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

Status
Not open for further replies.
The chances engineering/quality sample silicon was ready and available for dev kits is very slim. We know historically that dev kits with actual production silicon don't show up until the last year before release.
True. Except if company aimed for 2019 release and had to delay the product for some reason.

We know last gen late 2012 had dev kits with actual silicon (maybe even before for 1st party) so who knows.
 
The chances engineering/quality sample silicon was ready and available for dev kits is very slim. We know historically that dev kits with actual production silicon don't show up until the last year before release.

PS5 was targeting potential H1 2020 release so it's not unusual timeline wise to see final silicon now.
 
HBM2 controllers are much bigger.

Which areas of the footprint of the GPU and HBM stacks in a Vega 7 MCM are you considering to be the memory controllers?
The PHY of a GDDR6 was described in a video earlier in the thread to be 1.5-1.75x bigger for HBM2 of the same bandwidth.
The 256-bit subsystem of Navi takes up ~3/4 of two long sides of the die, while a roughly equivalent two stacks of Vega 7 take up one side, with a gap between them that may or may not be entirely controller-related.
I haven't seen a clear description of which blocks would be the actual memory controller logic, or if Vega's higher-end features and HBCC would exist in similar form for Navi.
 
Which areas of the footprint of the GPU and HBM stacks in a Vega 7 MCM are you considering to be the memory controllers?
The PHY of a GDDR6 was described in a video earlier in the thread to be 1.5-1.75x bigger for HBM2 of the same bandwidth.
The 256-bit subsystem of Navi takes up ~3/4 of two long sides of the die, while a roughly equivalent two stacks of Vega 7 take up one side, with a gap between them that may or may not be entirely controller-related.
I haven't seen a clear description of which blocks would be the actual memory controller logic, or if Vega's higher-end features and HBCC would exist in similar form for Navi.

I've measured out 1024 bit controllers to be bigger than a 64 bit gddr6 controller. I can do a measurement again just to confirm. You're right HBM2 is prolly smaller per bandwidth.
 
Gamescom begins August 20th. I wonder if Sony will share more of the PS5 at the event...

Gamescom: Opening Night Live
Gamescom said:
Will be livestreamed from Cologne on Monday, 19 August 2019 at 8:00 p.m. CEST, 2:00 p.m. EDT and 11:00 a.m. PDT.

The brand new show will feature announcements from major publishers and independent developers as well as world premiers with new footage and guest appearances by stars from the international games industry.
 
Last edited:
If PS5 is launching in late 2020, they will start promoting it in the first half of that year. Possibly again just before GDC in March.
 
I think Sony will start 2020 with PS5 as they did with PS4 in 2013.

PlayStation Meeting in February with Cerny presenting PS5 hardware innards and controller.

E3 will be where the console design is shown. Release date, price, etc.
 
I think Sony will start 2020 with PS5 as they did with PS4 in 2013.

PlayStation Meeting in February with Cerny presenting PS5 hardware innards and controller.

E3 will be where the console design is shown. Release date, price, etc.

Eh?

PS3 launched...
  • Nov. 11, 2006 in Japan.
  • Nov. 17, 2006 in North America.
  • Mar. 23, 2007 in PAL countries.
I doubt they will do a split launch like that in the major markets again, so it's more likely that they'll launch WW just ahead of or at the start of the Western Holiday shopping season (mostly NA).

They transitioned to doing WW launches at that time of the year as the Western Market became more important to their fianancial success than their traditional Japanese market. The Japanese market doesn't have anything like the Western Holiday Shopping season so the PS1 and PS2 basically just released in Japan when the console were ready (Dec. for PS1 and July for PS2). But they released each console at the start of the Holiday Shopping season for the NA market.

PS3 was the weird one with PAL countries getting it in March versus the traditional Holiday season launch (PS1 and PS2 both launched at the start of the Holiday season in Europe). That was likely due to problems with manufacturing due to component shortages.

That said, I think there's a possibility (albeit small) that they might look at NSW's success launching at the start of the year and seek to do something similar. But, I think it's more likely they'll look towards a Holiday shopping season launch.

A holiday season launch could allow them to launch at a higher price (it's going to be supply limited no matter what at that time of the year) then reduce price to the normal price once sales slow down to a sustainable level. That would allow them to potentially offset higher hardware costs.

Regards,
SB
 
All we need is some solid leaks which should be more available by year end or around PSX if they're doing one. By then most third parties would have the more updated kits readying their demos for the Feb PS Meeting reveal and E3 2020, it would be damn near impossible not to leak something.
 
If PS5 is launching in late 2020, they will start promoting it in the first half of that year. Possibly again just before GDC in March.

Which makes sense because why would Sony want to market an unreleased product during the holiday season? Sony wants to maximize PS4 sales during this time and any marketing now for the PS5 isn’t going to have appreciable impact on sales during holiday season 2020.
 
Also, next year they will again provide lots of AAA exclusive PS4 software even though they have to promote PS5. They did that back in 2013, and will have to repeat it in 2020.

What remains to be seen is if they have timed their earlier releases right so that they could actually provide enough nextgen software for launch and 2021.
 
HBM2 controllers are much bigger.
Speaking of which what is the reason for the gap between your earlier GDDR6 memory controller estimate (4.5mm2) and Liabe's graphic 8.16mm2?
Is this because of the empty space which you accounted on the 140mm2?
Per unit BW is the most relevant metric.
Problem with that is that for the same die budget you'll get a smaller but slightly faster memory pool compared to GDDR6
 
Status
Not open for further replies.
Back
Top