• 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.

Disco_

Member
You mean fixing backwards?
I mean "fixing". PS5 is mostly data copied from 5700xt the rest is github. Sacrlett is data copied from xonex and rest github. It's placeholder info given the validity of final spec. It's silly.
 

Marlenus

Member
No it's not. It's progressing the same as N7. There is a huge difference between having the same maturity and being on a similar maturation tragectory. There is no way N7+ has defect densities as low as 0.07/cm2. It only recently went into volume production.

There is also the fact that wafer supply for N7+ is lower than N7 - because it's a newer node. Each of those wafers are gonna be very expensive and AMD will want to use as much of those wafers as they can on Milan. Whatever is left will go to Ryzen and later Threadripper, because they leverage the same chiplet design. What's left will probably go to desktop GPU.
Then after all that, will consoles be considered.
Why? Because Epyc (and Radeon Instinct) has the highest margins. Then comes Threadripper. Then comes Ryzen. Then comes Radeon (though it seems AMD is following suit with Nvidia in keeping prices high).

Consoles have the lowest margin of all the chips they sell. AMD are no longer back in 2012, reeling from the devastatingly bad Bulldozer, and the underwhelming Tahiti. They're not as desperate for console volume, because they are making money in desktop and they're making money in datacenter. They're having difficulty meeting demand in fact.

It is for that reason that I seriously doubt either console will be N7+. At best N7P because it uses the same design rules as N7 and won't need brand new photomasks.

I mean maybe AMD have given N7+ wafers to Microsoft and/or s
Sony, but you can bet your ass they're going to be paying for them, cause AMD want their gross margin to be on the up. They are a publicly held business after all. They don't operate for the benefit of gamers.
It's possible, but unrealistic.

According to Anandtech TSMC said that N7+ had caught up with N7 yields after just 2 quarters and they put it down to the learning they had when bringing N7 online.

MS/Sony will sort their own wafer supply out and AMD have nothing to do with it. At best AMD might make a recommendation based on the provided targets and they can design with a process in mind but securing the wafers and the supply contract is not something AMD do for their semi custom clients. AMD get their money via upfront costs and then via licensing for the IPs used in the custom SoC.
 
Last edited:

R600

Banned
I mean "fixing". PS5 is mostly data copied from 5700xt the rest is github. Sacrlett is data copied from xonex and rest github. It's placeholder info given the validity of final spec. It's silly.
Yea but nowhere in Github did Arden have 384bit bus that is why.
 

SlimySnake

Flashless at the Golden Globes
Well the 1st question is can 1 shader engine support 15 wgps? The max we have seen is 12 in navi14 but only 11 are active in the 5500xt.

If not it means we need 3 engines for 60CUs and that is 225mm² with all 96 ROPs and no ray tracing.

370mm² is only doable if a shader engine can support 15 wgps otherwise it will be closer to 400mm².
the 5500xt has 24 CUs, 2 disabled right? how many work groups is that? 12? so 1 wgps = 2 CUs or 1 dual cu?

a console with 2 shader engines should be able to have at least 48 cus then. 52 if there are 12?
 

R600

Banned
According to Anandtech TSMC said that N7+ had caught up with N7 yields after just 2 quarters and they put it down to the learning they had when bringing N7 online.

MS/Sony will sort their own wafer supply out and AMD have nothing to do with it. At best AMD might make a recommendation based on the provided targets and they can design with a process in mind but securing the wafers and the supply contract is not something AMD do for their semi custom clients. AMD get their money via upfront costs and then via licensing for the IPs used in the custom SoC.
I think AMD is actually direct customer. They cant give anyone x86 IP.
 

Marlenus

Member
I think AMD is actually direct customer. They cant give anyone x86 IP.

They dont give x86 IP. They licence a specific chip design that their clients can have manufactured. MS mentioned this previously.

