Red Gaming Tech (RGT): The PS5 has a custom feature to be found in future RDNA 3 Cards.

What do you think? Is it true that the PS5 has a custom feature to be found in future RDNA 3 cards?


  • Total voters
    431
We will know for sure when the die shots come out after the console releases. Even 8MB of Sram would be obvious.

infinity-cache-100863932-large.jpg


FYI, RX 6800 XT's 128 MB Infinity Cache surrounds the stream processors instead of XBO style's single block 32MB ESRAM design. AMD avoids repeating XBO's mistakes.
 
So your answer to me saying that the ROPS don't add up with SX having 64ROPS with 56CUs vs RX 6800 96ROPS with 60CUs is to show me a RDNA1 card with 64ROPS with only 36CUs?
RX 5700 has 36 CU with 64 ROPS. https://www.amd.com/en/products/graphics/amd-radeon-rx-5700

From https://www.amd.com/en/products/graphics/amd-radeon-rx-6800-xt
RX 6800 XT has 128 ROPS. 128 MB infinity cache

RX 6800 has 96 ROPS. 128 MB infinity cache

BiG NAVI's read/write ROPS units are pretty good.

XSX GPU's 64 MB cache...points to 64 ROPS.
 
Last edited:
No fun ruined here. Just information people have known for a while. Thanks for making it seem Sony didn't co-develop, though. Do you want me to link the Road to PS5 where Mark Cerny talks about just this too?
You stated, "Sony has been co-developing Navi with AMD since the beginning" which is not true.

I'm using the primary source from AMD's David Wang's statements.

AMD's David Wang's statements override the narratives from Jason Evangelho (speculated Scarlett using Vega? LOL) and Anthony Garreffa (the article is Sony centric).
 
Last edited:
You claimed, "Sony has been co-developing Navi with AMD since the beginning" which is not true.

I'm using the primary source from AMD's David Wang's statements.

AMD's David Wang's statements override the narratives from Jason Evangelho and Anthony Garreffa
I'm sorry if Sony wasn't there right at the very start. Does that make you feel better? Good Lord the nitpicking. The point I was making is this.

Straight from Cerny's mouth from The Road to PS5.
"if you see a similar discrete GPU available as a PC card at roughly the same time as we release our console, that means our collaboration with AMD succeeded in producing technology useful in both worlds. It doesn't mean that we at Sony simply incorporated the PC part into our console."

This means Sony has been there co-developing from VERY EARLY IN DEVELOPMENT. Better?
 
I'm sorry if Sony wasn't there right at the very start. Does that make you feel better? Good Lord the nitpicking. The point I was making is this.

Straight from Cerny's mouth from The Road to PS5.
"if you see a similar discrete GPU available as a PC card at roughly the same time as we release our console, that means our collaboration with AMD succeeded in producing technology useful in both worlds. It doesn't mean that we at Sony simply incorporated the PC part into our console."

This means Sony has been there co-developing from VERY EARLY IN DEVELOPMENT. Better?
Don't waste your time. Emotional intelligence makes fans (on both sides) look stupid.
 
I'm sorry if Sony wasn't there right at the very start. Does that make you feel better? Good Lord the nitpicking. The point I was making is this.

Straight from Cerny's mouth from The Road to PS5.
"if you see a similar discrete GPU available as a PC card at roughly the same time as we release our console, that means our collaboration with AMD succeeded in producing technology useful in both worlds. It doesn't mean that we at Sony simply incorporated the PC part into our console."

This means Sony has been there co-developing from VERY EARLY IN DEVELOPMENT. Better?
1. You stated, "Sony has been co-developing Navi with AMD since the beginning".

2. That's another argument.

Don't waste your time. Emotional intelligence makes fans (on both sides) look stupid.
Dave_at_Home stated "Sony has been co-developing Navi with AMD since the beginning".

There are no emotions in my direct AMD's quote.
 
1. You stated, "Sony has been co-developing Navi with AMD since the beginning".

2. That's another argument.


Dave_at_Home stated "Sony has been co-developing Navi with AMD since the beginning".

