Support NeoGAF

[DeepEnigma]

I go to bed and it’s SX > PS5.

I wake up and it’s PS5 > SX.

I check now and suddenly it’s all...

 

[geordiemp]

What?? Microsoft are rumoured to be making two chipsets because they want to make two different consoles, how is that in any way relatable to Sony using two different chips for one console as per your assumption?

Where did it say I assuming Sony are making 2 different consoles.

Is basic reading that hard ?

I said explore, do you think Sony only looked at 1 chip design and called it a day ?

 

[ANIMAL1975]

First of all this

+ if oberon is related to the PS5 it also means they need to push it to 2Ghz to emulate what a 48CU's can deliver in graphics perf @1700Mhz, that's what explains the V Devs Kit (2Ghz is a lot), final product WILL NOT run @2Ghz.
- The Dual Chiplet theory is CRAZY, when i read it i was WOOOOOOOOOOOOOOO, the amount of power that would need is beyond 2GHZ, to communicate between chip to chip, travel (pikajul, i think that's the right name) to travel to a point to another needs more power.
-buying addons like a cooler will never happen.
-their business is to make it easier for consumers, not turning it into PC, they might make an easy SSD swap that's it.

I don’t know man adding 20 cu is much cheaper compared to dual gpu . We shouldn’t go all crazy yet haha

I will be repeating myself (already posted this before but there was no feedback), what if it is something in the line of a 3D stack layout approach, to reduce dye size and maintain the standard console form factor?

 

[Tenka Musou]

In order to hit the rumored 12~TF for PS5, how many CUs is needed, in theory? Assuming it's clocked at 2.0GHz.

 

[jroc74]

All evidence so far on how Sony handle BC is by directly mimicking the old hardware as close as possible. They did it with PS4 Pro, they have patents suggesting it, and then there is Oberon which does the same. PS4 to PS4 Pro was 18 to 36CU's with ability to disable half for BC.

One thing to consider is that they will also be planning ahead right now with how to do PS5 BC on either PS5 Pro or PS6. So quite possibly it's the same method. 36CU's to 72CU's on PS5 Pro / PS6. It's a straight forward tidy system of doubling the CU's and achieves their goals.

This is one of the theories. It makes sense.

But....that leaves the PS5.

18CUs for PS4, 18 more for Pro.

How about 18CUs?

That would give the PS5 54, a PS5 Pro 72., etc, etc.

This is one reason why to me github results don't make sense. Sony went to 36 CUs for the Pro. And just said the hell with it, we'll stay at 36....for their next gen baby?

Github is real, the results are real, but IMO it's not the entire picture. Or it could be. I just don't think it is.

Maybe Sony's BC patent would mimic the old hardware 100%. If MS and Sony do BC so different....isn't it reasonable that test results could also be different?

 

[SlimySnake]

In order to hit the rumored 12~TF for PS5, how many CUs is needed, in theory? Assuming it's clocked at 2.0GHz.

48

 

[ANIMAL1975]

“dual gpu can be terrible in pc but in a closed hardware optimized by creating faster connection buses than pci express and another story”




“as for the cooling system nn it is so difficult to design it just do consol to sendwich heatsink that occuba the part below and above the peripherals
nn and that there is a thermal scientist and then made in aluminum nn would cost a lot also because the efficiency of an aluminum heat sink nn and second to none with the fan put in the middle type ps3 solve everything”


L liviopangt , usa questo sito:

Google Translate

Google's service, offered free of charge, instantly translates words, phrases, and web pages between English and over 100 other languages.

translate.google.com

Deed you read the translation Deep? My brain hurts Google translate is worse than Stadia confirmed!

I don't think he understands your reply now since it's in English. Google translate would have helped him with that post had he wanted it to.

I think he understands (he joined the discussion), what ever translator he's using works for him, but then doesn't translate what he writes.

They dont have to release 2 consoles, but Sony would of been wise to EXLORE to power levels unless they knew the future costs of the chips in the 2 years or whatever cycle. Things change.

Do you think they would just look at 1 option ? mmmm.

What if that option was not as good as they though, do they blisfully carry on ?