This came about after the updated to the Intel AMD cross licensing deal allowing AMD to sell the fab side of their business and properly setup their semi custom business.
 

Marlenus

Member
the 5500xt has 24 CUs, 2 disabled right? how many work groups is that? 12? so 1 wgps = 2 CUs or 1 dual cu?

a console with 2 shader engines should be able to have at least 48 cus then. 52 if there are 12?

2CUs per wgp. Navi 14 has 12 wgps (24CUs) but the 5500xt only has 11 wgps(22CUs) enabled. Not sure if there are any fully enabled Navi 14 examples but apple probably has them if they exist.

That means the max wgps per shader engine we have seen is 12 so the emperical max for a 2SE chip is 24 which is 48 CUs. A single shader engine may be able to support more than 24 CUs but we do not have evidence to support that.
 

R600

Banned
Like you? Keep pushing that 9.2TF narrative but you don’t have the balls to bet, hilarious.
Bet who? Same song and dance as 2012 when Durango was leaked 1.2TF

"No way is MS gonna let them have 50% more TF"

"Document is fake"

"Its early"

Look, bookmark this, I HOPE PS5 is better, even if by 1%. Why? Because I only have PS4 and I want my games there, and I want best version of 3rd party games on my default console. Aaaand because I would buy lockhart for MS exclusives.

BUT its not what numbers show. 6 month data is NOT old. If V dekvkits were out in summer, chip was already inside, there is nothing they can do. Clocks are high up anyway, so they cant raise it. Not that they have to, console looks beast. If MS had single SKU strat they would not be able to beat that, because its near the top we expected.

Understand its not narrative. Its internal AMD testing document for next gen consoles that was removed from twitter and github for a reason. I am trying to be objective, if Oberon was MS I would say the same thing.
 

Marlenus

Member
Yea but nowhere in Github did Arden have 384bit bus that is why.

A 320 bit bus limits the amount of ram you can use. Unless the consoles are going to have a lot more ram than we think (say 40GB for a 320 bit bus) I think 384 bit offers the most reasonable amount without needing to have very high ram clocks to hit enough bandwidth to feed a 12 TFlop GPU.
 

R600

Banned
A 320 bit bus limits the amount of ram you can use. Unless the consoles are going to have a lot more ram than we think (say 40GB for a 320 bit bus) I think 384 bit offers the most reasonable amount without needing to have very high ram clocks to hit enough bandwidth to feed a 12 TFlop GPU.
I think 512bit bus is even better :) but yea, 384bit bus is 16mm² bigger then 320bit bus, and thats not small amount.

In addition, they might not even need it. 320bit bus gives you max 20GB, which if using 18Gbps memory is 720GB/s. Much more then enough.

Oberon shows 256bit bus and 18Gbps chips for 576GB/s max theoretical BW. Why would they go for higher if they dont need more RAM or BW? There is a sweetspot and I think they have hit it.
 

StreetsofBeige

Gold Member
Bet who? Same song and dance as 2012 when Durango was leaked 1.2TF

"No way is MS gonna let them have 50% more TF"

"Document is fake"

"Its early"

Look, bookmark this, I HOPE PS5 is better, even if by 1%. Why? Because I only have PS4 and I want my games there, and I want best version of 3rd party games on my default console. Aaaand because I would buy lockhart for MS exclusives.

BUT its not what numbers show. 6 month data is NOT old. If V dekvkits were out in summer, chip was already inside, there is nothing they can do. Clocks are high up anyway, so they cant raise it. Not that they have to, console looks beast. If MS had single SKU strat they would not be able to beat that, because its near the top we expected.

Understand its not narrative. Its internal AMD testing document for next gen consoles that was removed from twitter and github for a reason. I am trying to be objective, if Oberon was MS I would say the same thing.
Yup.

The old Durango and Orbis leaks from June 2012 had them exactly where they ended up as. Xbox One said 1.1-1.2. Ended up being 1.3TF. And PS4 was dead on at 1.8TF.

