Of course, and that's equally as silly
Of course, and that's equally as silly
If the masks are compatible between the nodes it is highly likely you can just jump to the better node without much effort. The thing is I do not know if that is the case.Yes, it could as N7P is design compatible to standard N7. My personal speculation is that the steppings moved to N7P as well as introduced the RDNA 2.0 features such as RTRT.
This is how I honestly see all this...What are insiders basing their information on? Do you know?
Based on:
1) It's a real AMD document.
2) The APU in the document is emulating the PS4 and PS4 Pro.
Do you doubt the Oberon APU that is tested in these files is a Sony APU?
I guess everyone are just bored so everything keeps floating back up over and over again. Sometimes it's the Github leak, sometimes it's HBM and sometimes it's spinning kick being redeemed :)
If the masks are compatible between the nodes it is highly likely you can just jump to the better node without much effort. The thing is I do not know if that is the case.
Locuza Do you know that?
Well, I doubt any other company will make a BC mode for PS4 and PS4 Pro other than Sony.It might be id say its likely one sure if that makes you feel better but i certainly don't claim to know it as a fact.
I'll take anything over the past few days Optimus Prime GIF-fest and zero next-gen talk and speculation :)1. I hate to keep bringing up his name, but Klee himself said that the Github leak literally did nothing to change the spec information he knew and he was adamant about it. Considering he also predicted XSX being shown at TGA before he mysteriously asked to be banned is only the cherry on top.
2. Yes. We know the information is real. What we don't know in the grand scheme is how that applies to PS5. We've already discussed it ad nauseum. I'm inclined to agree I'd rather read pages of unveil dates at this point than retread info we've already poured over months ago.
So thats off the table .... ?Everything I read states 7nm and 7nm euv have incompatible masks.
If the masks are compatible between the nodes it is highly likely you can just jump to the better node without much effort. The thing is I do not know if that is the case.
Locuza Do you know that?
TSMC's N7P uses the same design rules as the company's N7, but features front-end-of-line (FEOL) and middle-end-of-line (MOL) optimizations that enable to either boost performance by 7% at the same power, or lower power consumption by 10% at the same clocks. The process technology is already available to TSMC customers, the contract maker of chips revealed at the 2019 VLSI Symposium in Japan, yet the company does not seem to advertise it broadly.
Your right its unlikely for that to happen however this does not make it a fact, we can not change the definitions of words just because we wish toWell, I doubt any other company will make a BC mode for PS4 and PS4 Pro other than Sony.
I'll take anything over the past few days Optimus Prime GIF-fest and zero next-gen talk and speculation :)
Respecto al debate next-gen power, ya tienen los últimos kits de las 2, en terminos de raw-power va a ser un 360-ps3 más que un Ps4-One original.
Me ha dicho que están muy muy equiparadas, la gpu es ligeramente superior en one, en el procesador empate técnico y en el resto, ps5. Le encanta el mando por cierto, aunque muy continuista.
El está metido en un triple AAA de los gordos de los que ya han salido rumores, que se desmintieron sobre cierta época...pues son reales
Seems you have some confusion. As I said before, 7N and 7NP are design compatible. 7N and 7N+ are not design compatible.Everything I read states 7nm and 7nm euv have incompatible masks.
Just look above 1 comment above yours :)Seems you have some confusion. As I said before, 7N and 7NP are design compatible. 7N and 7N+ are not design compatible.
I'll give you that, let's change it to 99.99% sure it's a face while the 0.01% is a case where Sony is allowing some other company to build an APU which runs PS4 and PS4 Pro games :)Your right its unlikely for that to happen however this does not make it a fact, we can not change the definitions of words just because we wish to
Now THAT'S how you appeal to 7 billion customers!I'll give you that, let's change it to 99.99% sure it's a face while the 0.01% is a case where Sony is allowing some other company to build an APU which runs PS4 and PS4 Pro games :)
From spanish forum
Developers already have the latest dev-kits. Talking about raw-power, consoles are like PS3 vs 360 (really close) and there is no gap like PS4-ONE.
GPU is more powerful on Series X (only a little bit)
CPU is same PS5 and Series X
Other stuff is better on PS5 (I think he is talking about the SSD...)
The controller is really cool but almost same as PS4
The source is working on a big AAA, there was rumors about that AAA but it was claimed to be fake (but it's not, it's real. Idk what game can be)
The guy can not show any proof so....one rumor more.
It's got one thing in common with every developer leak I've seen : that both consoles are close in performance. I'm gonna just go with that.
My point is something either is a fact or it is not,we have very few facts to go on around here, barring the highly unlikely situation you layed out and assuming this is a chip made for sony you would have to once again assume if you want to say its whats in the ps5I'll give you that, let's change it to 99.99% sure it's a face while the 0.01% is a case where Sony is allowing some other company to build an APU which runs PS4 and PS4 Pro games :)
Could also be Sony piggybacking on the most similar chip available at the time to do some BC testing.I'll give you that, let's change it to 99.99% sure it's a face while the 0.01% is a case where Sony is allowing some other company to build an APU which runs PS4 and PS4 Pro games :)
BG's, Red Tech gaming, plus all the "insiders" from the two forums.
