• 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.

Next-Gen PS5 & XSX |OT| Console tEch threaD

Status
Not open for further replies.

FeiRR

Banned
Sadly, it seems to be blocked by Sony associating with our internet provider. Seems like our internet provider signed an NDA.
:messenger_anxious:
Try VPN or let me know which ones you want. I can help :messenger_beaming:

BZGoqan.jpg
 
Some bits to add...

- All Oberon test in GitHub are regression tests for Ariel... that means they are making sure everything that worked in Ariel works in Oberon... it is weird but there is no Oberon profile tests leaked.

- Sony doesn’t produce the chips... TSMC does it by AMD request, Sony only buy them from AMD.

- 7nm EUV is included in AMD 7nm.... so everything is possible here.

- What makes find weird with AMD probably not using the 7nm EUV is the fact they said 50 perf. per watt. AMD already have a big hint when they moved from GCN 7nm to RDNA 7nm so they already had 50% increase in the same node. They want more 50% that continuing in the same DUV? That means 75% due architecture improvements only? I don’t buy that... nVidia did the miracle Maxwell without using a new node they got 30% increase in perf. per watt.
So I can only see that 50% increase over RDNA with a more efficient node and that why they didn’t discard 7nm EUV... they just said everything is 7nm for them DUV or EUV.

I maybe forget something to add but my 10 months old son wake up ;)

Yes, I'm hoping some Oberon profile tests come out relatively soon, because it is odd and that deserves to be questioned. Sony and MS got out of the chip production game a long time ago but I think some people still believe they manufacture and assemble the chips for these machines in-house and on their own fab plants. But those days have been gone a long time ago. They defiitely architect the systems in-house and work with engineering samples in-house (plus make requests to AMD to support certain features in silicon for their designs), but the actual manufacturing and fabbing isn't done by them.

I kind of see AMD's cheeky wording regarding RDNA2 (and RDNA3) as basically them saying RDNA2 will have DUV and EUV products (as you've mentioned here), and that they are working on moving quickly to 5nm for RDNA3. So I guess that kind of makes 7nm EUV a bit of a stopgap for them but the question is how much so and for how long. There's still financial reasons however why I think both consoles are 7nm DUV process-wise, though.

We don't know a lot on AMD's wording when they say the "+" regarding 7nm is in terms of further process refinements. Whatever it is, I'd suspect the 7nm DUV the consoles are most likely using are both RDNA2 and somewhat more efficient than earlier 7nm DUV GPU cards from AMD, but not necessarily at 7nm EUV levels. And they might be hyping up some of their numbers just a tad in terms of perf-per-watt increase of RDNA2 over RDNA1. Either that, or the 50% claim is the high-end for chips on the EUV process, and it'd be lower for those on the DUV process.

But go back and take care of little junior there. Heck he might have the answers we're all looking for!
 

HerjansEagleFeeder

Gold Member

With all due respect, FuCk China! I have a few Chinese friends and they all agree this situation is a shit show mostly created by their irresponsible communist government. If they had asked the world for help as soon as the epidemic started things probably wouldn’t as bad as they are. Goddamn bat and pangolin eaters.

Only a very small minority of the Chinese eat wild animals. Most find it just as disgusting as you and me. Totalitarian regimes have a tendency to try to hide incidents like this until its no longer possible because shits already hitting the fan left and right. Remember how USSR dealt with Chernobyl in the beginning? I don't think our Western countries would have been better prepared for this if China gave us a few weeks more time to contemplate the outbreak. It seems to me at least that all of Western Europe underestimated the problem.
 

Bo_Hazem

Banned
3 cured? Tell the hospital there to send me some of that cure lol.

Forgot to say that when I had H1N1 for the first time I felt like dying, it was insane and lasted like 3-4 weeks with the first early days being insanely hard. Second time was still tough but a lot less powerful. Now the H1N1 is a resident virus in my region and many have dealt with it with no direct vaccine. Would say have like 2-4 showers a day, eat honey and try to eat good and you'll be fine.
 

Hatsuma

Member
Only a very small minority of the Chinese eat wild animals. Most find it just as disgusting as you and me. Totalitarian regimes have a tendency to try to hide incidents like this until its no longer possible because shits already hitting the fan left and right. Remember how USSR dealt with Chernobyl in the beginning? I don't think our Western countries would have been better prepared for this if China gave us a few weeks more time to contemplate the outbreak. It seems to me at least that all of Western Europe underestimated the problem.

Yes unfortunately. Chinese government is a.. piece of **** for concealing this until it grew out of control. And it is unfortunate that the governments have been so slow to react and lack proper countermeasures for this. This whole event has shined a bright light on the incompetence in some of our governments.
 