Whomever did the leaks even got the ram boost leak right saying each system had 2-4 gb of ram, but they might be boosted to 8gb. Spot on again. Both were 8gb.

And that was about a year and a half before launch!
 

demigod

Member
Bet who? Same song and dance as 2012 when Durango was leaked 1.2TF

"No way is MS gonna let them have 50% more TF"

"Document is fake"

"Its early"

Look, bookmark this, I HOPE PS5 is better, even if by 1%. Why? Because I only have PS4 and I want my games there, and I want best version of 3rd party games on my default console. Aaaand because I would buy lockhart for MS exclusives.

BUT its not what numbers show. 6 month data is NOT old. If V dekvkits were out in summer, chip was already inside, there is nothing they can do. Clocks are high up anyway, so they cant raise it. Not that they have to, console looks beast. If MS had single SKU strat they would not be able to beat that, because its near the top we expected.

Understand its not narrative. Its internal AMD testing document for next gen consoles that was removed from twitter and github for a reason. I am trying to be objective, if Oberon was MS I would say the same thing.

So now you're being what you accused bitbydeath of, obtusive. The document was created in 2018, it's old. Matt has already implied that its not the final specs and he's glad "people are over the github", but here you are still pushing that narrative that it will be final specs. But you're too scared to bet because you yourself don't even believe in the bs you're spouting.
 

Marlenus

Member
I think 512bit bus is even better :) but yea, 384bit bus is 16mm² bigger then 320bit bus, and thats not small amount.

In addition, they might not even need it. 320bit bus gives you max 20GB, which if using 18Gbps memory is 720GB/s. Much more then enough.

Oberon shows 256bit bus and 18Gbps chips for 576GB/s max theoretical BW. Why would they go for higher if they dont need more RAM or BW? There is a sweetspot and I think they have hit it.

I am just not sure if 20GB will last a 7 year gen so unless they go with 40GB then 320 feels limiting when 384 gets you 24GB and you can use slower ram that costs less.
 

demigod

Member
Yup.

The old Durango and Orbis leaks from June 2012 had them exactly where they ended up as. Xbox One said 1.1-1.2. Ended up being 1.3TF. And PS4 was dead on at 1.8TF.

Whomever did the leaks even got the ram boost leak right saying each system had 2-4 gb of ram, but they might be boosted to 8gb. Spot on again. Both were 8gb.

And that was about a year and a half before launch!

The difference is, we knew those machines were going to be weak as fuck. There was no other leak that said it was more powerful.
 
  • LOL
Reactions: Dim

R600

Banned
Take a week off from this thread. Your dismissive attitude now regressing to name calling needs to be dialed back. This isn’t the R600 show.
So now you're being what you accused bitbydeath of, obtusive. The document was created in 2018, it's old. Matt has already implied that its not the final specs and he's glad "people are over the github", but here you are still pushing that narrative that it will be final specs. But you're too scared to bet because you yourself don't even believe in the bs you're spouting.
It was created in 2018 because Navi 10, which is in document, was released in july 2019 (and has most info by far actually). Last time it was updated was in June, and given that we have Ariel A0, Oberon A0 and B0 results, we know Oberon couldnt be from 2018 even if Ariel was (and it doesnt seem to be, its more likely to be January).

As for Matt or anyone from Resetera role playing insiders, I dont give a shit. If he knows the specs he should put them like they are in Github and not be vague, which he always is (even when he "denied" the leak).

Bet where moron? Yes, I believe the bs like AMDs official testing data and not in Oracle that is Matt from Reee.
 
Last edited:

demigod

Member
It was created in 2018 because Navi 10, which is in document, was released in july 2019 (and has most info by far actually). Last time it was updated was in June, and given that we have Ariel A0, Oberon A0 and B0 results, we know Oberon couldnt be from 2018 even if Ariel was (and it doesnt seem to be, its more likely to be January).