The other interesting note is that there are two design styles within the 7N/7NP family: a high density and a high efficiency design.
BG's, Red Tech gaming, plus all the "insiders" from the two forums.
I think high efficiency makes more sense from a clock and performance standpoint.The other interesting note is that there are two design styles within the 7N/7NP family: a high density and a high efficiency design.
So the high efficiency design is larger in general. But who knows what the 5700 UC was based on....?
We don't talk about the old forumI don't think you should trust any of the insiders from the old forum.
This is how I honestly see all this...
On one hand, we have insiders saying what they are saying. And everything every insider is saying suggests that both machines are close. Some say PS5 is stronger others say its the XSX. The operative word here being "close".
On the other hand, we have these leaks. And whether we choose to admit it or not, those leaks are all without context. Which literally makes them open to interpretation. I am not doubting Eras detective/technical chops...but this has to be the greatest display of tunnel vision I have ever seen. I mean folks could end up being right, but it would just be outta pure luck, and not because there is anything conclusive on what shaped their theories.
Long story short, we can speculate and theorize all we want, but we don't know shit.
1st rule of forum club
The most economical thing to do with a fixed die size is to clock the piss out of it. Given wafer costs, there's a logic to pursuing high clocks. They control the thermal and power delivery completely. They're not beholden to arbitrary chassis design, standard slot sizes, or ATX power standards.
- 36 cu at 2.0 ghz would be a 150w+ gpu. add 40-50w for the cpu and their apu alone is 200w.
- if their goal was to hit 9.2 tflops, they would get there for less than 100w if they went with 1.35ghz and 54 CUs.
- so to save 50mm2 or $10-15, they decided to take a cool a 200w apu that cant even hit 10 tflops.
its hilarious to me that the team that was supposedly team realistic and team thermals for over a year is now putting all their chips in the oberon 2.0 ghz basket. i can actually see sony going with a 9.2 tflops 1.35ghz 54cu gpu. that to me at least makes some sense even if it doesnt line up with other rumors from within the industry. because even then it would be a 150w apu that would require some kind of crazy cooling solution considering the entire ps4 pro consumed 150w.
with n7p and other arch improvements that result in 20% power savings, i can see 2.0 ghz being somewhat viable at around 120w for the gpu alone. but the same should apply to the 54cu configuration which would be under 80w for the same amount of tflops.
i wouldnt be surprised if the oberon 2.0 ghz tests are for BC with some kind of native boost mode that runs all pro games at native 4k 60 fps while disabling the remaining CUs and using the power left from disabling those CUs into increasing the clocks on the 36CUs. sony was afraid to put boost mode out at launch, but the mere 111 mhz clockspeed boost didnt cause any issues on ps4 games with no pro support. now imagine what they can do with 1,200 mhz with the same amount of CUs.
One look at the dev kit design should tell you they pushing clocks.The most economical thing to do with a fixed die size is to clock the piss out of it. Given wafer costs, there's a logic to pursuing high clocks. They control the thermal and power delivery completely. They're not beholden to arbitrary chassis design, standard slot sizes, or ATX power standards.
From spanish forum
Developers already have the latest dev-kits. Talking about raw-power, consoles are like PS3 vs 360 (really close) and there is no gap like PS4-ONE.
GPU is more powerful on Series X (only a little bit)
CPU is same PS5 and Series X
Other stuff is better on PS5 (I think he is talking about the SSD...)
The controller is really cool but almost same as PS4
The source is working on a big AAA, there was rumors about that AAA but it was claimed to be fake (but it's not, it's real. Idk what game can be)
The guy can not show any proof so....one rumor more.
my point is that whatever savings they might get from a smaller die will be offset by the need to have a more elaborate cooling solution.The most economical thing to do with a fixed die size is to clock the piss out of it. Given wafer costs, there's a logic to pursuing high clocks. They control the thermal and power delivery completely. They're not beholden to arbitrary chassis design, standard slot sizes, or ATX power standards.
I am just ready for some amazing $499 machines with NVMe SSD, Modern CPU, ~10 TF GPU with next gen RDNA extras....
As much as the GitHub details seem conclusive, I can't help questioning the notion of a 36cu chip.
PS4 had 18, Pro doubled the CU count to 36. The X came out a year later with 40CUs and a clock increase.
But Sony's plan for PS5, 4 years after the Pro, 4 years after specs for the X were revealed, with all their BC patents, is a 36 Navi CUs clocked ridiculously high?
The only way it makes sense is if Sony and AMD absolutely couldn't develop any way to increase CUs while preserving Pro BC. I have to imagine they've been considering Pro BC since they started work on the Pro. So that's like 7 years? And couldn't crack it?
Basically, yeah. And I'm not saying it's not a possibility, not at all. But I find such lack of skills a little hard to believe coming from Cerny and his engineering team. Or maybe they had a ridiculously small budget for development of the PS5 and they did their best? Even harder to believe, but yeah, I guess it's possible. Maybe the PS5 is indeed somehow a $399 machine and a "niche product" for the hardcore, all at the same time? Yeah, maybe, it could be I suppose 🤷As much as the GitHub details seem conclusive, I can't help questioning the notion of a 36cu chip.
