Baseless Next Generation Rumors with no Technical Merits [post E3 2019, pre GDC 2020] [XBSX, PS5]

[HBRU]

IMO, if it is indeed clocked at 2 GHz, I'm thinking it'll likely be in the 225-275 watt range. And that is assuming they get some relatively significant gains WRT to power efficiency. If power efficiency gains are relatively minor then it could be higher even than that.

That's basically going to keep it beyond the knee of the power efficiency curve.

Regards,
SB

There is also a commercial curve....

A 150 dollar 4K console (ps4pro diskless + HD) could sell really a lot extending ps4 commercial life....

Then a "best bang for the buck" 250 dollars 8.2 TF (+SSD fast loading) ps4pro2 also can extend ps4 commercial life till almost "forever"..

Ps5 will be the flagship console utilizing best uncutted chips... and going to be bought by enthusiast people... also with the 4k blue ray player...

 

[fehu]

So? Are we at 15TF yet?

 

[PSman1700]

So? Are we at 15TF yet?

Half way, about.

 

[Shifty Geezer]

I find also strange the ps-plus games for January have not been announced... something is boiling at Sony... ?!?

No.

Uncharted: The Nathan Drake Collection headlines January PlayStation Plus line-up. Also, Goat Simulator.

 

[ultragpu]

Any new news on PS5 yet? At this point I'll be glad if they can just reach 10 TF so it looks and sounds better, damn this TF business. Also I'm sure this was discussed here somewhere but if Sony goes wide and slow like 56 CU at 1700mghz, is it really impossible to do BC with PS4 and Pro that way? Is it not so much easier and cost effective to reach higher TFs using the wide approach, why banking on the few percentage of BC whores and go hellbent on extreme clock, expensive cooling solution yet not even reaching 10 TF in the end?

 

[London Geezer]

This TF malarkey is so weird and funny. Gamers bitching about a missing 800GFlops, as if we can actually see what that looks like? 9.2, 10, it’s literally all the same in the grand scheme of things, especially when the biggest generational jump will be on the CPU and SSD side of things. Now THAT will actually be very visible to everyone.

Anyone coming from the base PS4 will see an insane jump. Coming from Pro a little less.

I’ll tell you what I want, what I really really want. 60fps on *most* games. I don’t care if cinematic, open, closed, RT, no RT, this or that, I don’t care. 60 fps please, we have earned it. The CPU will surely give us that.

 

[PSman1700]

The CPU will surely give us that.

Now, will devs give us that?

 

[HBRU]

This TF malarkey is so weird and funny. Gamers bitching about a missing 800GFlops, as if we can actually see what that looks like? 9.2, 10, it’s literally all the same in the grand scheme of things, especially when the biggest generational jump will be on the CPU and SSD side of things. Now THAT will actually be very visible to everyone.

Anyone coming from the base PS4 will see an insane jump. Coming from Pro a little less.

I’ll tell you what I want, what I really really want. 60fps on *most* games. I don’t care if cinematic, open, closed, RT, no RT, this or that, I don’t care. 60 fps please, we have earned it. The CPU will surely give us that.

True

 

[iroboto]

Any new news on PS5 yet? At this point I'll be glad if they can just reach 10 TF so it looks and sounds better, damn this TF business. Also I'm sure this was discussed here somewhere but if Sony goes wide and slow like 56 CU at 1700mghz, is it really impossible to do BC with PS4 and Pro that way? Is it not so much easier and cost effective to reach higher TFs using the wide approach, why banking on the few percentage of BC whores and go hellbent on extreme clock, expensive cooling solution yet not even reaching 10 TF in the end?

no one knows.
you don’t know everything about the chip. You don’t know everything about the system.
Have all the flops you want but without a massive amount of bandwidth to feed it; it’s all going to be bandwidth bound anyway. You can rely on other technologies to make up a a difference but we saw how that turned out for 4Pro and X1X.
big TF means nothing without big bandwidth. And big bandwidth and big TF means big price.

 

[PSman1700]

MS doubles the One X, Sony the Pro.

 

[HBRU]

MS doubles the One X, Sony the Pro.

Sony a bit more even IMHO.
I think that -at the end- we will discover that the TF/bandwidth ratio is a bit better for Sony.... also Sony I think will clock the CPU slower

 

[AbsoluteBeginner]

Any new news on PS5 yet? At this point I'll be glad if they can just reach 10 TF so it looks and sounds better, damn this TF business. Also I'm sure this was discussed here somewhere but if Sony goes wide and slow like 56 CU at 1700mghz, is it really impossible to do BC with PS4 and Pro that way? Is it not so much easier and cost effective to reach higher TFs using the wide approach, why banking on the few percentage of BC whores and go hellbent on extreme clock, expensive cooling solution yet not even reaching 10 TF in the end?

Reading what Jim said about fastest transition to next gen PS ever, I would think BC is of extreme importance to Sony. Inability to play your PS4/Pro library on PS5 would be massive fuck up. Imagine MS giving you game pass and hundreds of last gen game on it, and PS5 doesn't even give you ability to play your disk or digital games from last gen on it? Its probably much more important for them then ~5% higher TF number.

If I could speculate, I think this is big reason why Sony started earlier to work on Navi with AMD. They absolutely have to nail BC for last gen, otherwise pool of 100+ million current costumers might just buy next Xbox.