Seems a hard concept to grasp by the looks of it lol

Even PS4 Pro (a mid gen refresh) was rumored to have 2 skus

Per favore, anche se l'inglese non è la lingua madre di molti qui intorno, il consenso è per tutti di usarlo.

Giuseppe Verdi

 

[bhunachicken]

48

So, that would mean to do PS4 Pro BC they'd need to disable 1/3 of the CUs, and for PS4 BC they'd need to disable 2/3 of them...?

 

[Gamernyc78]

Deed you read the translation Deep? My brain hurts Google translate is worse than Stadia confirmed!


I think he understands (he joined the discussion), what ever translator he's using works for him, but then doesn't translate what he rights.


Even PS4 Pro (a mid gen refresh) was rumored to have 2 skus


Giuseppe Verdi!

Even Google Translate has issues translating even Spanish and messes up certain tenses and words but it is quite good.

 

[DeepEnigma]

Deed you read the translation Deep? My brain hurts Google translate is worse than Stadia confirmed!


I think he understands (he joined the discussion), what ever translator he's using works for him, but then doesn't translate what he rights.


Even PS4 Pro (a mid gen refresh) was rumored to have 2 skus


Giuseppe Verdi!

I read it after and yes, it broke my cervello.

 

[Sinthor]

So, that would mean to do PS4 Pro BC they'd need to disable 1/3 of the CUs, and for PS4 BC they'd need to disable 2/3 of them...?

I think 54-56 CU's makes more sense. 2ghz is awfully hot to run in a closed form factor, no? Considering the cooling needed, wouldn't it then be cheaper to go with more CU's and lower clock speed?

 

[BGs]

Man, you're late, we already discussed Shu's answer to try to figure out if he was trolling with the typo or the date.

Sorry, I haven't had time to respond so far. I've been very busy at work these days. That is why my last answers have been so brief.

I saw Shu's answer and thought maybe it meant something else. That's why I published the date with a question. I didn't have time to look at the last pages to know if the subject had already been discussed. Too bad it's a joke, it would have been nice to know a specific date. Although who knows, Shu may surprise us.

I have researched the study of Manchester closed and it does not seem that it will affect any development, but I don't know much more. No comments.

Let's see if we soon have all the news, personally, I don't think Sony is handling the information situation well. I think I understand why they do it, but I hope the wait really compensates the user. Otherwise it could be an epic fail for everyone.

See u. Child crying.

 

[Neo Blaster]

Giuseppe Verdi!

Tendi nada, cara!!

 

[bhunachicken]

I think 54-56 CU's makes more sense. 2ghz is awfully hot to run in a closed form factor, no?

I literally have no idea..! I guess they can use a combination of heatpipes, passive cooling, and fans?

 

[bhunachicken]

Personally, I don't think Sony is handling the information situation well.

Could it be that the final specs still aren't quite locked down and there could be an 11th hour update?

 

[DeepEnigma]

Hi, I am new to this forum and while I follow conversations/speculations here, I am wondering if anyone commented on the XSX press release from Dec. 2019? It states;

Power Your Dreams with Xbox Series X, Available Holiday 2020 - Xbox Wire

Xbox Game Studios unveils Senua’s Saga: Hellblade II, a new chapter in the award-winning Hellblade story from Ninja Theory For eighteen years and three console generations, the Xbox community has shown us the power both games and fun can have on all of us. As we enter a new generation of gaming...

news.xbox.com

That is referring to the CPU, not the GPU.

 

[Koopatrol]

This is one of the theories. It makes sense.

But....that leaves the PS5.

18CUs for PS4, 18 more for Pro.

How about 18CUs?

That would give the PS5 54, a PS5 Pro 72., etc, etc.

This is one reason why to me github results don't make sense. Sony went to 36 CUs for the Pro. And just said the hell with it, we'll stay at 36....for their next gen baby?

Github is real, the results are real, but IMO it's not the entire picture. Or it could be. I just don't think it is.

Maybe Sony's BC patent would mimic the old hardware 100%. If MS and Sony do BC so different....isn't it reasonable that test results could also be different?

