Next Generation Hardware Speculation with a Technical Spin [post E3 2019, pre GDC 2020] [XBSX, PS5]

[BRiT]

At least the Taiwanese leak is a fake. AquariusZi told Xbox Series X is 350mm2 and Oberon 300mm2. One fake insider down, next one now.

Okay, the res/cap arrays are almost assured to be spaced based on the underlying BGA pitch to fit with vias (so either 0.40 or 0.50 quantized) and it seems to be 0.40. The space between each little chip in a string is spaced exactly 2.4mm or 2.5mm.

Measuring each string of 7 or 10 as close to the chip,as possible the image has a slight vertical stretch. I averaged the string on each side, with a tiny difference.
Vertical 22.231 pix/mm
Horizontal 21.758 pix/mm

So I get a rather precise 350 mm2.

No that makes no sense. But that's what I have. Maybe it shifts 0.1 per chip to align with a 0.50 bga and fan out. So 2.5 per chip means... 380 mm2.

So does this put the Taiwanese leak from AquariusZi back into the Possibly Legit Leaker and out of the Fake category?

 

[MrFox]

So does this put the Taiwanese leak from AquariusZi back into the Possibly Legit Leaker and out of the Fake category?

Edited... it was either 2.4 or 2.5 to fit with 0.40 or 0.50 quantized.

Much more confident of the 380 mm2 figure, BGAs are almost always 0.50. But it remain two possibilities.

 

[BRiT]

So 350 mm2 or 380 mm2, what's a few 30 mm2 between chips?

Maybe it was the camera, or maybe the stripes or maybe the color. They say some or all of those things can make one look bigger than they are.

 

[MrFox]

So 350 mm2 or 380 mm2, what's a few 30 mm2 between chips?

Maybe it was the camera, or maybe the stripes or maybe the color. They say some or all of those things can make one look bigger than they are.

Well if anyone can measure the BGA pitch of modern GPU from AMD and it's 0.40mm then I would say 350 mm2 is worth a bet. Some sort of bet.

And 0.50 would mean 380 obviously.

 

[Tkumpathenurple]

That chip size leak could still be legitimate.

√350 = 18.71
√380 = 19.49

I think a difference along each axis of <1mm is a reasonable margin of error for eyeballing the thing.

 

[HBRU]

So 350 mm2 or 380 mm2, what's a few 30 mm2 between chips?

Maybe it was the camera, or maybe the stripes or maybe the color. They say some or all of those things can make one look bigger than they are.
[emoji38]

380 mm2 ? Big

 

[DavidGraham]

TBP, not TDP. Very different metric.

Not that different, TBP includes VRAM + GPU + VRM circuitry, and considering that a GPU in a console will still have VRAM + GPU, you can discard VRM circuitry from the equation (Which usually means 30-40w of power), so my expectation of 280w will turn into maybe 240w. Note also if the console has higher clocked GDDR6 memory power consumption will go up even more.

Precisely. The 5700M, which has the same number of CUs, probably similar clocks, but the wider memory bus of its 5700 parents, has a TDP of 120W.

Without knowing the clocks, that's a moot point. Besides, the information in the link is based on rumors, TPU has it at 180w, not 120w.

 

[goonergaz]

It's varying by country and voltage. Having a universal 100v to 240v PSU is more difficult to meet all requirements everywhere from a single design. It's not necessarily just the connector versus the wattage.

Technically if they have different PSU models for US, EU, and Japan, I think they can use the small connector and meet individual country electical code.

Or maybe the laws changed.

Thanks, maybe a law change because the change in power socket on the Pro didn't see a drop in wattage...although DF seem to agree with me.

 

[AzBat]

Interesting animated comparison of XBSX & XB1X:

https://twitter.com/x/status/1214609722105647104

Tommy McClain

 

[chris1515]

So does this put the Taiwanese leak from AquariusZi back into the Possibly Legit Leaker and out of the Fake category?

The leak of AquatiusZi is 350 mm2 for Xbox Series X. All other people track it between 380 mm2 to 40 mm2. It is a huge chip. I think people thinking the PS5 is 36 Cus at 2 Ghz Tflops will be surprise. The PS5 is maybe 9.2 Tflops but not 36 CUs and 2Ghz for example 48 CUs at 1.5 Ghz is a better choice. the Xbox Series X die is so huge I doubt the PS5 will be more powerful maybe the PS5 will be just above 10 Tflops and less powerful than the Xbox Series X but 100 dollars cheaper and probably with a faster SSD.

https://www.gamesradar.com/what-is-...discuss-the-potential-of-the-next-generation/

6 months ago some devs were talking about pretty low frequency for the PS5 GPU. The article is from december 2019 but the devs talks in june 2019.

 

[chris1515]

Jim Ryan told this element will help differentiate the PS5 maybe the raytracing is different than on Xbox Series X.

 

[Karamazov]

The infamous Ray Synthesizer

 

[AbsoluteBeginner]

The leak of AquatiusZi is 350 mm2 for Xbox Series X. All other people track it between 380 mm2 to 40 mm2. It is a huge chip. I think people thinking the PS5 is 36 Cus at 2 Ghz Tflops will be surprise. The PS5 is maybe 9.2 Tflops but not 36 CUs and 2Ghz for example 48 CUs at 1.5 Ghz is a better choice. the Xbox Series X die is so huge I doubt the PS5 will be more powerful maybe the PS5 will be just above 10 Tflops and less powerful than the Xbox Series X but 100 dollars cheaper and probably with a faster SSD.

https://www.gamesradar.com/what-is-...discuss-the-potential-of-the-next-generation/

6 months ago some devs were talking about pretty low frequency for the PS5 GPU. The article is from december 2019 but the devs talks in june 2019.