Last edited:

gspat

Member
Doesn't seem to be spreading in the hot equatorial climates. There have not been major outbreaks in the southern hemisphere.
Experts believe, like influenza it survives longer in cold weather.

Edit: on another note they are shutting the port of Los Angeles, the port will get way behind but should be caught up by this summer. Them union guys love that OT.
Not so sure about that, the only reported case near me is someone that just got back from Egypt. That's pretty equatorial.
 

Bo_Hazem

Banned
I do wonder once we have the exact specs of the PS5, if we will still be discussing the github leak.

Sony have officially confirmed hardware ray tracing yet they still cling to it so hard. Now we are debating if Brie Larson's butt is Flat vs Spherical.
 
At least his posts make sense, and while his points may not always be something you agree with, he offers good conversational points and is factual in his discussion, while still trying to remain somewhat balanced.

He’s not... like... other posters........
Makes me realize I'm sane. I knew I wasn't just a "fanboy" (god I hate that term) . It really is people being complete asses with an agenda for PS5 to be less than. I have no problems with others opinions but damn, confusion doesnt come from GOD for sure. Thanks again all of you true Gafers
 

Bo_Hazem

Banned
Plus eat two lemons a day one in the morning and one in the evening

Well that's good for vitamin C, but you can shortcut that and eat hot chili as one packs around 110% of daily vitamin C. Just try to have milk, yogurt, or more oily food with it as it can backfire if you're not used to spicy food. It doesn't need to be hot ones anyway, or you can go with 1 paprika that packs around 170% of daily need of vitamin C.

I like their honesty here:

1918__94628.1562998566.jpg
 

Null_Key

Neo Member
I often wonder about people in this thread and other gaming website panicking more from Coronavirus pandemic or lack of news from Sony about PS5. It hilarious that IGN and other garbage website having catastrophic meltdown over lack of info from Sony, over a virus that is causing mass hysteria.
 

Lng0004

Member

With all due respect, FuCk China! I have a few Chinese friends and they all agree this situation is a shit show mostly created by their irresponsible communist government. If they had asked the world for help as soon as the epidemic started things probably wouldn’t as bad as they are. Goddamn bat and pangolin eaters.
Yes, and the US govt is doing much better job.
 

thelastword

Banned
Because it's pastebin. Anybody can post any BS there.
That doesn't mean anything, it's leak country out there...….The right leak can come out of anywhere, yet I see people have tendency to believe leaks from Githaub, twitter and reddit, like it's no tomorrow.....Some people even swear on GitHub for a particular reason......
,
I'll be too. $799 16TF 24GB PS5 pro.
If base is 15TF, I'm willing to pay $800 for 18TF with the same 32GB Vram, so some extra power, but no difference in Vram just as PS4 was to PRO....Maybe this can aid 120fps and 8K games a bit more on the PRO version....

He said one of the pastebins was pretty accurate, so no.
One of the pastebins is 15TF, so that's on the cards too surely? This is going to get interesting in a minute.....
 
I am still believing the final GPU of the PS5 has 54 CU (18 ps4, 36 ps4 and test of github) clock to 1.8 Ghz which is possible to clock to 2 Ghz for few periods of time if the temperature is good enough.

While the XSX has 56 CU but the clock cannot be so hight as 2 Ghz but not should be necessary, the thing I dont understand if the RDNA use the AMD patent for
raytracing the shader unit of the same GPU.

How insiders like OsirisBlack OsirisBlack told us the raytracing was better in the PS5 if is looks possible the new PS5 has less CU? that means the way of who the patent
works is better with a high clock speed or PS5 has more Shader units by CU or the use amd technology working with a third company which has been mention
in last pages?
 

devilNprada

Member
Not so sure about that, the only reported case near me is someone that just got back from Egypt. That's pretty equatorial.

Purely anacdotal... Look up the numbers for Egypt which has a lot of tourism and it's again not a big outbreak... 110 cases of 100 million people.
 
That doesn't mean anything, it's leak country out there...….The right leak can come out of anywhere, yet I see people have tendency to believe leaks from Githaub, twitter and reddit, like it's no tomorrow.....Some people even swear on GitHub for a particular reason......
,

If base is 15TF, I'm willing to pay $800 for 18TF with the same 32GB Vram, so some extra power, but no difference in Vram just as PS4 was to PRO....Maybe this can aid 120fps and 8K games a bit more on the PRO version....


One of the pastebins is 15TF, so that's on the cards too surely? This is going to get interesting in a minute.....