I don't think anybody should be denying the authenticity of the Github leak, but yes, I feel like it's hard to relate the Github leak to the PS5 when there's no context at all. Clocking the GPU at 2 GHz does not make sense; AegonSnake on the other site seems to have pretty reasoning as to why more CU's would be better rather than clocking super high, as the additional costs are super minimal. Also there's the other stuff like lack of ray tracing in the tests.

And I mean honestly, an intern was working with this information. The Github was uploaded in summer of 2019, but there's no indication IIRC as to how old the tests themselves are. And when you see the mix of PS5 code-names and Xbox code-names in the files, it makes it even more weird IMO. I mean if it was an intern doing these tests, he obviously wasn't super high up in the food chain; if anything, he could've been using old PS5 related files as a set up template for some coming Xbox related tests? I'm not super tech savvy so I don't know if that makes sense, but I also thought that the Github leak was lacking Xbox-related tests when compared to the PS5-related tests.

I dunno, I'm not involved in the software and hardware industries. But I was an intern in the engineering industry, and there were plenty of times where the engineers would be working on new projects, and they would told me to search old files and set up templates for reports, analyses, etc. for their new projects.

 

[James Sawyer Ford]

I think Sony is going wide and fast

GitHub is true for BC only, but the clocks match the final console

54 CU @ 2.0 GHz

SlimySnake quick, how many TFlops is that?

 

[BGs]

Could it be that the final specs still aren't quite locked down and there could be an 11th hour update?

No, I want to say that, personally, I don't think Sony needs to get away with so much secrecy and ambiguity.

Don't know if I translete it correctly.

Spoiler: in SPANISH

No, quiero decir que, personamente, no creo que Sony tenga necesidad de andarse con tanto secretismo y ambigüedad.

 

[Moses85]

I feel like Bill Murray:

 

[Ellery]

Could it be that the final specs still aren't quite locked down and there could be an 11th hour update?

It is way too late now to change anything about the actual chip (APU = GPU+CPU+ maybe ray tracing cores). They could adjust minor things, but even minor things could make the whole thing unbalanced. Like maybe slightly upping clockspeed, but that means more power = more heat and your power supply has to be able to handle it and the cooling needs to handle it.
Right now the consoles are final in terms of hardware. Sony and Microsoft know exactly what hardware you are going to get.

I think 54-56 CU's makes more sense. 2ghz is awfully hot to run in a closed form factor, no? Considering the cooling needed, wouldn't it then be cheaper to go with more CU's and lower clock speed?

Neither makes sense, because more CU = more die space = much higher cost for Sony/MS.
Higher clock = way more power draw = way hotter = way harder to cool.

It is all theoretically possible, but it is a math equation where you can't use magic or trickery to get to your desired end result. If you change 1 thing then it changes everything.

It depends on what they are ready to give us and think the market is ready to buy/pay for. Technically they could give us a 56 CU at 1700MHZ or 60CU at 1600MHZ but that would still be A LOT and by A LOT I mean A LOT of power (which is heat).

They have to make it work in a console form factor, they have to have the right cooling and all the other components which cost money aswell.

It is much easier to start thinking with the costs and what is theoretically possible instead of starting with the wish of hitting certain TF numbers. Biggest problems are the costs of the chips and power/heat.

 

[ANIMAL1975]

Even Google Translate has issues translating even Spanish and messes up certain tenses and words but it is quite good.

I was being mean, nothing is worse than Stadia

I read it after and yes, it broke my cervello.

this part must be an insider riddle leak
shit!

and second to none with the fan put in the middle type ps3 solve everything”

 

[pawel86ck]

I think Sony is going wide and fast

GitHub is true for BC only, but the clocks match the final console

54 CU @ 2.0 GHz

SlimySnake quick, how many TFlops is that?

54CU (3456 SP) 2GHz = 13.8TF

 

[Reindeer]

Where did it say I assuming Sony are making 2 different consoles.

Is basic reading that hard ?

I said explore, do you think Sony only looked at 1 chip design and called it a day ?

Where did I say that you said that Sony are making two different consoles? I said you assume that they're making two chips for one console.

Is basic reading that hard?

 

[JAMMA]

.

