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

[DavidGraham]

It's very hard to argue against the PS5 leak with it's backward compatibility modes, If both systems are close, then they are both around 10TF.

 

[mpg1]

It's very hard to argue against the PS5 leak with it's backward compatibility modes, If both systems are close, then they are both around 10TF.

The thing about the leaks is the CU's. If it's correct I don't see how PS5 is 10Tflops.

More likely is that XSX is a bit under 12 given that the words "targeting 12" have been used. 9.2 and something like ~11.5 might be considered "close"

 

[MrFox]

The thing about the leaks is the CU's. If it's correct I don't see how PS5 is 10Tflops.

More likely is that XSX is a bit under 12 given that the words "targeting 12" have been used. 9.2 and something like ~11.5 might be considered "close"

Yeah, with 36CU there's no way it can be clocked high enough for 10TF.

It's also a small possibility the 36CU for testing compatibility modes were just the BC boost mode. But if that was the case, we would see an 800MHz 18CU mode too.

 

[BRiT]

But if that was the case, we would see an 800MHz 18CU mode too.

We have, have haven't we?

 

[iroboto]

We have, have haven't we?

Yes, I believe the DF video covered this.

 

[Deleted member 13524]

Will be fun to see how these "insiders" with their conflicting info are going to spin it when the specs are known.

Probably nothing because no one in the internet will be keeping tabs for too long.

I do doubt he somehow has final info and everybody else doesn't, though.

Some leakers are actual insiders, probably devs who've been given target specs and actual hardware. Others are data miners who caught some lost log from AMD but realistically can't know for sure what they're looking at.

 

[MrFox]

We have, have haven't we?

Yes, I believe the DF video covered this.

They cover the test clocks, but do they indicate the CU enabled at 800MHz is down to 18?

That would indicated the CU enabled are part of each of the three clocks, therefore a possibility of having more than 36, otherwise it would be more clear the 36CU is the true physical count, regardless of the test setup.

 

[BRiT]

They cover the test clocks, but do they indicate the CU enabled at 800MHz is down to 18?

That would indicated the CU enabled are part of each of the three clocks, therefore a possibility of having more than 36, otherwise it would be more clear the 36CU is the physical count, not the test setup.

Yes. From DFs article covering the leak: https://www.eurogamer.net/articles/...laystation-5-xbox-series-x-spec-leak-analysed

So how can we be confident that this processor is actually a semi-custom AMD product for Sony and not another partner like Microsoft? The giveaway is the fact that the GPU can be switched to three different modes in order to provide hardware backwards compatibility with PS4 and PS4 Pro. While a 2.0GHz GPU clock is used for what is described as the fully unlocked 'native' or 'Gen2' mode, the processor is also tested in what is referred to as Gen1 and Gen0 modes. The former is explicitly stated as running with 36 compute units, a 911MHz core clock, 218GB/s of memory bandwidth and 64 ROPs - the exact specifications of PlayStation 4 Pro. The latter Gen0 mode cuts the CU and ROP counts in half and runs at 800MHz, a match for the base PS4. The indications are that back-compat is an integral part of the silicon, which in turn raises some interesting questions about the makeup of the Navi GPU and the extent to which older GCN compatibility may be baked into the design.​

 

[Betanumerical]

They cover the test clocks, but do they indicate the CU enabled at 800MHz is down to 18?

That would indicated the CU enabled are part of each of the three clocks, therefore a possibility of having more than 36, otherwise it would be more clear the 36CU is the physical count, not the test setup.

The leaks indicate that at 800mhz theres only 18CUs enabled.

 

[wco81]

Do you suppose Sony and MS are reading the tea leaves like people are doing on forums like this one, to divine the capabilities and specs of these systems?

Or are the specs likely finalized and probably too late to change?

 

[BRiT]

Do you suppose Sony and MS are reading the tea leaves like people are doing on forums like this one, to divine the capabilities and specs of these systems?

No. They know what they can build with AMD and what it would cost them. No need to go anywhere else about capabilities and specs.

 

[AbsoluteBeginner]

It would also be good to keep in mind how long it takes from chip A0 tapeout to full production, because people saying May 2019 is old data for console coming out in October/November 2020 might think again.