15TF is possible but not likely. I think that the critical number is 13. A 13 TF PS5 would have over seven times the TF of the PS4 and would have over three times the TF of the PS4 Pro. My guess is that the PS5 will be somewhere between 13 and 14TF.

Combine that with the SSD and the better CPU and you have a very competent system.
 
So can we assume the following:

- The source of WhispersintheWind a is legit PS5 dev.
- The source debunks, once again, the PS5 9.2tf ficction.
- WhispersintheWind shared his well-funded "opinion" after talking with a legit PS5 dev.
- We dont know yet what leak the source points as correct.
- WhispersintheWind has promised that he will share more info if his source replies back.

The idea the PS5 would be 9.2 is ridiculous. What people need to get through their heads is that the GPU in GitHub is not the GPU that will be in the PS5. However, this concept is too difficult for a LOT of people.
 

gspat

Member
Purely anacdotal... Look up the numbers for Egypt which has a lot of tourism and it's again not a big outbreak... 110 cases of 100 million people.
I'm sure that will make the guy that caught it from Egypt feel much, much better.

A friend of mine is flying down to Cuba in a couple days. They just got confirmed infections from a group of Italian tourists.

I'm not worried for me, I'm worried for my friend.

On a side note, the store I work at just got cleaned out of toilet paper two days straight. People are idiots. It's almost like they are buying TP instead of food.
 
Makes me realize I'm sane. I knew I wasn't just a "fanboy" (god I hate that term) . It really is people being complete asses with an agenda for PS5 to be less than. I have no problems with others opinions but damn, confusion doesnt come from GOD for sure. Thanks again all of you true Gafers

Why would someone intentionally promote a fake agenda for PS5? You can spread as many lies as you want about widget X, at the end of the day widget X gets released as whatever it is, regardless of the lies spread. Seems pointless.

Sharing ideas that are likely incorrect is something different, as a laymen, I post ideas that probably make the devs on the board laugh. It is what it is, but it's not part of some kind of plot to spread misinformation. I'm sure that's true for a lot of the posters here. Don't hate on the uninformed too bad. LOL
 

POTUS

Banned
I am way way way more interested in Low latency wifi direct for PS VR2 than I am in the the PS5 Teraflop number.

I hope ps5 has a low latency direct wifi chip built in that can stream 4k resolution wirelessly to psvr2.

A wireless psvr2 similar to the oculus quest that gets the image from ps5 would be amazing.

Pair dual 4k screens with surround sound and haptic controls and you can have vr experience that are damn near life like simulations.
 
Last edited:

POTUS

Banned
I already had a valve index but the oculus quest is the best gaming purchase I made in years. Wireless is a breath of fresh air. Just wish the screen was higher resolution but beat saber is amazing as are vr 180 degrees videos that let me be whereever i want and see whatever i want.

The ps5 and psvr2 need wifi direct and 4k displays and it would be THE killer app.
 
Last edited:

Bo_Hazem

Banned
I can't speak for others who utilize Github and the testing data as a reference point, but I'll answer what I can in your post.

-I always thought the initial testing data was doing BC testings for PS4 and Pro. This was even before I learned about the Ariel iGPU testing profile and therefore meant even well before bringing that up into discussion, I felt that testing data was only reflective of BC settings and not the full actual chip

-The 36/40 is either the full Oberon CU counts, or a count that's incomplete. It's not the fault of the data miners why that particular question can't be clearly answered; you'd have to ask Sony and AMD staff running these tests and storing the data about that.

-You are actually misinterpreting what RDNA2 actually is. RDNA2 isn't exclusively 7nm EUV. AMD's wording clouded conclusions on that but they shortly clarified after the conference that it was a nomenclature thing, so 7nm"+" can be thought of more as further improvement on the 7nm node process

-Another important thing to keep in mind is that RDNA2 does not necessarily mean "large chip" out of necessity. AMD have never stated this, not even once, and they never will. Why would they keep RDNA2 features like RT, VRS, VRR etc. off their mobile APUs, which have very low CU counts compared to these consoles? Some people have automatically made the correlation when that is just them jumping through an unnecessary hoop. It just so happens there will be chips on RDNA2 that are "big" chips, but RDNA2 features can be implemented on smaller chips as well. Otherwise, AMD have completely failed in designing a smart, scalable GPU architecture

-There's nothing inherently stopping Sony from going with a bigger chip, but that doesn't mean a bigger chip was their goal, either. I actually speculated why there's a chance they've gone with a relatively smaller chip, but there are also cost-related reasons, plus the fact Sony is very likely going to manufacture a very big chunk of units for the first fiscal quarter (I would guess at least 10 million) and then you have to weigh all of that against the size of Sony as a corporation itself and what money they have on hand (and what can be borrowed through big loans from banks or investors) to cover those costs. That alone could have led them down a path of wanting a relatively "smaller" chip.