 

[PaintTinJr]

This is one of the theories. It makes sense.

But....that leaves the PS5.

18CUs for PS4, 18 more for Pro.

How about 18CUs?

That would give the PS5 54, a PS5 Pro 72., etc, etc.

This is one reason why to me github results don't make sense. Sony went to 36 CUs for the Pro. And just said the hell with it, we'll stay at 36....for their next gen baby?

Github is real, the results are real, but IMO it's not the entire picture. Or it could be. I just don't think it is.

Maybe Sony's BC patent would mimic the old hardware 100%. If MS and Sony do BC so different....isn't it reasonable that test results could also be different?

Looking a the V shaped design – which is iconic like the launch ps1, ps2, p3 designs (IMHO), so possibly the real shape – it would suggest cooling symmetry beyond cooling symmetrical memory layout.

I’m now thinking the Github leak is real for PS5, but for Sony to again play to their EE design strengths (like Ps2, PS3 and Ps4) they are going to use Dual Oberon chips and downclock at whatever level they can, to beat MS using a single RDNA2 XsX chip (with them rebinning defects as XsS).

A dual chip setup for PS5 would allow them to hit much higher performance – like their PS2 Dual coprocessor design and intended PS3 dual Cell processor design. They would also maximize their waffer yield and could aim lower power and quieter if downclocking heavily to hit 7TF from each. A solution like this might also explain why someone mentioned they are still negotiating price with AMD, and why general consensus of their SSD solution being faster and them being better on the RT side of things, and suggestions they might be above RTX 2080 TI. I know it is wishful thinking , but (IMHO) Sony’s real strength is how they configure the items available and how many they use – like the GDDR5 bandwidth on PS4. I resigned to them not having a special super secret APU design, as re-living the risks of a cutting edge Cell solution (even when they had Toshiba and IBM expertise partnering) isn’t something I think they’d risk, now, meaning they’d automatically be behind trillion dollar MS on what they can afford to buy from AMD if going single chip.

I also think the 3D audio chip inside the APU is an AMD version of the Cell processor, making BC completely doable in hardware, but that's more wishful thinking by me, I suspect

 

[sonomamashine]

A few points of contention here. The 36CU chip @ 40CUs wouldn't actually quite give the performance levels of a 48CU chip hitting around the sweetspot levels unless the 36 CU chip is pushed to around 2250MHz - 2400MHz, which at that point would probably just melt the chip. I agree tho that whatever PS5's final specs end up at, it absolutely won't be running at 2GHz on the GPU.

I agree the dual-chiplet idea is somewhat on the crazier side; have discussed why it would be insane in a way with dual Oberons (clocked waaaaay below sweetspots, giving "only" 12 or 13TFs when they could give a lot more, no insider rumors honestly trying to push something like 16TF (about what dual Oberons or a 72CU Oberon in sweetspot frequencies would give) etc.); it was just wild speculation on my end. It's pretty much the least likely scenario tbh.

That said, the Oberon chip having a chunk of CU's disabled is a lot more likely, so in actuality it could be a possible 48 CU chip, but they're working some bugs out most likely (maybe there is an Oberon stepping waiting to be datamined with more active CUs on it, hopefully it comes up within the next few weeks leading up to GDC). That said, I've prepared myself for a "worst-case" scenario if it really is just 36 active CUs max (I think this is less likely than a later Oberon stepping having more CUs active, but it's a potential possibility). Because as you say, there's no way either system's clocking their retail GPUs @ 2GHz, even high-end PC GPU cards don't push clocks that high and they have pretty insane cooling.

40CU's @2Ghz would deliver aproximatively the same perf as 48CU's @1700Mhz, either way oberon being 36CU's or 40CU's doesn't matter, that's why you always see crappy games at console launch compared to others, they can't give Dev's retail chip.
-There is another assumption why data miners didn't get retail PS5 chip info's only because Sony decided to test it in intern and went with the other test chip -40CU's running online benches to keep quiet, this is a billion dollar market why wouldn't they keep the final chip secret in intern to surprise competition ?

I will be repeating myself (already posted this before but there was no feedback), what if it is something in the line of a 3D stack layout approach, to reduce dye size and maintain the standard console form factor?