There are no emotions in my direct AMD's quote.
Again, let it go. Maybe they weren't there at the very start, but it's absolutely clear they were there early in development. There may be no "emotions" here, but you're splitting hairs for...reasons
 
Last edited:
During his live stream, tonight Moore's Law Is Dead YouTuber known for predictions on tech called out MS over their full RNDA 2 statement and also pointed out if you want to go down that road the Xbox would be 1.5 RNDA because it doesn't have infinity cash and PS5 has. Also claims PS5 has RDNA 3 features.



I don't fault MS for the marketing hustle. You can see it works on their hardcore fans. That however doesn't mean when we're trying to have a discussion on it that logic should fly out the window cause daddy MS said it. More info will trickle down the more the microscope is placed over these topics. There's gonna be a lot of salt and invisible walkbacks.
 
Last edited:
RX 5700 has 36 CU with 64 ROPS. https://www.amd.com/en/products/graphics/amd-radeon-rx-5700

From https://www.amd.com/en/products/graphics/amd-radeon-rx-6800-xt
RX 6800 XT has 128 ROPS. 128 MB infinity cache

RX 6800 has 96 ROPS. 128 MB infinity cache

BiG NAVI's read/write ROPS units are pretty good.

XSX GPU's 64 MB cache...points to 64 ROPS.

Series X has 76MB of SRAM across the whole SoC & 64 ROPS been known for awhile.
 
Infinity cache 128mb takes up 100+mm^2 on RX 6000 series.

The PS5 SOC is ~310mm^2.

Do the math.
There's "a very large SRAM" (on-chip RAM) as Cerny calls it, in the PS5 SoC. He didn't go into much detail at the time in his Road to PS5, nor did he mention its size. I presume it was due to NDA and it's very likely a variant of infinity cache? Perhaps it's 64 MB?
 
There's "a very large SRAM" (on-chip RAM) as Cerny calls it, in the PS5 SoC. He didn't go into much detail at the time in his Road to PS5, nor did he mention its size. I presume it was due to NDA and it's very likely a variant of infinity cache? Perhaps it's 64 MB?
Sadly we might not know this unless a 1st party Dev talks about it for their game. Sony does not seem to care about tech PR/buzzwords at this point

Considering the GPU is half the CUs as a 6800xt, 64MB would make sense, no?
 
Last edited:
There's "a very large SRAM" (on-chip RAM) as Cerny calls it, in the PS5 SoC. He didn't go into much detail at the time in his Road to PS5, nor did he mention its size. I presume it was due to NDA and it's very likely a variant of infinity cache? Perhaps it's 64 MB?

I thought it was some sort of cache for the I/O subsystem... 🤔
 
Sadly we might not know this unless a 1st party Dev talks about it for their game. Sony does not seem to care about tech PR/buzzwords at this point

Considering the GPU is half the CUs as a 6800xt, 64MB would make sense, no?

Yes i wonder why Sony does not care this time round. They always do techy tallks before.

Just a do a video conference with DF or something.
Unusually quiet about their tech even after Amd has revealed rDNA2? 🤷‍♀️
 
There's "a very large SRAM" (on-chip RAM) as Cerny calls it, in the PS5 SoC. He didn't go into much detail at the time in his Road to PS5, nor did he mention its size. I presume it was due to NDA and it's very likely a variant of infinity cache? Perhaps it's 64 MB?
You guys are setting yourselves up for disappointment. While it would be nice, it feels like if PS5 had any form of infinity cache then it would have been announced with the AMD conference
 
See, this is where holding your tongue and trying to stay out of these bullshit dick waving conversations gets really difficult. We have a guy here, in this video calling out MS about claims they have made about their architecture.
Now putting aside the fact that I have no idea what the credentials are of this fella in the field of computer science and engineering are, if any, why would guys like Jason Ronald and Andrew Goossen risk damaging their professional reputations by bullshitting about hardware designs they have, in some way or another, been involved with?
It makes absolutely zero sense for them to do so, if you apply even the tiniest bit of common sense in trying to understand the narrative that videos like this one are selling.