-You're right, pretty much anyone speculating Oberon is an RDNA2 chip is doing so with information outside of Github. But for myself personally, I've been using a swell of other sources besides that in my own speculation for a long time now, and more to the point, it's not like most of the insiders really know what the PS5 is, either. If so, we could actually get some more nuanced specifications beyond simply a TF range. And when you have guys like Tommy Fischer commit reputation suicide by almost duping a mod with doctored controller photos, that just puts an extra bit of questioning on other rumors as a whole: even when pressed with proof some of these dudes may try bullshitting their way to the top because the balls are just that brass.

-I never questioned PS5 having hardware ray-tracing, not once. But it wasn't just Githubbers speculating about PS5 being RDNA1; many people in general were. Many were under the assumption Sony had their own RT ASIC solution either integrated into the APU as an IP block (some went as far as to bring up a team member having worked on PowerVR as Sony maybe using PowerVR RT instead of AMD's), or connected to the APU, likely with some PCIe interconnect or something to that effect. And that was mainly Sony people speculating this. Even I did for a while but once it became apparent how inefficient a separate RT ASIC would be to the design (bandwidth issue, potentially) I just accepted that PS5 would have AMD's RT in one form or another, and that's while still relating Oberon to PS5. In other words, I dropped PS5 being RDNA1 a long time ago and that was pretty much at the time Cerny clarified things and when the RT ASIC idea fell through.

-Gotta keep in mind that a lot of us were also uncertain about RDNA2's node process. I assumed both PS5 and XSX were RDNA1 because I thought they would be 7nm. In fact, given AMD's recently said 7nm+ isn't necessarily EUV but rather improvements on the 7nm process, it means almost for certain PS5 and XSX are still 7nm but with some efficiency gains (just not EUV levels). I even doubted XSX was fully RDNA2 even when Microsoft themselves mentioned RDNA2 in that Twitter post, because I assumed RDNA2 was exclusively 7nm EUV and knew that would be even costlier than regular 7nm.

So basically, I think PS5 and even XSX will still be 7nm, but be RDNA2 chips. That should be pretty clear-cut.

-I think people are overestimating RDNA2 efficiency. The 50% perf-per-watt doesn't actually mean 50% IPC over RDNA1. And we now know RDNA2 is not exclusive to EUV, and the "7nm+" is actually indicative of small improvements over the 7nm process, notable improvements but not to the level of what actual 7nm EUV brings. So if anything it suggests both PS5 and XSX are 7nm but benefit from slight efficiency improvements while not being on the EUV process.

-Wasn't the interpretation of the testing data always meant to indicate BC of PS4 and PS4 Pro? Even with that said it asks the question of why 2GHz needed to be tested for that except to perhaps ensure PS4 Pro titles could hit native 4K60 natively on the chip. Probably a question worth asking and one we'll hopefully get an answer for soon (through Sony themselves).

-So I guess that brings us back to the high clock. You ask why would they do it, well, I'm still asking that too. I have always said Github and the testing data do not represent everything in terms of PS5's performance, not even once. However, I think with clarification on what RDNA2 actually is and the fact it can be on both 7nm and 7nm EUV (and the "7nm+" was apparently more about minor improvements on the 7nm process, probably through things AMD have done themselves on the architecture pipeline side) that there might be some justification for the 2GHz clock testing on Oberon outside of acting as a free FPS stability boost for Pro-enhanced games.

What would justify that? Well, the easiest answer is that Oberon might be a 36/40CU chip after all. Or, it could be a slightly larger chip (the Dec revision adjusted the memory bus controller IIRC, upping it higher) but still getting a relatively high clock. That's why I think maybe the best answer to what PS5 actually is (for now) is to go somewhere in the middle. A 44 - 48 active CU chip on a more recent revision, pushing relatively high clocks (anything between 1800MHz and 1900MHz if 2000MHz can't be sustained) might fit that bill.

But I can't have a grounded speculation like that by simply considering one source type (insiders) as gospel and the other (pertinent & persistent testing data) as throwaway junk. Unfortunately a lot of the people who say things like "Team Github" are doing it unironically, so as to discredit anyone even partially referencing the Github stuff or any of the datamined testing data whatsoever, for any purpose at all. That type of reasoning is nothing short of insanity.