No, 7nm, 5nm, 3nm, 3D stack.

I think 54-56 CU's makes more sense. 2ghz is awfully hot to run in a closed form factor, no? Considering the cooling needed, wouldn't it then be cheaper to go with more CU's and lower clock speed?

I think Sony is going wide and fast

GitHub is true for BC only, but the clocks match the final console

54 CU @ 2.0 GHz

SlimySnake quick, how many TFlops is that?

No.

 

[James Sawyer Ford]

54CU (3456 SP) 2GHz = 13.8TF

pretty close to the 14.2 TF leak...15% more powerful than XSX...with insiders hinting PS5 is more powerful than XSX and the story has not changed...both consoles being “close”

I want to believe

 

[SlimySnake]

I think Sony is going wide and fast

GitHub is true for BC only, but the clocks match the final console

54 CU @ 2.0 GHz

SlimySnake quick, how many TFlops is that?

13.8 tflops. ive seen that number come up before.

54 cus will be 3 shader arrays of 20 cus, 6 disabled CUs gives us 54 cus. they likely initially went wide and slow-ish around 1.7 ghz which is the sweet spot for navi 10 cards, and then increased the clocks to see how far they can push it. MS aiming for 12 tflops likely forced their hand to push it higher.

im guessing they will land somewhere around 1.85-1.9ghz which should be around 13.0 tflops.

but for that to happen, a lot of other things have to happen. they need amd engineers to improve power efficiency on navi cards by 20% so that 1.7 ghz sweet spot becomes 2.0 ghz. then n7p to get further 7% in power efficiency. that should bring the tdp of that gpu down under 130w. lastly, hbm to bring the total tdp of the system down under 225w. preferably around 200w.

 

[darkinstinct]

This is one reason why to me github results don't make sense. Sony went to 36 CUs for the Pro. And just said the hell with it, we'll stay at 36....for their next gen baby?

If they went with RDNA (Navi10), there is nothing else. RDNA only has a max of 40 CU, so they deactivate 4 for yields and end up at 36. The only way to get more performance out of RDNA is to increase clock speeds. Microsoft went with next-gen RDNA (Navi21, some call it RDNA2), which can go up to 80 CU. This narrative (that is supported by the AMD leak, the codenames match) fits with Jez Corden saying on Twitter that there was a reason why Microsoft was substantially behind Sony when getting devkits out. That reason would be that Navi21 wasn't available until November 2019 in volume production.

AMD is already producing Navi 21 at 7nm+ with the double size of Navi 10

Halfway through last year, AMD launched the Navi 10, at the heart of the Radeon RX 5700 XT, as the first sample of its new RDNA architecture, and even though not much time passed, the

optocrypto.com

It also makes sense regarding rumors that Sony originally planned for a 2019 launch. That original launch APU would've been Ariel, when they delayed it they upped the clocks from 1.8 GHz to 2 GHz and it turned into Oberon. That clock speed increase leads to loads of heat, which explains the devkit design.

https://www.overclock3d.net/news/systems/tsmc_is_rumoured_to_be_creating_a_new_7nm_console_chip/1 (April 2018)
https://www.neogaf.com/threads/poss...3-2ghz-clock-navi-10-based-gpu.1471267/page-3 (January 2019)

At least that makes a lot more sense than the alternative, that Sony designed one APU for a 2019 launch, then skipped it to create Ariel, then skipped that to create Oberon, did four more iterations of Oberon, while secretly working on the actual PS5 APU. That's a lot of money spent for creating designs you don't intend to use.

 

[PocoJoe]

Where did I say that you said that Sony are making two different consoles? I said you assume that they're making two chips for one console.

Is basic reading that hard?

well sometimes it is safe to assume things if they are based on history and facts. I have read that they even prototyped bulldozer cpu cores for ps4.

Imo it is more crazy to assume that they didnt prototype multiple chips vs. they did

 

[sonomamashine]

thicc_girls_are_teh_best yes i did get the dual chip theory at lower clock (1300 etc...) but still, it will run hotter than 2GHz.