I've said in prior post people are dismissing the fact that AMD's logo is on that tweet. I highly doubt if Microsoft were saying something untrue AMD would allow their logo be used. Especially in a tweet that's indirectly aimed at their other console customer.


Also here is what Sebastien Nussbaum the Corporate Vice President at AMD said about Series X back March.
The Xbox Series X is the biggest generational leap of SoC and API design that we've done with Microsoft, and it's an honor for AMD to partner with Microsoft for this endeavor. The Xbox Series X is going to be a beacon of technical innovation leadership for this console generation and will propagate the innovation throughout the DirectX ecosystem this year and into next year.
 
The hints for AMD's Infinity Cache for the GPU was given at XSX's reveal.

MS claimed 76 MB cache for SoC and Zen 2 CPU's cache configuration was the mobile version, hence leaving 64 MB cache for the GPU with two SE. MS didn't elaborate on the GPU's 64 MB cache for obvious reasons.

Quad SE (80 CU scale) version with 128 MB cache can be speculated.

This is something fishy but nice question.
Renoir apu only has 12mb cache.
Why does Series X soc get 76mb?
My full desktop 16c/32t 3950x has only 72mb cache? :messenger_medmask:

I wonder what you get pairing 352b memory bandwidth with 64mb soc cache in Series x? lots more power?

can someone with twitter account ping/dm MS guys? Let's get an official answer from official sources.
 
Last edited:
Yes i wonder why Sony does not care this time round. They always do techy tallks before.

Just a do a video conference with DF or something.
Unusually quiet about their tech even after Amd has revealed rDNA2? 🤷‍♀️
almost as if they're confident and dont need to tweet disingenuous PR
 
You guys are setting yourselves up for disappointment. While it would be nice, it feels like if PS5 had any form of infinity cache then it would have been announced with the AMD conference
All good. I expect PS5 to consistently come out with lower resolutions in multiplat comparisons against SX despite PS5 having some advantages to its GPU.

I'm waiting for the all important first-party exclusives. And how teams like Naughty Dog, Guerrilla Games will leverage the Zen 2, new GPU features like geometry engine and what this massive jump from 2.5in crappy laptop HDD to blazing fast PCIe 4.0 SSD allows them to achieve (load times won't be the only improvement😉).

That's what I'm most excited about.
 
Last edited:
It's like the title and first sentence were out of breath. Let me try and understand it.

YouTube channel named "Moore's Law Is Dead", who makes predictions, has new predictions.

Prediction 1: Xbox Series X|S doesn't have the full set of RNDA 2.0 features.

(There's something about infinity cash I don't really understand in the middle)

Prediction 2: PlayStation 5 has features that will be in the RNDA 3.0 set.

(There's already been a thread about PS5 having custom features that may make it to 3.0).
2013 misterxmedia level secret sauce ..reversed
 
I'm sorry if Sony wasn't there right at the very start. Does that make you feel better? Good Lord the nitpicking. The point I was making is this.

Straight from Cerny's mouth from The Road to PS5.
"if you see a similar discrete GPU available as a PC card at roughly the same time as we release our console, that means our collaboration with AMD succeeded in producing technology useful in both worlds. It doesn't mean that we at Sony simply incorporated the PC part into our console."

This means Sony has been there co-developing from VERY EARLY IN DEVELOPMENT. Better?
I'll tell you again dosnt matter what they developed with. is just irrilevant if the result is an inferior ..bigger that drain more watt and with less features set machine that cost the same of a better one
 
Last edited:
What's the 9TF soc?
oberon....the soc that mark cerny said they was working on clocking it at 2 ghz ... and straight from his mouth ... they couldn't keep the 2ghz stable. Now we have a 2.2ghz variable that everyone thinks will keep 2.2 99% of the time ..........
PS. take note that ps5 soc have the same AMD code wrote over it and is the same oberon code
 
Last edited:
oberon....the soc that mark cerny said they was working on clocking it at 2 ghz ... and straight from his mouth ... they couldn't keep the 2ghz stable. Now we have a 2.2ghz variable that everyone thinks will keep 2.2 99% of the time ..........
PS. take note that ps5 soc have the same AMD code wrote over it and is the same oberon code

How do you quote Mark Cerny and take his words out of context and then say, "everyone thinks will keep 2.2 99%" of the time?