How are you going to arrange front end for 48CU Navi?

If they can get 2.0GHZ on 36CUs, its most certainly better then 48CU part. It will be smaller, ROPs/cache/CU count fit like a glow and they also perfect fit for PS4/Pro BC. With 48CUs, not so much. Also, performance scale better with clocks and you will get 30% higher pixel fill-rate with 36CU chip.

I think there is a reason we haven't seen 40-56CU part from AMD yet, because it doesn't fit. Rumors about big Navi being 80CUpart, and smaller one being 60CU lend credence to this.

In any case, I would not dismiss this rumor because you can see people counting pixels, measuring and coming up with widely different results. That forum post reads more as an estimate, then actual measuring, so even something like 320mm2/380mm2 would make sense. He has a good history of getting correct info on AMD stuff (Vega, Radeon 7, 250-260mm2 die for Navi 10 back in January etc), I would not just hand wave his info.

As for the dev quote, it makes no sense. Anything below 1.5GHz will be wasted silicon, as this is well below Navi efficiency sweet-spot. From where I am sitting, you either go with 3 x 10WGP (60CUs full chip) or 2 x 10WGP (40CUs full chip). Anything in between will not fit.

 

[chris1515]

How are you going to arrange front end for 48CU Navi?

If they can get 2.0GHZ on 36CUs, its most certainly better then 48CU part. It will be smaller, ROPs/cache/CU count fit like a glow and they also perfect fit for PS4/Pro BC. With 48CUs, not so much. Also, performance scale better with clocks and you will get 30% higher pixel fill-rate with 36CU chip.

I think there is a reason we haven't seen 40-56CU part from AMD yet, because it doesn't fit. Rumors about big Navi being 80CUpart, and smaller one being 60CU lend credence to this.

In any case, I would not dismiss this rumor because you can see people counting pixels, measuring and coming up with widely different results. That forum post reads more as an estimate, then actual measuring, so even something like 320mm2/380mm2 would make sense. He has a good history of getting correct info on AMD stuff (Vega, Radeon 7, 250-260mm2 die for Navi 10 back in January etc), I would not just hand wave his info.

As for the dev quote, it makes no sense. Anything below 1.5GHz will be wasted silicon, as this is well below Navi efficiency sweet-spot. From where I am sitting, you either go with 3 x 10WGP (60CUs full chip) or 2 x 10WGP (40CUs full chip). Anything in between will not fit.

I never said than it will be under 1.5 Ghz, they said it will not be high frequency where did you said this I said 48 CUs at 1.5 GHz. And same thing every rumors with die size of Xbox Series X was far from the reality at the end.

I don't say than the PS5 will not be 9-10 Tflops I think this will be the range of the PS5 performance but not at a stupid 36 CUs 2 Ghz rumors. Imo knowing the power consumption of Navi this is not realistic at all. And it seems Oberon had at least three new steppings since the github leak.

For PS4 Pro for PS4 compatiblity half of the CUs were disabled. They can disable 12 Cus for PS4 Pro BC or 30 for PS4 BC.

There is 99.9% the chance the PS5 is weaker in Tflops side than the Xbox Series X but rumors seems to be false for the moment.

EDT: I think PS5 for factor will be tinier than the Xbox Series X and it will be impossible to have a 36 CUs 2 Ghz Navi GPU inside.

 

[Betanumerical]

If they can get 2.0GHZ on 36CUs, its most certainly better then 48CU part. It will be smaller, ROPs/cache/CU count fit like a glow and they also perfect fit for PS4/Pro BC. With 48CUs, not so much. Also, performance scale better with clocks and you will get 30% higher pixel fill-rate with 36CU chip.

I don't think they have to 100% match the CU count of the PS4 for BC they can disable CU's for BC as far as I am aware, this means they should be fine as long as the CU count is >= 36.

 

[chris1515]

Imo the Ps5 will probably be in the low 10 Tflops, they will probably try the best to push it there for marketing and PR reason but 100 dollars cheaper and a form factor less bulkier than Xbox Series X.

And not a very hot 36 CU 2 Ghz even using a 5700XT 40 CU with 40 CUs at 1.8 Ghz is better or a 44 CUs at 1640 Mhz is better for reach 9.2 Tflops.

 

[anexanhume]

Without knowing the clocks, that's a moot point. Besides, the information in the link is based on rumors, TPU has it at 180w, not 120w.

This is based on historical TDP of the laptop form factor. 120W is a very common thermal envelope for high-end notebooks graphics. Additionally, the desktop 5700 is 150W, so any number above that (or within 10W in my opinion), is completely bogus. 180W is just them quoting the TDP of the desktop 5700XT, which has more CUs and obviously higher clocks. No TBPs invoked in any of these numbers, by the way.

I will also note the inherent problem in quoting mobile numbers, which could potentially create a sense of optimism. These are binned parts for low volume. Consoles won’t have the same luxury.

 

[DavidGraham]

I will also note the inherent problem in quoting mobile numbers, which could potentially create a sense of optimism. These are binned parts for low volume. Consoles won’t have the same luxury.

I agree, that's why my analysis is resting solely on desktop parts.

 

[RDGoodla]

Jim Ryan told this element will help differentiate the PS5 maybe the raytracing is different than on Xbox Series X.

It seems that PS5 uses 100GB Blu-ray disk while Xbox media is not confirmed?

I speculate all of the five features of PS5 are something unique.

 

[anexanhume]

It seems that PS5 uses 100GB Blu-ray disk while Xbox media is not confirmed?

I speculate all of the five features of PS5 are something unique.

The Xbox One X already has a UHD drive. It would be very surprising if Series X didn’t. Especially since Xbox supports Atmos and Sony doesn’t. It’s a legit option for home theater 4K player.