Next Generation Hardware Speculation with a Technical Spin [pre E3 2019]

Status
Not open for further replies.
I think Navi GCN mid-end GPU in some form is a very realistic assumption, with the rumors in mind. Arcturus seems a very intresting chip, heres hope this will land in PS5. If not that atleast in my future pc.
 
http://www.guc-asic.com/upload/media/event/PDF/HBM_PHY_Controller_2018Q1.pdf

There is a company selling memory controller IPs promising hbm2 at 2.8gbps on 7nm for Q2 2019 (current IPs are 2.4 on 7nm, 2.0 on 16nm) That's a nice 716GB/s with only 2 stacks.

Once they get cheaper non-silicon interposers working with HBM, I wonder what the cost difference would be between two stacks HBM and 384bit gddr6.

I have questions:

What are the thermal management considerations of having the processing elements and the memory in such close proximity as opposed to spread out over a wider area?

What is the effect of different package sizes of HBM when it comes to presenting a level surface for good contact with the heatsink? I remember some inconsistencies with the early Vega cards based on from where the HBM2 dies were sourced (or was it where the final GPU was packaged?).
 
Last edited:
I remember some inconsistencies with the early Vega cards based on from where the HBM2 dies were sourced (or was it where the final GPU was packaged?).
Both.
Vega 56 used slower SK Hynix memory and was packaged without a resin filling around the dies. Vega 64 used faster Samsung memory and used the filling for higher/safer mechanical resistance. They were packaged at different places.

When custom AIB versions started coming out, things got mixed (Vega 56 with Samsung HBM, SK Hynix HBM reaching 1.9 Gbps, etc).
 
True, that is if PS5 will be BC with spiderman then you could be playing it, with HBM memory :D
HBM offers massive bandwidth, akin to PS2's for its time?
 
Both.
Vega 56 used slower SK Hynix memory and was packaged without a resin filling around the dies. Vega 64 used faster Samsung memory and used the filling for higher/safer mechanical resistance.

I like this filling idea, let's brand it the LurpaK method, would go nice with some Hovis for a tasty PR piece

Next we can implement real thermal paste and call that Colman's (English mustard brand for non UK members) for that piquant finish to the new chip.

Perhaps I need more coffee and more importantly breakfast.
 
3.2Ghz boost clock/1Ghz base. Rather big difference between base and boost? Navi10LITE, mid range Navi i assume, even though LITE sounds like low power but why do that for a GPU?
Nice with 8 cores @3.2Ghz, if those are full cores and not like last time, that will probally the biggest jump over current gen.
 
What do the numbers mean, Mason.

View attachment 2832

xCO90g0.png
 
Ok ok that explains alot. 1.6ghz as base means the cpu will not draw much power in idle etc, but has grunt when needed (3.2ghz). No idea what todays Ryzens have for operating speeds, but 3.2ghz for a x86 console isnt too shabby.
 
So Sony said fuck the CPU advances, give me 2x Jaguar cores, run them at 1.6 Ghz base clock, with boost speeds to 3.2 Ghz for Patched games for the PS5.

Riiiiiiiiiiiiiiiiiiiiiiiiiiiiiiight...
 
Do you remember the Cerny patent about a future system with backwards compatibility? Well... I am sure that the idea is that CPU in PS4 mode underclocks itself to 1.6 Ghz.
 
So Sony said fuck the CPU advances, give me 2x Jaguar cores, run them at 1.6 Ghz base clock, with boost speeds to 3.2 Ghz for Patched games for the PS5.

While GPU power is more important to gaming, i also doubt they wont abandon jaguar, hope they wont use it again atleast. Or sony will use the jaguar, aside from a ryzen cpu. The jaguar can double for BC and background tasks etc,
 
Status
Not open for further replies.
Back
Top