He said himself that it will stay at that clock most of the time.

Trouble having fixed clock at 2GHz was BEFORE they implemented smart shift.

If you want to quote Mark Cerny, you can't only accept words you want to hear.
 
Yes.. Because they have tested it and stated that it holds this frequency the vast majority of the time and drops below only last a moment.
we will see from the df comparation if we will have just 18 % difference vast majority of the time ...or was just a way to sell the console
 
we will see from the df comparation if we will have just 18 % difference vast majority of the time ...or was just a way to sell the console

that's nice, point is you are acting as if people are coming to some crazy conclussion based on nothing when they are using the same person in the same interview that you just used to try to imply that it would.

Cerny said they couldn't maintain that frequency until they used variable power delivery so that they could maintain the clocks 99% of the time.
WHY ARE PEOPLE ACTING LIKE IT"S SUDDENLY GOING TO MAINTAIN IT 99% OF THE TIME.
 
that's nice, point is you are acting as if people are coming to some crazy conclussion based on nothing when they are using the same person in the same interview that you just used to try to imply that it would.
did you see the test on smartshift laptop? you think is all that incredible innovation? it help but that's it

what I mean is that people should sip what is said for PR purposes as well. from the ms slides presented at hotchips Microsoft tells us that SFS will allow you to load x3 times less data .. only this without even taking directsrorage should already fill the gap of difference with the sony ssd? should i believe it just because jason and ms said so? I have very strong doubts ... very strong doubts .. that it can approach the ssd of the ps5. same thing applies to the variable clock.
we have seen many times both companies try to oversell their products and innovations.
 
Last edited:
oberon....the soc that mark cerny said they was working on clocking it at 2 ghz ... and straight from his mouth ... they couldn't keep the 2ghz stable. Now we have a 2.2ghz variable that everyone thinks will keep 2.2 99% of the time ..........
PS. take note that ps5 soc have the same AMD code wrote over it and is the same oberon code
He never said 9TF's and neither does it go to 9TF's when and if it drops to 2Ghz.
 
did you see the test on smartshift laptop? you think is all that incredible innovation? it help but that's it

what I mean is that people should sip what is said for PR purposes as well. from the ms slides presented at hotchips Microsoft tells us that SFS will allow you to load x3 times less data .. only this without even taking directsrorage should already fill the gap of difference with the sony ssd? should i believe it just because jason and ms said so? I have very strong doubts ... very strong doubts .. that it can approach the ssd of the ps5. same thing applies to the variable clock.
we have seen many times both companies try to oversell their products and innovations.

That's a really terrible, Microsoft never said that SFS fills the gap between XBSX and PS5. They said that using that technology gives a a 2.5x boost over not using it at all. they never stated that they didn't have something similar before that wasn't already giving some of that boost and they definitely never stated that Sony won't have something similar.

Skepticism is healthy and companies very often talk around things and speak in ways that make people make assumptions. But a statement like "Our clocks maintain X frequency for X% of the time it leaves little room for doubt on what he's saying.
 
That's a really terrible, Microsoft never said that SFS fills the gap between XBSX and PS5. They said that using that technology gives a a 2.5x boost over not using it at all. they never stated that they didn't have something similar before that wasn't already giving some of that boost and they definitely never stated that Sony won't have something similar.

Skepticism is healthy and companies very often talk around things and speak in ways that make people make assumptions. But a statement like "Our clocks maintain X frequency for X% of the time it leaves little room for doubt on what he's saying.
they said that will load 1/3 of datas....
if I have to load 10 gb with a ssd that load 2.5 gb/s it take 4 secs, if with SFS those 10gb become 3.3 it will take 1 second and something (i know that is a little bit different from this because they were talking just about textures and game assets it was to give u an idea on how company say things).
What percentage Cerny talked about ? he gave a raw number of how much % of the time it will stay at 2.2ghz ? really? i don't remember ..I'm being honest
 
Last edited:
When will the 9tf die?

