Baseless Next Generation Rumors with no Technical Merits [pre E3 2019] *spawn*

Status
Not open for further replies.
ultragpu said:
I wonder which one favors the GPU space more, interesting rumors.
Click to expand...

If we take those pastebins as being from the same source, it says pretty clearly the PS5 has the better GPU, and the Xbox Scarlet has the better CPU. But I think that's probably just someone glomming onto the Zen 3 BS with a side of Sony Navi exclusivity.
 
I'm curious if we'll get DF article about the hardware this time. I was critical of some of the Xbox stuff coming from MS but appreciated DF conducting the interview. It would be pretty cool if we see both companies take slightly different approaches to allocating their resources so we can see the impact on performance. If Sony goes for streaming with less system memory and MS delivers more but slower memory for example, we can judge the application of those choices.
 
Shortbread said:
[...]the supposed APU sizes don't seem plausible towards reasonable wattage/TDP in the console space [...]
Click to expand...

Yeah, and unless the yield and wafer cost rumors that were thrown around for 7nm DUV are totally wrong such huge dies seem unlikely for consoles from as cost standpoint as well. The estimations I've seen for the wafer price of TSMC so far are 4000-6000 USD per wafer for 16nm and 12000-14000 USD per wafer for 7nm.

If we assume a yield of 90% (which seems unrealistic for such a huge chip on a new node) then we would get:

Code: 
16nm
4000 dollar wafer price -> 525mm² with 90% yield = ~42 USD per die
6000 dollar wafer price -> 525mm² with 90% yield = ~63 USD per die

7nm
10000 dollar wafer price -> 525mm² with 90% yield = ~105 USD per die
12000 dollar wafer price -> 525mm² with 90% yield = ~126 USD per die
14000 dollar wafer price -> 525mm² with 90% yield = ~147 USD per die

I included the 10000 dollar per wafer because the demand for 7nm DUV from Apple and other customers seems to have been lower then expected which could mean TSMC offers better prices to fill the capacity.

Considering that there are rumors that AMD only get's 70% yield for their tiny ~74mm² chiplets and that the defect density (#/sq.cm) is 0.3 it could look worse (for Vega 20 this would mean a yield of around ~40%). Let's say with yield boosting measures and one additional year of process maturity they have a yield of 70% then it could look like this:

Code: 
7nm
10000 dollar wafer price -> 525mm² with 70% yield = ~134 USD per die
14000 dollar wafer price -> 525mm² with 70% yield = ~189 USD per die

10000 dollar wafer price -> 525mm² with 40% yield = ~236 USD per die
 
Shortbread said:
Bored, thought I search some Reddit pastebin speculation and nonsense.

Speculation wise, I could see another Sony/Rambus partnership happening, not so much the HBM3 (maybe HBM2 for such a memory setup). And the last pastebin seems not out of this world, however, the supposed APU sizes don't seem plausible towards reasonable wattage/TDP in the console space (I could be wrong though).

https://pastebin.com/dd1WUpjb (2 or more of these)
No salt required.

- Both are monolithic APU designs.
- Midrange CPU desktop clocks.
- Navi is an architecture. RDNA is a philosophy towards a more flexible and customizable core implementation design. RDNA can be implemented in prior or undisclosed architecture designs outside of the Navi architecture.
- MNXB: CPU is multitier architecture oriented for large AI and machine learning datasets. Larger L1 cache.
- MNXB: CPU takes up a good portion of the APU design.
- MNXB: GDDR5 memory (Hynix) and PCIe 4 busing system.
- MNXB: GPU sparse rendering acceleration.
- SPS5: GPU has a distinct advantage in feature sets and performance.
- SPS5: HBM3 memory and busing system (Rambus partnership)
- SPS5: SSD advantage in terms of performance.

https://pastebin.com/12Yj1KT8
System 1 APU ~525mm²
System 2 APU ~534mm²
More to come.
prior info https://pastebin.com/09ddeXvn

https://pastebin.com/09ddeXvn
Both the core systems have monolithic APU designs. One favoring more CPU space usage for larger cache sizes and the other favoring more GPU space for an additional but smaller non-AMD GPU based co-processor. More to come.
Click to expand...
How do you even find those pastebins ? I can't find the last one using the search function.
 
Globalisateur said:
How do you even find those pastebins ? I can't find the last one using the search function.
Click to expand...