 

[Betanumerical]

It would also be good to keep in mind how long it takes from chip A0 tapeout to full production, because people saying May 2019 is old data for console coming out in October/November 2020 might think again.

Arden and Sparkman don't even have silicon results in the leaks, they appear to be all theoretical values with no actual measured performance. How long roughly would it take to tapeout from A0/B0 to full production?.

 

[Nisaaru]

The "obvious" solution to the Sony's CU crysis is that they are double pumped

P.S. Let's see if that reaches the desperate at resetera/neogaf

 

[BRiT]

4Ghz and beyond!

 

[AbsoluteBeginner]

Arden and Sparkman don't even have silicon results in the leaks, they appear to be all theoretical values with no actual measured performance. How long roughly would it take to tapeout from A0/B0 to full production?.

~12 months I'd say, probably even longer for 300+ mm² APUs on 7nm.

The fact that Arden A0/B0 results are not there does not mean there wasnt A0 stepping of a chip at the time. Certainly, both Ariel and Arden were added to PCI ID repo in January at similar time.

Sony obviously had chip earlier, as V shaped dev kits have been with developers since ~July, so they are probably ahead by few months vs where they were with PS4.

Going by what Albert Panello said, tapeout of XSX must be close now. I assume same for PS5 is true.

This is why I think if May was B0 stepping of Oberon, and V shaped dev kits were sent do developers in July, that must be pretty much a final chip.

 

[AzBat]

I heard PS5 will be stronger, cheaper, and also run PC and Xbox games.

So the PS5 is just another Xbox SKU? Xbox Series P? For PC & PS?

Tommy McClain

 

[Proelite]

The 18WGP data is under Gen2 2ghz. All the results under that category are indicates peak performance of the GPU. Numbers are in between 5700 and 5700XT.
All the results for the m100, renoir, oberon, ariel, arden, sparkman have peak perf data, and in cases of Xbox / PS5, BC test results.

It's puzzling to me why the test results, which are comprehensive, would be missing only peak theoretical performance for Oberon, not for anything else.

Also in BC2 modes, the unused GWP are clearly marked as unused, if 2ghz is BC boost mode, I would expect the "missing" WGPs to be there and marked as unused.

Putting it frankly, I see no evidence that AMD hid the peak theoretical data for Oberon / Ariel. The most logical and probable explanation is that Oberon is 18WGP active total.

I can see Sony making all 20 WGPs active if the yields are great. Imo that's the only path for PS5 to be > 10TF.

Clocked at 2.15ghz, they might be able to hit 11TF.

 

[Betanumerical]

The 18WGP data is under Gen2 2ghz. All the results under that category are indicates peak performance of the GPU. Numbers are in between 5700 and 5700XT.
All the results for the m100, renoir, oberon, ariel, arden, sparkman have peak perf data, and in cases of Xbox / PS5, BC test results.

It's puzzling to me why the test results, which are comprehensive, would be missing only peak theoretical performance for Oberon, not for anything else.

Also in BC2 modes, the unused GWP are clearly marked as unused, if 2ghz is BC boost mode, I would expect the "missing" WGPs to be there and marked as unused.

Putting it frankly, I see no evidence that AMD hid the peak theoretical data for Oberon / Ariel. The most logical and probable explanation is that Oberon is 18WGP active total.

I can see Sony making all 20 WGPs active if the yields are great. Imo that's the only path for PS5 to be > 10TF.

Clocked at 2.15ghz, they might be able to hit 11TF.

It could be for the exact same reason that the repo is missing the silicon and measured results for Arden and Sparkman. Or the same reason the repo contains partial results for PS4 Pro SoC (Gladius). Or the same reason the Navi10 testing isn't even finished in the spreadsheets. Just because, the repo contains only partial data on a number of things that it really should have.

 

[Proelite]

It could be for the exact same reason that the repo is missing the silicon and measured results for Arden and Sparkman. Or the same reason the repo contains partial results for PS4 Pro SoC (Gladius). Or the same reason the Navi10 testing isn't even finished in the spreadsheets. Just because, the repo contains only partial data on a number of things that it really should have.

Repo has comprehensive test results for Oberon across many tests.
The repo is one that helped further confirm what I had already knew in October / November. I already made a few wagers on the 36/40 CU prior to this.