I also think it's because we really do have a group of people who are desperate for a power crown battle at any costs, unwilling to even consider the worth or validity of any source that isn't for 'their team" coming out on top in a proverbial e-penis TF pissing contest, and assume anyone and everyone who refers to anything besides insiders exclusively as being against their preferred brand or "team". The truth is you may get that with literally a very small minority of people but the vast majority of that type aren't here on these forums.

Yes it's unfortunate people with actual troll-like behavior in the past such as TimDog (and even some people here like Longdi who I hope has calmed that type of posting down) are clinging to the Github 9.2TF stuff to antagonize PS fans, but I'm not TimDog and TimDog is not representative of the majority of us who simply see it as a valuable asset in discussing the possibility of next-gen PlayStation. Hell, most of us are even doing so while still listening and giving consideration to the rumors as well, it doesn't have to be an either/or situation.

In the end, I ultimately agree with your end point: Github doesn't provide the full picture on PS5. It didn't provide the full picture on XSX either and barely provides anything on Lockhart (speaking of which, I'll actually take a minute here and say how funny it was to see some Xbox people try arguing the 22/24 CU APU leak that very clearly listed Microsoft in it, with a 950MHz TDP-down, was in fact not related to Lockhart. It took until DF did a video simulating a 4TF Lockhart running games at 1080p and mostly 60FPS clocks for them to accept that particular leak. However I took that leak to pretty much be Lockhart upon first seeing it and it also has me thinking Lockhart is not a "traditional" game console, but that's something for another time).

However, there're people who will take anyone merely stating simply that, and interpret it to mean "Github is useless. The testing data is useless. None of these chips are PS5 related at all", and some of them know what they're doing when they do this. So IMHO, they're just as guilty as having an agenda in dismissing Github as some people are in using Github and the testing data to antagonize Sony and PS fans (and pushing 9.2TF like it's a derogatory insult). It goes both ways on this, and as much as I don't want to give the fanatical, trolling Xbox fanboys ammunition (in needing to clarify things on the Github info and testing data from my own POV), I don't want to give the fanatical, trolling PlayStation fanboys ammunition in attacking all aspects of that info and testing data as being FUD that only folks with an agenda against Sony and PS would bother bringing up in the first place, because for the vast majority of us that is NOT the case whatsoever.

I just simply talk about what I observe, and I try to observe from multiple sources, and be fair in doing so to see where they can meet in the middle and agree upon. However, I can't help if one of those source types has data on its side that I can look at, check up on/research, verify on my own (in terms of following any coding the producers of the chips the data comes from would follow), and itself has been very consistent. And I can't help if that comes off to some people as me "choosing" one side over another when that isn't me doing so out of some emotional need or negative bias towards a preference. And I honestly think most of the other people who regard the Github info and testing data with at least some weight would feel similar.

I just gave you a like for how long that essay is. Are you a writer? Do you have a book? That's some serious typing diarrhea :messenger_medmask::messenger_winking_tongue: (joking)

source.gif
 

icerock

Member
I can't speak for others who utilize Github and the testing data as a reference point, but I'll answer what I can in your post.

-I always thought the initial testing data was doing BC testings for PS4 and Pro. This was even before I learned about the Ariel iGPU testing profile and therefore meant even well before bringing that up into discussion, I felt that testing data was only reflective of BC settings and not the full actual chip

-The 36/40 is either the full Oberon CU counts, or a count that's incomplete. It's not the fault of the data miners why that particular question can't be clearly answered; you'd have to ask Sony and AMD staff running these tests and storing the data about that.

-You are actually misinterpreting what RDNA2 actually is. RDNA2 isn't exclusively 7nm EUV. AMD's wording clouded conclusions on that but they shortly clarified after the conference that it was a nomenclature thing, so 7nm"+" can be thought of more as further improvement on the 7nm node process

-Another important thing to keep in mind is that RDNA2 does not necessarily mean "large chip" out of necessity. AMD have never stated this, not even once, and they never will. Why would they keep RDNA2 features like RT, VRS, VRR etc. off their mobile APUs, which have very low CU counts compared to these consoles? Some people have automatically made the correlation when that is just them jumping through an unnecessary hoop. It just so happens there will be chips on RDNA2 that are "big" chips, but RDNA2 features can be implemented on smaller chips as well. Otherwise, AMD have completely failed in designing a smart, scalable GPU architecture

-There's nothing inherently stopping Sony from going with a bigger chip, but that doesn't mean a bigger chip was their goal, either. I actually speculated why there's a chance they've gone with a relatively smaller chip, but there are also cost-related reasons, plus the fact Sony is very likely going to manufacture a very big chunk of units for the first fiscal quarter (I would guess at least 10 million) and then you have to weigh all of that against the size of Sony as a corporation itself and what money they have on hand (and what can be borrowed through big loans from banks or investors) to cover those costs. That alone could have led them down a path of wanting a relatively "smaller" chip.