 

[jackets with pockets]

I feel like Bill Murray:

 

[Gavin Stevens]

FOURTEEN POINT TWO TERROR FLOPPIES.

Excuse me, I’m just about to go take a hammer to my 2080ti For being so underpowered and so expensive compared to the almighty PlayStation. Hail hydra.

Wait didn’t some of you tits say just recently that Ms couldn’t possibly have anything as high as over 12tf because it wouldn’t work due to heat/power/cockroaches...?

Make up your damn minds people

 

[BGs]

Deleted

 

[James Sawyer Ford]

Seems the magic number

we’ve nailed it, folks

benji was right

 

[liviopangt]

A few people have asked me about raytracing now, so I thought I would enquire at our latest drinking session. There isn't a lot I can tell you right now. But what I can say is when turning on raytracing in engine on PS5 devkit the frame rate cost is around 30%. Bearing in mind this is before any optimisation. The Dev team think they will be able to get this down to 15-20%, which should be enough to hit their target frame rate without too much trouble. On the PC turning on the same effects on Nvidia hardware costs around 35%. I can't give you comparison with Series X as the team are still yet to receive their kits. Apparently they may have to wait till April! Which would give Sony a considerable leg-up in terms of hands on time with the hardware. This may also explain why no Series X exclusives for launch, if teams simply haven't had long enough to aquiant themselves to the hardware.

but are you a programmer working on ps5 dev kits?

 

[darkinstinct]

Looking a the V shaped design – which is iconic like the launch ps1, ps2, p3 designs (IMHO), so possibly the real shape – it would suggest cooling symmetry beyond cooling symmetrical memory layout.

I’m now thinking the Github leak is real for PS5, but for Sony to again play to their EE design strengths (like Ps2, PS3 and Ps4) they are going to use Dual Oberon chips and downclock at whatever level they can, to beat MS using a single RDNA2 XsX chip (with them rebinning defects as XsS).

A dual chip setup for PS5 would allow them to hit much higher performance – like their PS2 Dual coprocessor design and intended PS3 dual Cell processor design. They would also maximize their waffer yield and could aim lower power and quieter if downclocking heavily to hit 7TF from each. A solution like this might also explain why someone mentioned they are still negotiating price with AMD, and why general consensus of their SSD solution being faster and them being better on the RT side of things, and suggestions they might be above RTX 2080 TI. I know it is wishful thinking , but (IMHO) Sony’s real strength is how they configure the items available and how many they use – like the GDDR5 bandwidth on PS4. I resigned to them not having a special super secret APU design, as re-living the risks of a cutting edge Cell solution (even when they had Toshiba and IBM expertise partnering) isn’t something I think they’d risk, now, meaning they’d automatically be behind trillion dollar MS on what they can afford to buy from AMD if going single chip.

I also think the 3D audio chip inside the APU is an AMD version of the Cell processor, making BC completely doable in hardware, but that's more wishful thinking by me, I suspect

You do realize that for a dual Oberon setup they would need to not only spend $200 per console on the APUs themselves but also double the memory amount, right? So if they intend to use 20 GB total (16 GB GDDR6, 4 GB DDR4 for OS) they would need 36 GB of RAM. Which is also doubling the RAM cost pretty much, from around $60 to $120. That console you have in mind would cost upwards of $699. It would consume around 350 Watt. Just so you know what a dual Oberon design means.

 

[ANIMAL1975]

Edit: too slow

 

[SlimySnake]

FOURTEEN POINT TWO TERROR FLOPPIES.

Excuse me, I’m just about to go take a hammer to my 2080ti For being so underpowered and so expensive compared to the almighty PlayStation. Hail hydra.

Wait didn’t some of you tits say just recently that Ms couldn’t possibly have anything as high as over 12tf because it wouldn’t work due to heat/power/cockroaches...?