PS4 had 18, Pro doubled the CU count to 36. The X came out a year later with 40CUs and a clock increase.
But Sony's plan for PS5, 4 years after the Pro, 4 years after specs for the X were revealed, with all their BC patents, is a 36 Navi CUs clocked ridiculously high?
The only way it makes sense is if Sony and AMD absolutely couldn't develop any way to increase CUs while preserving Pro BC. I have to imagine they've been considering Pro BC since they started work on the Pro. So that's like 7 years? And couldn't crack it?
right, and while i trust matt, i have no idea what he means by that. is it 100% confirmation? no. is it mostly/partially right? is it only part of the picture (of course)? who knows, its purposely vague. also, i dont know the context of his info.
Basically, yeah. And I'm not saying it's not a possibility, not at all. But I find such lack of skills a little hard to believe coming from Cerny and his engineering team. Or maybe they had a ridiculously small budget for development of the PS5 and they did their best? Even harder to believe, but yeah, I guess it's possible. Maybe the PS5 is indeed somehow a $399 machine and a "niche product" for the hardcore, all at the same time? Yeah, maybe, it could be I suppose 🤷
Hi all, I am a new user that has been lurking on here for a long time.
I have been watching you all twisting about and trying to figure out how all of these specs makes sense.
I am speculating here. I am not an insider, but I would like to take a crack to solve the puzzle.
I believe that everyone is correct on specs. How is this possible? Well... This is how it can be all possible at once...
Question: Why the different specs?
Answer: There are development kit specs and there are final retail specs. The development kits that are 12 TF are because it uses an older chipset (VEGA?) before NAVI came out and these are the development kits that are currently available to third parties. The 12 TF are actually GCN TF, not NAVI TF.
So you have 12 TF GCN which is equal to that of 9 TF NAVI. So, the insiders are getting third party development kit information from other people telling them that it's close to Xbox, which in raw numbers it's true, but the real truth is that those are GCN flops not NAVI flops, so the final retail console will be around 8-9 NAVI TF.
Sony's first party is always going to have the latest development kits by the way, later this year they will all (first and third party) have the final development kit available to everyone.
Sony is doing this to keep their "real specs" private for security reasons.
So, people are right, you still might get the same information from GDC. :-)
This all makes sense for both the insiders as well as the GitHub leaks. So, Klee and the rest of the insiders are not lying, they are telling the truth, but they just don't have all the details to bring to the table to see the entire big picture. The GitHub leak is true and the original consumer target was 9 TF, we will see if they hit their target or not.
Question: Okay, that explains that very well, but what about the part about ray-tracing?
Answer: The PS5 will have hardware ray-tracing no doubt. Software ray-tracing is done via the CPU using software. Mark Cerny said it was hardware but we don't see it in the github leaks. Well, that's because it's a built in feature of the GPU and not a separate feature.
Sony is going to "probably" be using AMD"s own brand of ray-tracing and it's built inside of the hardware of the GPU and there will be drivers that are written for this to use this hardware.
Microsoft has been doing ray-tracing research via Microsoft Azure when they were doing that cloud research for the Xbox One. This was mentioned in a tweet from Phil Spenser in 2015, so they have years and years of experience before they started writing DXR and they also worked with Nvidia to get their solution out so that Microsoft could start getting developers building on ray-tracing games for the next Xbox.
My speculation says that Microsoft will have dedicated ray-tracing cores in hardware that are of course dedicated towards real-time ray-tracing.
So, if I am correct, both will offer hardware ray-tracing but Microsoft will go one step further and offer dedicated ray-tracing hardware cores that are not part of the GPU hardware itself but dedicated for those kinds of tasks like that of Nvidia.
If you want to play games on the PC, I would recommend Nvidia's new Ampere chips coming out this year, it should have some of the newer tech for ray-tracing (2nd generation tech) that will probably be in the next generation Xbox (again, I am all speculating here).
Anyway, I really won't be posting all that much unless I have something to really say. As I said, all speculation of course, but something to think about at least huh?
Thank you for allowing my opinions here in this community. :-)
My internet soul 100 acre wood friend! Also sorry it's so late of a responseThis seems like a reasonable description of most insider YouTubers.
Also I love a good normalized relational database :)
This is how I honestly see all this...
On one hand, we have insiders saying what they are saying. And everything every insider is saying suggests that both machines are close. Some say PS5 is stronger others say its the XSX. The operative word here being "close".
On the other hand, we have these leaks. And whether we choose to admit it or not, those leaks are all without context. Which literally makes them open to interpretation. I am not doubting Eras detective/technical chops...but this has to be the greatest display of tunnel vision I have ever seen. I mean folks could end up being right, but it would just be outta pure luck, and not because there is anything conclusive on what shaped their theories.
Long story short, we can speculate and theorize all we want, but we don't know shit.
my point is that whatever savings they might get from a smaller die will be offset by the need to have a more elaborate cooling solution.
My internet soul 100 acre wood friend! Also sorry it's so late of a response
As many as it needs to be to render The Lion King remake graphics in real time