-You're right, pretty much anyone speculating Oberon is an RDNA2 chip is doing so with information outside of Github. But for myself personally, I've been using a swell of other sources besides that in my own speculation for a long time now, and more to the point, it's not like most of the insiders really know what the PS5 is, either. If so, we could actually get some more nuanced specifications beyond simply a TF range. And when you have guys like Tommy Fischer commit reputation suicide by almost duping a mod with doctored controller photos, that just puts an extra bit of questioning on other rumors as a whole: even when pressed with proof some of these dudes may try bullshitting their way to the top because the balls are just that brass.

-I never questioned PS5 having hardware ray-tracing, not once. But it wasn't just Githubbers speculating about PS5 being RDNA1; many people in general were. Many were under the assumption Sony had their own RT ASIC solution either integrated into the APU as an IP block (some went as far as to bring up a team member having worked on PowerVR as Sony maybe using PowerVR RT instead of AMD's), or connected to the APU, likely with some PCIe interconnect or something to that effect. And that was mainly Sony people speculating this. Even I did for a while but once it became apparent how inefficient a separate RT ASIC would be to the design (bandwidth issue, potentially) I just accepted that PS5 would have AMD's RT in one form or another, and that's while still relating Oberon to PS5. In other words, I dropped PS5 being RDNA1 a long time ago and that was pretty much at the time Cerny clarified things and when the RT ASIC idea fell through.

-Gotta keep in mind that a lot of us were also uncertain about RDNA2's node process. I assumed both PS5 and XSX were RDNA1 because I thought they would be 7nm. In fact, given AMD's recently said 7nm+ isn't necessarily EUV but rather improvements on the 7nm process, it means almost for certain PS5 and XSX are still 7nm but with some efficiency gains (just not EUV levels). I even doubted XSX was fully RDNA2 even when Microsoft themselves mentioned RDNA2 in that Twitter post, because I assumed RDNA2 was exclusively 7nm EUV and knew that would be even costlier than regular 7nm.

So basically, I think PS5 and even XSX will still be 7nm, but be RDNA2 chips. That should be pretty clear-cut.

-I think people are overestimating RDNA2 efficiency. The 50% perf-per-watt doesn't actually mean 50% IPC over RDNA1. And we now know RDNA2 is not exclusive to EUV, and the "7nm+" is actually indicative of small improvements over the 7nm process, notable improvements but not to the level of what actual 7nm EUV brings. So if anything it suggests both PS5 and XSX are 7nm but benefit from slight efficiency improvements while not being on the EUV process.

-Wasn't the interpretation of the testing data always meant to indicate BC of PS4 and PS4 Pro? Even with that said it asks the question of why 2GHz needed to be tested for that except to perhaps ensure PS4 Pro titles could hit native 4K60 natively on the chip. Probably a question worth asking and one we'll hopefully get an answer for soon (through Sony themselves).

-So I guess that brings us back to the high clock. You ask why would they do it, well, I'm still asking that too. I have always said Github and the testing data do not represent everything in terms of PS5's performance, not even once. However, I think with clarification on what RDNA2 actually is and the fact it can be on both 7nm and 7nm EUV (and the "7nm+" was apparently more about minor improvements on the 7nm process, probably through things AMD have done themselves on the architecture pipeline side) that there might be some justification for the 2GHz clock testing on Oberon outside of acting as a free FPS stability boost for Pro-enhanced games.

What would justify that? Well, the easiest answer is that Oberon might be a 36/40CU chip after all. Or, it could be a slightly larger chip (the Dec revision adjusted the memory bus controller IIRC, upping it higher) but still getting a relatively high clock. That's why I think maybe the best answer to what PS5 actually is (for now) is to go somewhere in the middle. A 44 - 48 active CU chip on a more recent revision, pushing relatively high clocks (anything between 1800MHz and 1900MHz if 2000MHz can't be sustained) might fit that bill.

But I can't have a grounded speculation like that by simply considering one source type (insiders) as gospel and the other (pertinent & persistent testing data) as throwaway junk. Unfortunately a lot of the people who say things like "Team Github" are doing it unironically, so as to discredit anyone even partially referencing the Github stuff or any of the datamined testing data whatsoever, for any purpose at all. That type of reasoning is nothing short of insanity.

