid of something in your Next Gen Architecture - you dont ask why - you fire the one who implemented it in first place.
Have trust ..
 
Or because Sony is the better selling platform by a lot developers put 80% of the effort on the Sony side. It only makes sense to makes sense to polish to best selling sku and get the others running. Control perfect example using the Sony custom io and other features series x was a quicky pc port.
No, there's not just one reason, its always a combination.

Each multiplat will be different, yes lead platform is a factor, but will vary.

One engine prefers one consoles and visa versa.

There are many variables that will affect results.

As matt hargett said himself.



They will continue to swing per title meaning hell for fanboys on both sides as one weeks 'win' is followed by a 'loss'.

Listen to the devs who said closest consoles ever, it's really that simple.
 
Last edited:
So why is it performing as well as it has?

Something seems wrong.
if everything IS true what he says than it is also simple i guess. They got rid of what is missing because they were confindent about ther very sophisticated I/O Co Processors and array to make up for it. And i think the fact that PS5 keeps up or even beat Series X with the later having almost 2 Tflop more is prooving them right.
 
Told people last year to hold on with all the infinity/shared cache stuff until it was confirmed, but people love that secret sauce taste.

Honestly this makes the performance comparisons even more confusing.

if everything IS true what he says than it is also simple i guess. They got rid of what is missing because they were confindent about ther very sophisticated I/O Co Processors and array to make up for it. And i think the fact that PS5 keeps up or even beat Series X with the later having almost 2 Tflop more is prooving them right.

Maybe Sonys goal was to eliminate all bottlenecks and design an incredibly efficient system?

That's all that I can think of to be honest.
 
Last edited:
Or because Sony is the better selling platform by a lot developers put 80% of the effort on the Sony side. It only makes sense to makes sense to polish to best selling sku and get the others running. Control perfect example using the Sony custom io and other features series x was a quicky pc port.
Yes that must be it. Devs favour Sony therefor spend more time on their games to make sure they run the same if not better
 
HOLY SHIT

Damn is neogaf even allowing this post? Damn someone is going to get in a lot of troubles :lollipop_fearful:
Just sold all my sony stocks

before the masses understand how fucked up this situation is

If you're interested in money you should have sold them last week anyway.
 
Its true though. XSX (so far) has shown that specs on paper mean little. Its about the games and PS5 will hold its own there for me.
Regardless, PS5 and XSX are way better cosoles at launch than PS4 and XB1 were at launch. This will be a good gen once it gets going.
Also closer, even on specs, than last gen. Which is something everyone should be pleased with.
 
Could be the additional CUs that cause extra heat to be produced so they have to reduce the clocks.

I would really love to see a comparison between the CUs to determine if they are the same or not.

So how can a 72-80CU card can clock at 2.5GHz with common paste, not even liquid metal?

As we saw with the RX 6800 and 6800 XT, AMD has now started to be a bit more conservative in its official boost clock figures. All of the RX 6000 GPUs have exceeded their boost clocks in our Metro Exodus testing, and we definitely noticed other games going above 2250 MHz. Overclocked, the card even managed a cool 2.5 GHz, which is pretty freaking awesome. We can't wait to see what the extreme overclockers fare with liquid nitrogen.

 
Last edited:
So how can an 72-80CU card can clock at 2.5GHz with common paste, not even liquid metal?

As we saw with the RX 6800 and 6800 XT, AMD has now started to be a bit more conservative in its official boost clock figures. All of the RX 6000 GPUs have exceeded their boost clocks in our Metro Exodus testing, and we definitely noticed other games going above 2250 MHz. Overclocked, the card even managed a cool 2.5 GHz, which is pretty freaking awesome. We can't wait to see what the extreme overclockers fare with liquid nitrogen.

It's just for lower temps, normal paste would probably work here too, but console would be noisier and that would sucks. Also higher temp = biggesr chance of failure, bugs, etc...
 
Broken tools?

Nope.

Look at the structure of the PS4 Pro and the X1X, compare those to the GPU and memory arrangements of the PS5 and XSX. Then remember that all the games so far are based on last gen engines and "next gen" versions are upgraded PS4 Pro and X1X versions.

It's incredibly simple, the simplest explanation is almost always the correct one.
 
So how can a 72-80CU card can clock at 2.5GHz with common paste, not even liquid metal?

As we saw with the RX 6800 and 6800 XT, AMD has now started to be a bit more conservative in its official boost clock figures. All of the RX 6000 GPUs have exceeded their boost clocks in our Metro Exodus testing, and we definitely noticed other games going above 2250 MHz. Overclocked, the card even managed a cool 2.5 GHz, which is pretty freaking awesome. We can't wait to see what the extreme overclockers fare with liquid nitrogen.