Make up your damn minds people

2080ti is over 17 tflops. dont go by the nvidia tflops, they have the clocks at 1.35 ghz on the official specs. you only hit those clocks during main menus. all rtx gpus are 1.9 to 19.5 ghz during gameplay. that gives us 11.4 tflops for the rtx 2080 and 17.5 tflops for the 68 cu 2080 ti.

a 14 tflops gpu wont come close to matching the performance of a 2080 ti. maybe a little bit better than the 2080 super.

also, it looks like ms is going with a gigantic apu. they can easily have 70 cus in there as long as they cut down the zen 2 cache by 75%. it seems they are going wide and slow at 1.47 ghz and 64 cus. they can easily hit 1.5-1.6 ghz and get to 13 tflops themselves.

which is what im guessing what they will do if sony comes around 12 tflops.

 

[liviopangt]

Dude you got to translate that shit if you want to be understood... or i will start to reply to you in portuguese, percebido?

I don't come here to say that English is shit you understand?

 

[James Sawyer Ford]

2080ti is over 17 tflops. dont go by the nvidia tflops, they have the clocks at 1.35 ghz on the official specs. you only hit those clocks during main menus. all rtx gpus are 1.9 to 19.5 ghz during gameplay. that gives us 11.4 tflops for the rtx 2080 and 17.5 tflops for the 68 cu 2080 ti.

a 14 tflops gpu wont come close to matching the performance of a 2080 ti. maybe a little bit better than the 2080 super.

didnt somebody (schreir?) recently suggest both consoles aiming for better performance than rtx2080 or was it the 2080 ti?

 

[semicool]

I don’t know man adding 20 cu is much cheaper compared to dual gpu . We shouldn’t go all crazy yet haha

Not to mention 2x more efficient in performance per shader, and that's not even mentioning energy efficiency and like you alluded to...MUCH cheaper etc....dual GPU is a lost cause

 

[SlimySnake]

didnt somebody (schreir?) recently suggest both consoles aiming for better performance than rtx2080 or was it the 2080 ti?

schrier did indeed say hes heard both consoles are even more powerful than the 2080.

but he didnt want to be quoted on message boards because he didnt believe it himself. its on a podcast. let me find it. very interesting.

that would put both consoles at 12 tflops.

 

[Reindeer]

well sometimes it is safe to assume things if they are based on history and facts. I have read that they even prototyped bulldozer cpu cores for ps4.

Imo it is more crazy to assume that they didnt prototype multiple chips vs. they did

Yeah...I highly doubt they would be prototyping another chip that late, it's either a final chip or it's been replaced by a new one very late into console's development. It could be that they pushed the 36CU chip to the max and realised that it wasn't good enough and replaced it with more powerful chip.

 

[sonomamashine]

schrier did indeed say hes heard both consoles are even more powerful than the 2080.

but he didnt want to be quoted on message boards because he didnt believe it himself. its on a podcast. let me find it. very interesting.

that would put both consoles at 12 tflops.

Can you put timer

 

[SlimySnake]

Can you put timer

didnt somebody (schreir?) recently suggest both consoles aiming for better performance than rtx2080 or was it the 2080 ti?

Google Podcasts

With Google Podcasts, you can find and listen to the world's podcasts for free.

podcasts.google.com

timestamp: 38:05-38:50.

 

[liviopangt]

however if sony nn has still released the full specifications and because they don't want microsoft engineers to bring improvements at the last minute to the gpu and ram everything will be revealed as soon as the games are done and the chips in production

 

[James Sawyer Ford]

however if sony nn has still released the full specifications and because they don't want microsoft engineers to bring improvements at the last minute to the gpu and ram everything will be revealed as soon as the games are done and the chips in production

those items are already locked down

Sony doesn’t need that as an excuse to not reveal their console yet. We are well past the point of needing to be secretive for competitive advantage

the only thing they may want to be secretive on is price, but specs can’tbe adjusted at this point other than minor clocks or RAM quantity depending on the memory setup (if it allows for additions)

 

[Neo Blaster]

those items are already locked down

Sony doesn’t need that as an excuse to not reveal their console yet. We are well past the point of needing to be secretive for competitive advantage

the only thing they may want to be secretive on is price, but specs can’tbe adjusted at this point other than minor clocks or RAM quantity depending on the memory setup (if it allows for additions)

Well, both can still adjust clock and mem size.

Would it be too crazy to think Oberon is just a PS4 Pro revision using zen2+navi, 2GHz being just a stress test?