I also think it's because we really do have a group of people who are desperate for a power crown battle at any costs, unwilling to even consider the worth or validity of any source that isn't for 'their team" coming out on top in a proverbial e-penis TF pissing contest, and assume anyone and everyone who refers to anything besides insiders exclusively as being against their preferred brand or "team". The truth is you may get that with literally a very small minority of people but the vast majority of that type aren't here on these forums.

Yes it's unfortunate people with actual troll-like behavior in the past such as TimDog (and even some people here like Longdi who I hope has calmed that type of posting down) are clinging to the Github 9.2TF stuff to antagonize PS fans, but I'm not TimDog and TimDog is not representative of the majority of us who simply see it as a valuable asset in discussing the possibility of next-gen PlayStation. Hell, most of us are even doing so while still listening and giving consideration to the rumors as well, it doesn't have to be an either/or situation.

In the end, I ultimately agree with your end point: Github doesn't provide the full picture on PS5. It didn't provide the full picture on XSX either and barely provides anything on Lockhart (speaking of which, I'll actually take a minute here and say how funny it was to see some Xbox people try arguing the 22/24 CU APU leak that very clearly listed Microsoft in it, with a 950MHz TDP-down, was in fact not related to Lockhart. It took until DF did a video simulating a 4TF Lockhart running games at 1080p and mostly 60FPS clocks for them to accept that particular leak. However I took that leak to pretty much be Lockhart upon first seeing it and it also has me thinking Lockhart is not a "traditional" game console, but that's something for another time).

However, there're people who will take anyone merely stating simply that, and interpret it to mean "Github is useless. The testing data is useless. None of these chips are PS5 related at all", and some of them know what they're doing when they do this. So IMHO, they're just as guilty as having an agenda in dismissing Github as some people are in using Github and the testing data to antagonize Sony and PS fans (and pushing 9.2TF like it's a derogatory insult). It goes both ways on this, and as much as I don't want to give the fanatical, trolling Xbox fanboys ammunition (in needing to clarify things on the Github info and testing data from my own POV), I don't want to give the fanatical, trolling PlayStation fanboys ammunition in attacking all aspects of that info and testing data as being FUD that only folks with an agenda against Sony and PS would bother bringing up in the first place, because for the vast majority of us that is NOT the case whatsoever.

I just simply talk about what I observe, and I try to observe from multiple sources, and be fair in doing so to see where they can meet in the middle and agree upon. However, I can't help if one of those source types has data on its side that I can look at, check up on/research, verify on my own (in terms of following any coding the producers of the chips the data comes from would follow), and itself has been very consistent. And I can't help if that comes off to some people as me "choosing" one side over another when that isn't me doing so out of some emotional need or negative bias towards a preference. And I honestly think most of the other people who regard the Github info and testing data with at least some weight would feel similar.

A very well written post, but I'm not sure if you're providing me any answers which I don't know? Or I didn't allude to in my original post? In fact, you're in agreement with pretty much everything I have written that Github provides us with very little if we perceive Oberon to be RDNA2.

When I talk about RDNA2, I'm limiting myself to the benefits arriving out of using that specific architecture. What node they are using or how expensive it is, I'm not concerned with, as AMD have been pretty muddy about the details. Also, when I say efficiency gains, I'm NOT talking about performance gains, simply performance per watts gain. Keeping that in mind considering this:

1) We know Sony (for reasons known to them) did manufacture an APU (Gonzalo), housing an iGPU based on architecture which is not going to be in PS5. (Ariel) So, it is very much possible, they have enough in the coffers, to spend on more iterations than we are privy to. Reason why I say is this, that many think Oberon is PS5 iGPU (I think so too), but there remains a small possibility, of a different PS5 APU, which hasn't been discovered. Just like Xbox Series S, we know its out there, but no one knows the details.

2) I never said RDNA2 automatically implies = bigger chip, I pointed out the efficiency gains (as in more performance for every watt) allows AMD to build a bigger GPU. This is pretty evident when you consider, that AMD put out a 9.7TF card last year, which never mind competing with an RTX 2080TI, doesn't even stack up to something like a RTX2070 Super. Why do you think they limited themselves to a 40CU product? For a whole year no less? One look at power consumption and heat dissipation of those cards will answer everything. AMD decided to bide their time and have a severely under-powered card in the market, while they worked to sort out bigger NAVI, which could only happen if efficiency per watts was improved. So, when I say RDNA2 allows one to build a bigger chip. I mean to say, it gives them an option to do so, not that it is mandatory for them. Cerny and his engineering team, might have very well decided to just push the clocks due to efficiency gains, instead of taking the radical route of redesigning the hardware by adding more CUs. Point is, we just don't know. Github doesn't help us get any of these answers.