Seriously... What does it take for it to die?
A first party playing a game for 1h on a debug ps5 with the profilers running on screen?
 
When will the 9tf die?

Seriously... What does it take for it to die?
A first party playing a game for 1h on a debug ps5 with the profilers running on screen?
the fact that ps5 have the same leaked soc code of oberon?
probably after we have some perf data on both console we will find out ..we are weeks aways
 
Last edited:
they said that will load 1/3 of datas....
if I have to load 10 gb with a ssd that load 2.5 gb/s it take 4 secs,

1/3 of the data compared to a GPU that does not have anything resembling this tech. Not compared to even the ps4 that we know have similar techniques and definitely not compared to the PS5.

Anyway Like you said, we are days away not gonna waste more time.
 
1/3 of the data compared to a GPU that does not have anything resembling this tech. Not compared to even the ps4 that we know have similar techniques and definitely not compared to the PS5.

Anyway Like you said, we are days away not gonna waste more time.
there are no gpu that have hardware sfs not even the new rdna2 gpu presented by amd. what you talking about
 
Last edited:
You guys don't get it. They NEED the PS5 to be 9 instead of 10 Tflops because 3 is a bigger number than 2 when comparing it to XSX and it sounds worse! Now... on the flip side... the Series S is a GREAT deal and won't hold ANYTHING back since EVERYTHING can scale... just keep all that in mind while we tell you a 3 Teraflop deficit is a big deal (with no proof there IS a 3 Teraflop deficit)
 
You guys don't get it. They NEED the PS5 to be 9 instead of 10 Tflops because 3 is a bigger number than 2 when comparing it to XSX and it sounds worse! Now... on the flip side... the Series S is a GREAT deal and won't hold ANYTHING back since EVERYTHING can scale... just keep all that in mind while we tell you a 3 Teraflop deficit is a big deal (with no proof there IS a 3 Teraflop deficit)

Can't wait to see comparisons between Xbox 4 and PS5
 
Can't wait to see comparisons between Xbox 4 and PS5

I know, even with a 2 Tflop deficit things like variable clocks and i/o throughput could really matter, nevermind APIs. There are so many factors here that even if these guys were right it wouldn't be the be-all, end-all on what performs better in the end. Honestly my money is on PS5 being the winner for performance in most cases and even if they often lose they'll have the best looking games because their first party is unbeatable. It could be PS3 all over again, Call of Duty looks like a blurry stain on the screen but then Killzone 2 makes the 360 cry. We'll see, well, the best looking launch game is a PS5 game so it's sort of already true but if they win the multi-plat argument, too? LOL, what a gen this will be.
 
I know, even with a 2 Tflop deficit things like variable clocks and i/o throughput could really matter, nevermind APIs. There are so many factors here that even if these guys were right it wouldn't be the be-all, end-all on what performs better in the end. Honestly my money is on PS5 being the winner for performance in most cases and even if they often lose they'll have the best looking games because their first party is unbeatable. It could be PS3 all over again, Call of Duty looks like a blurry stain on the screen but then Killzone 2 makes the 360 cry. We'll see, well, the best looking launch game is a PS5 game so it's sort of already true but if they win the multi-plat argument, too? LOL, what a gen this will be.

I just wanted to double down on your comment on how much people care about numbers by pointing out that Xbox has to come up with these ridiculous names just so their number isn't small.

I can't pretend to actually care about who 'wins" in tech that much. I just like seeing the innovations. right now the thing I'm enjoying most is playing Hyrule warrior at like 26fps average.

I'd probably guess PS5 will usually be 10% to 30% weaker "less res" on average but if it outperforms xbox I would experience 0 surprise .Simply don't got enough hard evidence for me to really make a statement strongly enough that I would care if I ended up being wrong. If Xbox were getting FF7:REmake first that's what I'd get.
 
Last edited:
Top Bottom