The ones I ever posted are links from replies from Reddit, Era, and I hate to admit it 4chan. But you can also use the magnifying glass search at the top of Pastebin (see pic).

hmgwvmb.jpg
 
You got to love it... o_O:runaway::LOL:

https://pastebin.com/axr3gvaR

Third party developer here with contacts at AMD from the old days. Just got our hands on the dev kits and I reached out to my contacts for some backchannel info. Here is what I've got:

PS5 is running on a semi-custom Navi at around 10TF. Performs well but actually squeezing power out of it takes a bunch of work. It's sort of a mess to deal with. Sony really wanted this card and was willing to overlook its flaws.

XBX is running on semi-custom Vega. More power at around 12TF, but obviously using Vega. MS really put a ton of work into it and it looks to be significantly outperforming the Navi out of the gate. From what I hear, the MS team passed on Navi because they felt that they could customize and optimize the Vega and not deal with the headaches that Navi would create. The passed on it.
 
Yes @Shortbread I know about the search function but the last one you posted is hard to find, the guy who wrote it must have linked it himself ! I thought it was you ! ;-)

The one I personnaly prefer is this one. It's a gem:

Personally, let me tell you Anaconda is literal beast mode. We managed to squeeze 4k/60fps out of this system with RT enabled in a short period of time. Also, in some cases Lockhart is outperforming PS5 (when RT and other effects are enabled). PS5's ray tracing solution is a software/hardware hybrid and it takes a HUGE hit on the system when turned on. Expect most AAA to keep it turned off, IMO.

SSD loading speeds go in this order: Anaconda > PS5 = Lockhart
Click to expand...

https://pastebin.com/YQuqh7Mv
 
Globalisateur said:
Yes @Shortbread I know about the search function but the last one you posted is hard to find, the guy who wrote it must have linked it himself ! I thought it was you ! ;-)

The one I personnaly prefer is this one. It's a gem:



https://pastebin.com/YQuqh7Mv
Click to expand...

Oh, sorry didn't understand question quite clear, but no I didn't post that last one. Or did I... :yep2::no::yep2::nope:

Seriously, my speculation would include more meaningful things towards conversations of bandwidth improvements and future CB (checkboard) usage. Not the pointless ongoing war of flops...
 
AzBat said:
Considering the dual SKU rumor for Xbox I would say Xbox.

Tommy McClain
Click to expand...
bgroovy said:
If we take those pastebins as being from the same source, it says pretty clearly the PS5 has the better GPU, and the Xbox Scarlet has the better CPU. But I think that's probably just someone glomming onto the Zen 3 BS with a side of Sony Navi exclusivity.
Click to expand...
You know what I'm gonna officially tap out at least until MS' conference finishes, these never ending pastebin rumors are killing me softly here.
 
Shortbread said:
Bored, thought I search some Reddit pastebin speculation and nonsense.

Speculation wise, I could see another Sony/Rambus partnership happening, not so much the HBM3 (maybe HBM2 for such a memory setup). And the last pastebin seems not out of this world, however, the supposed APU sizes don't seem plausible towards reasonable wattage/TDP in the console space (I could be wrong though).

https://pastebin.com/dd1WUpjb (2 or more of these)
No salt required.

- Both are monolithic APU designs.
- Midrange CPU desktop clocks.
- Navi is an architecture. RDNA is a philosophy towards a more flexible and customizable core implementation design. RDNA can be implemented in prior or undisclosed architecture designs outside of the Navi architecture.
- MNXB: CPU is multitier architecture oriented for large AI and machine learning datasets. Larger L1 cache.
- MNXB: CPU takes up a good portion of the APU design.
- MNXB: GDDR5 memory (Hynix) and PCIe 4 busing system.
- MNXB: GPU sparse rendering acceleration.
- SPS5: GPU has a distinct advantage in feature sets and performance.
- SPS5: HBM3 memory and busing system (Rambus partnership)
- SPS5: SSD advantage in terms of performance.

https://pastebin.com/12Yj1KT8
System 1 APU ~525mm²
System 2 APU ~534mm²
More to come.
prior info https://pastebin.com/09ddeXvn

https://pastebin.com/09ddeXvn
Both the core systems have monolithic APU designs. One favoring more CPU space usage for larger cache sizes and the other favoring more GPU space for an additional but smaller non-AMD GPU based co-processor. More to come.
Click to expand...

Well since this is the baseless next gen rumors thread...

IF that was true, man it'll be hell on multiplatform developers trying to maintain some kind of parity in releases.

Sort like the PS3/X360 days. Boy did the multiplatform devs have it easy this generation. :)

Regards,
SB
 
Status
Not open for further replies.

Similar threads

BRiT
  • Locked
Baseless Next Generation Rumors with no Technical Merits [post E3 2019, pre GDC 2020] [XBSX, PS5]
448 449 450
Replies
9K
Views
714K
Shifty Geezer
Shifty Geezer
Back
Top