I think it was pretty clear why...


images


PS5 sustain these clocks with one fan at low rpm (noise).
 
Last edited:
It's heavily overclocked, that's why they use liquid metal, and that's why it has the performance it has, for now ...
 
Last edited:
I don't quite understand this. The tweet seems to suggest PS5 is not using full RDNA2 features? Didn't we already establish that a while ago?

Told people last year to hold on with all the infinity/shared cache stuff until it was confirmed, but people love that secret sauce taste.

Cerny said it's RDNA2 based but decided not to use some features while adding others. And Cerny never mentioned Infinity Cache. So, please, educate me how this die shot is different from Cerny's presentation?

Or because Sony is the better selling platform by a lot developers put 80% of the effort on the Sony side. It only makes sense to makes sense to polish to best selling sku and get the others running. Control perfect example using the Sony custom io and other features series x was a quicky pc port.

Come on, this is getting pathetic real fast and it's offensive to devs.
 
It's heavily overclocked, that's why they use liquid metal, and that's why it has the performance it has, for now ...
Wrong.

A lot of test showed PS5 APU runs fine with thermal paste but it increase the fan rotation (noise).

Sony choose Liquid Metal because fan noise was a complain with PS4 Pro... with Liquid Metal the PS5 fan runs at low rpm (low noise).
 
Last edited:
Cerny said it's RDNA2 based but decided not to use some features while adding others. And Cerny never mentioned Infinity Cache. So, please, educate me how this die shot is different from Cerny's presentation?



Come on, this is getting pathetic real fast and it's offensive to devs.

When did I mention Cerny? You made that up.
I'm talking about the people here and on Twitter who were saying this unified L3 cache was "basically confirmed" and running with it as fact.
 
So infinite cache was a lie huh! RedGamingTech and Mlid kept feeding us lies all this time just for views
🤬
What ? Infinity Cache is a name for Desktop RDNA2 L3 cache. It was revealed during late october presentation. Navi21 has 128MB L3 cache, PS5 with 128MB l3 cache would have had +400mm2 APU instead of 300mm2
 
Oh no I was talking about the CUs.

20200329145127.jpg


If this die shot reveals that the CUs are RDNA1 CUs then Cerny would have been lying to us.

I don't think he would talk about RDNA2 CUs if they weren't in the PS5. It would be kind of silly to talk about something that's not even in the console.
This. RDNA 2 cu's should be legit unless Cerny told a massive lie. He was specific about this.

The tweet is intriguing tho.

What's FPU?
 
Last edited:
Wrong.

A lot of test showed PS5 APU runs fine with thermal paste but it increase the fan rotation (noise).

Sony choose Liquid Metal because fan noise was a complain with PS4 Pro.

Also people forget they went with liquid metal so they didn't have to use more expensive heatsink.

I think we can all write off PS5 overheating issues.
 
BTW where is coming that nonsense of Infinte Cache? It was a no go since we saw the size of the die... there is no way to have Infinite Cache.

Infinite Cache is a big part of Navi21 silicon.
 
Last edited:
And is that a bad thing?

I mean what matters is what we actually play right?
Sure , that's true . But wouldn't it be great if it was forward looking ? And supporting engine designs of the future? Or is it better that the market leader can dictate the market and how engines are build and designed ?
 
The hell you know if it's a better or a worst thing for the ps5 CPU perfomance? Doubt they removed this part for the worst. It's simple logic. ps5 has maybe a separated RAM/bandwidth as pc? Anyway I remind Cerny say something about removed chip parts compared the pc equivalent because useless in his opinion, somewhere in road to the ps5.

I'm consider myself a "sony fanboy", but I would be ashamed of making up excuses about this, like "it's to improve the BC". One of the biggest reasons why Zen 2 was so much better than Zen 1 for games was the improved FPU.
Power consumption isn't enough justification, this is still a mobile GPU, don't use that much power even more so low clocked like in these consoles. Why would I cut the FPU in half to have vector code running in double the time in a time critical task as running games in high fps?


Yet series x has struggled in some cpu limited areas, so we know what real world results have shown us.

We need a proper breakdown of this and really some info from Sony before we know anything.

If no unified cache then it only further adds questions as to why ps5 out performs in some areas where cpu is a limit. Youd have to put it down to efficiency of data transfer and all components working in harmony, with minimal bottlenecks. Or there could be many other customisations that this grainy die show doesn't even begin to show or explain.

The hardware is similar but the software is very different.
We know very well how Microsoft struggles with optimizing DirectX 12.
 
Top Bottom