 

[JoeJ]

MS doubles the One X, Sony the Pro.

It's often overlooked PS4 Pro has more fp16 TF than XBX.
Edit: XBX has no fp16, so if you can use it for most, PS 8 TFfp16 could be faster than XBX 6 TFfp32. But i don't know if this ever happened in practice.

 

[BRiT]

Never happened cause 4Pro is missing the needed bandwidth and RPM FP16 only helping in limited situations over the big picture.

 

[iroboto]

It's often overlooked PS4 Pro has more fp16 TF than XBX.
Edit: XBX has no fp16, so if you can use it for most, PS 8 TFfp16 could be faster than XBX 6 TFfp32. But i don't know if this ever happened in practice.

Nothing notable That I am aware of, But it was still useful I imagine.

 

[Globalisateur]

Never happened cause 4Pro is missing the needed bandwidth and RPM FP16 only helping in limited situations over the big picture.

Maybe it's happening on some games, like some that are running better on Pro ? Some even using the same resolution or very similar. But yes, in most cases, Pro games should be bandwidth starved.

 

[JoeJ]

Yeah, personally i have not used fp16 yet, but i think it will work for both lighting and geometry in my case. (Surfel hierarchy is an approximation anyways)
Optimistic i get more from it than just 20% for certain shaders.

I'm curious how widespread fp16 is at all, and for what it's used.
I assume it's often left behind for cross platform games? (No XB, rare PC support... focus on somethinng else?)

 

[AbsoluteBeginner]

So I've read pretty much entire repo and here are few points :

• Its legit as it can be, and a massive fuck up by intern at AMD (I don't even think this guy knew were this chips will end actually)
• Only one chip from the document has actually been released - Navi 10. All others are pending for release in 20'(MI100, Renoir, Arden, Oberon)
• Oberon is 100% PS5 related
  • Oberon is 36CU in native mode. Not only are there dozens of references to this (18 WGP - full chip result), theoretical texture filtrate firmly points to 288 GT/s (4*36*2GHz)
  • GEN0 and GEN1 matches BW, clocks and CUs for PS4 and PRO
  • Ariel was earlier version of Oberon (perhaps one is dev kit, other retail?)
  • Ariel/Oberon have 256bit bus
  • Total BW for chip went from theoretical 448GB/s (Ariel A0) to 530GB/s (Oberon B0). Indicating 14Gbps > 18Gbps (downclocked?) memory change
• Arden/Sparkman are almost certainly Xbox related
  • Data for them is insufficient, seems like Arden template is copied to both Arden and Sparkman tab
  • Sparkman can therefore either be Lockhart or Arden's "Ariel"
  • Most likely, Sparkman is Lockhart as team that is responsible for Arden and Sparkman are same
  • Arden and Sparkman match Xbox and Xbox One X 4RBs for GEN0 and GEN1 BC data, though Sparkman does NOT include XBX BC
  • Arden/Sparkman has 320bit bus
  • Total BW for chip (theoretical) is 560GB/s, likely indicating same default 14Gbps chip Oberon theoretical BW was calculated
  • Arden/Sparkman have 3584SPs - therefore 56CUs
  • RT and VRS are both referenced in Arden/Sparkman
  • Arden and Sparkman are both town's in Arkansas, USA

My opinion on this - there was clearly work being done on Oberon in June this year. Fact that RT and VRS are in Arden, but not in Oberon does not mean its not on chip. Data points to completely different tests being done on Oberon in comparison to Arden/Sparkman.

From this, I would say I am 99% sure chips specified here will end up in next gen consoles. Tapeout should be close now, and near final chip should have already been here by the end of last year.

Interesting tidbit from well known leaker from Taiwan - Aquariuszi, with likely ties to TSMC fab, points to Oberon and Arden being near finished.

https://www.ptt.cc/bbs/PC_Shopping/M.1570476662.A.7A0.html

Arden
OBR/Oberon
Sparkman

The first one should be on the Xbox.
The second full chance is PS5.

In addition
Arden's progress is relatively fast, and OBR is still in its early stages of verification
And Arden is more than one size larger than OBR (350 vs 300 mm2)

In Oberon's continuous revision and retest, can Holiday 2020 really catch up ...

(Issues with clocks?)

So, if Arden is ~50mm2 bigger then Oberon, and there is a chip named Oberon going into Fab by late 2019, what are the chances leaked Oberon is not that chip? I think chances are slim to none.

 

[HBRU]

Sure left behind on multyplat... but not so much on true exlusives

 

[iroboto]

Maybe it's happening on some games, like some that are running better on Pro ? Some even using the same resolution or very similar. But yes, in most cases, Pro games should be bandwidth starved.

Unlikely. At the highest level of AAA gaming we don't see the 4Pro beating out X1X. And they woudl be the ones to be able to fundamentally afford the time and have the talent to leverage the features to the maximum. Games on 4Pro that beat out X1X is because X1X is just being asked to brute force its way to its target. (since the hardware is technically capable of just 4X XBO settings) Some teams may find it easier to just leave it at that without pushing it further.

RPM is probably used a lot more than we know, and with it's introduction for all consoles next year anyway there is no reason to hold back on it. The feature is not that niche, it's only missing on Xbox but available everywhere else (like AMD and Nvidia cards).