3) Lastly, my main issue is with folks, whether knowingly or unknowingly, using Github to push something, which is never provided in there. Yes, the data is legit, but it has a lot of gaps, gaps which many knowledgeable people can't answer to. That vacuum is being filled by them, depending upon, what they perceive to be a PS5, not that of an actual PS5.
 

Jtibh

Banned
This thread should be on lockdown like all the countries at the moment.

Hundreds of pages later still no info on ps5 and there wont be any for a few months.
 

Bo_Hazem

Banned
At least his posts make sense, and while his points may not always be something you agree with, he offers good conversational points and is factual in his discussion, while still trying to remain somewhat balanced.

He’s not... like... other posters........

+ his avatars and username is something to agree with, at least.

No seriously, he's actually stuck a bit with the 36-40 and at best 48 CU's, but he always backs his speculations with good points wither you agree or not.
 
Last edited:

Bo_Hazem

Banned

With all due respect, FuCk China! I have a few Chinese friends and they all agree this situation is a shit show mostly created by their irresponsible communist government. If they had asked the world for help as soon as the epidemic started things probably wouldn’t as bad as they are. Goddamn bat and pangolin eaters.

GbPNc.gif


Relax, mate :messenger_tears_of_joy: :messenger_anxious:
 

01011001

Banned
15TF is possible but not likely. I think that the critical number is 13. A 13 TF PS5 would have over seven times the TF of the PS4 and would have over three times the TF of the PS4 Pro. My guess is that the PS5 will be somewhere between 13 and 14TF.

Combine that with the SSD and the better CPU and you have a very competent system.



you and the number 7 should get a room already!
 

ethomaz

Banned
Yes, I'm hoping some Oberon profile tests come out relatively soon, because it is odd and that deserves to be questioned. Sony and MS got out of the chip production game a long time ago but I think some people still believe they manufacture and assemble the chips for these machines in-house and on their own fab plants. But those days have been gone a long time ago. They defiitely architect the systems in-house and work with engineering samples in-house (plus make requests to AMD to support certain features in silicon for their designs), but the actual manufacturing and fabbing isn't done by them.

I kind of see AMD's cheeky wording regarding RDNA2 (and RDNA3) as basically them saying RDNA2 will have DUV and EUV products (as you've mentioned here), and that they are working on moving quickly to 5nm for RDNA3. So I guess that kind of makes 7nm EUV a bit of a stopgap for them but the question is how much so and for how long. There's still financial reasons however why I think both consoles are 7nm DUV process-wise, though.

We don't know a lot on AMD's wording when they say the "+" regarding 7nm is in terms of further process refinements. Whatever it is, I'd suspect the 7nm DUV the consoles are most likely using are both RDNA2 and somewhat more efficient than earlier 7nm DUV GPU cards from AMD, but not necessarily at 7nm EUV levels. And they might be hyping up some of their numbers just a tad in terms of perf-per-watt increase of RDNA2 over RDNA1. Either that, or the 50% claim is the high-end for chips on the EUV process, and it'd be lower for those on the DUV process.

But go back and take care of little junior there. Heck he might have the answers we're all looking for!
I’m tired :D
A bit because I have right now an infection that makes me a bit weak :D (it is not Corona).

Your last point is really interesting and probably what AMD will do... DUV in low/mid-end chips and EUV in high-end chips... it even makes sense why they are grouping both process as the name 7nm.
 

thelastword

Banned
15TF is possible but not likely. I think that the critical number is 13. A 13 TF PS5 would have over seven times the TF of the PS4 and would have over three times the TF of the PS4 Pro. My guess is that the PS5 will be somewhere between 13 and 14TF.

Combine that with the SSD and the better CPU and you have a very competent system.
Of course that'll be a fairly good system, but 15TF will light leagues of fire under asses for 7 years......If we were impressed with what Sony did for the last 7 years on 1.84TF......15TF will impress even more at the end of the gen, especially with that fast CPU and SSD in tow...…….Yet, no doubt, a 13-14TF GPU will not be contested in graphics either as long as PD, ND, Santa Monica, Guerilla are in SWWS...….
 

geordiemp

Member
I’m tired :D
A bit because I have right now an infection that makes me a bit weak :D (it is not Corona).

Your last point is really interesting and probably what AMD will do... DUV in low/mid-end chips and EUV in high-end chips... it even makes sense why they are grouping both process as the name 7nm.

When U read it I thought they would use EUV for the critical mask layers for the gates and fins and DUV for the rest. ie wherever the dimensions are smallest and critical and need precision.

You dont have to use one litho process for every layer....
 
Last edited:
Status
Not open for further replies.
Top Bottom