As for Matt or anyone from Resetera role playing insiders, I dont give a shit. If he knows the specs he should put them like they are in Github and not be vague, which he always is (even when he "denied" the leak).

Bet where moron? Yes, I believe the bs like AMDs official testing data and not in Oracle that is Matt from Reee.

Personal insults now?
 
They're already fixing things. Scalett had 64 rops yesterday, up to 96 today.

No way...

giphy.gif
 

Lort

Banned
I'm not sure how they managed to pull it off but they have successfully stuffed the Era moderation team with Xbox Ambassadors. After one is promoted they start nominating and recommending their friends to the moderation team I suppose.

Ghostcrew for the record is not an Xbox Ambassador but his promotion remains a puzzle to many because he was notorious for anti-Sony baiting and had numerous infractions to his name. Then one month he suddenly starts acting overtly pro Sony. A few of us joked that his account had been hijacked because it was most unusual behaviour. The reason became clear after he was promoted to the moderation team. His buddies must have nominated him and so he adopted an overt stance of neutrality while privately he issues warnings and bans to PlayStation enthusiastic forum members, all the time ignoring similar infractions on the other side.

Era is just a pretence. More and more people are beginning to see it.

So Its an XboxERA and SonyGAF then...

In truth since the great purge the new mods have been fair and I’m no longer pointing out to them how threads I create should be allowed since they are literally me copying a non banned Sony thread title. About the last overtly Sony MOD change was to insert “who would want to play these games” into the thread about the awesome backwards compatibility xbox offers. ( this was not inserted into any Sony ps5 bf threads which is odd because it was Sony who said it about Sony games).

So anyway all good here but not surprised to see all companies gorilla marketing where they can.
 

Imtjnotu

Member
Yeah, treat that with a fair amount of skepticism. They sometimes post the specs of cards that were never released.
i think 64 is good tho.

considering the x was able to get away with 32 and sony overkilled at 64 on the pro. 64 should be perfect for both to get away with 4k native
 
Too much bullshit out there for me to make a conclusion-acceptance of the actual specs of PS5 and Xsex vs Durango and Orbis back in 2012:

-Developers saying that PS5 and Xsex are pretty close and hardly any difference with one with slight edge (10% difference at most)
-RDNA vs RDNA 1.5 vs RDNA 2
-Game journalist-youtuber insider leak fuckery causing disruption and throwing a wrench contradicting developers.

Tired of this shit.
ChelyuG.gif
 

DForce

NaughtyDog Defense Force
You have confident people on both sides.

One side believes the leaks are it and there's nothing else to worry about.


Then you have insiders on the other side who are very confident in what they believe, even after the leaks.

I just hope we don't see people disappear when everything gets announced.
 

vpance

Member
Samsung going all in on 7nm. Is 600k WPM unprecedented? By comparison TSMC plans to be at 140k WPM in 2H 2020.

At present, Samsung's 7nm production capacity is roughly 150,000 wpm. It is also actively increasing 7nm capacity. According to industry rumors, Samsung plans to quadruple capacity in 2020. Nvidia and Qualcomm's next-generation products may be produced using Samsung's 7nm EUV process, but details remain to be seen.

TSMC's 7nm production capacity is fully booked. Relief may only come when Apple migrates to 5nm in 2H'2020. TSMC's 7nm capacity will increase to 140,000 wpm in 2H'2020. By order proportion, the ranking of customers using 7nm will be re-shuffled. AMD's orders are set to double, replacing Apple as the largest customer [for 7nm]. Huawei's HiSilicon and Qualcomm are similar by order proportion.

https://wccftech.com/amd-7nm-wafer-...-7nm-capacity-at-tsmc-currently-fully-booked/

Wonder if any console will go with Sammy. Would be interesting if Sony went with them for the ultimate APU/HBM/ReRam/1TB flash package deal.
 
Status
Not open for further replies.
Top Bottom