Next Generation Hardware Speculation with a Technical Spin [2018]

[Shifty Geezer]

It’s a patent application, so the single image is the only thing disclosed.

The whole thing is disclosed and described. Just needs someone fluently in Japanese to translate the pages describing the diagrams. I can see references to 5a in the what I take to be the claims - [00xx] statements.

 

[anexanhume]

The whole thing is disclosed and described. Just needs someone fluently in Japanese to translate the pages describing the diagrams. I can see references to 5a in the what I take to be the claims - [00xx] statements.

Where are you seeing the whole text? I can only see the abstract and claim 1.

Edit: never mind I see it. Digging in now.

Edit 2:

Ok, it describes the assembly as a system in package where 5c and 5d can be stacked die. 5a is explicitly a thermal pad. It says 11 can be filled with copper paste OR thermal grease in case EMI is a concern. It says heat sinks can be placed on both sides of the device, or alternatively only the bottom as depicted. Finally, 9 is not placed on the package, but rather in close proximity. It gives examples of it being an antenna, sensor, or storage device.

Based on the mention of antennas, sensors, and EMI concerns, I believe this could be a SoC for PSVR2.

 

[MrFox]

So it's a rather generic patent for getting heat through the other side of a PCB. It has examples with normal via stiching, rectangular holes filled with heat-conductive material, and even the heatsink itself poking through.

It's not specific to any sort of chip or stacks, but it mentions cpu and gpu. Logically it relates to both improving heat removal performance, and reducing the form factor.

I say it's ps4 portable. Because that's what I chose to believe.

Seriously, everything about it looks like it's for a portable device, not a large gpu. It even mention soldering the heatsink to the board, which would be fine only for a very small and light heatsink.

 

[Deleted member 13524]

Based on the mention of antennas, sensors, and EMI concerns, I believe this could be a SoC for PSVR2.

Why would psvr2 need a SoC? Sounds like bad design from the get go..

 

[MrFox]

Why would psvr2 need a SoC? Sounds like bad design from the get go..

For the foveated processing to the two 8K screens.

 

[anexanhume]

Why would psvr2 need a SoC? Sounds like bad design from the get go..

Sounds like you’re assuming it has to be in the headset. The breakout box for PSVR has a processor with RAM and a large aluminum heatsink.

https://www.ifixit.com/Teardown/PlayStation+VR+Teardown/69341

If PSVR2 is wireless as other patents suggest, it will need an antenna(s) as well. You can’t really integrate the BOB inside the PS5 if you need LOS to your headset to broadcast, nor would you want to bake that cost into the PS5.

For the foveated processing to the two 8K screens.

I would think you need to do this while rendering the frame. Otherwise, you’re rendering stuff you’re guaranteed to throw away, and a processing burden on a battery powered device. You’d need a fast way to get the necessary positional data from the user, though. I’ve not read the patents enough to know what their approach is.

 

[Shifty Geezer]

Haven't Sony said they're not doing another handheld?
https://www.eurogamer.net/articles/2018-09-20-no-plans-for-a-vita-successor-sony-says

But a break-out box VR thing makes little sense too as the size of that isn't restrictive. Either PSVR 2 is actually a standalone headset, no console required, or this patent is for something else. Not a handheld... PS5 is a tablet computer confirmed!!

 

[anexanhume]

Haven't Sony said they're not doing another handheld?
https://www.eurogamer.net/articles/2018-09-20-no-plans-for-a-vita-successor-sony-says

But a break-out box VR thing makes little sense too as the size of that isn't restrictive. Either PSVR 2 is actually a standalone headset, no console required, or this patent is for something else. Not a handheld... PS5 is a tablet computer confirmed!!

You think people aren’t going to care about the size of a device they have to place in line of sight, and likely don’t want to move repeatedly for losing calibration or optimal placement?

 

[MrFox]

It would receive a compressed sparse pixel field, and need to convert it back to a cartesian display controller, with good filtering. I was joking about 8K but so far foveated render needs processing directly on the panel controller to avoid ridiculous bandwidth from the console to the screen, it must be expanded as late as possible.

For wireless, the 60ghz chipsets are still very very expensive. It won't hit a reasonable price point soon enough. But would be nice of course.

 

[Shifty Geezer]

You think people aren’t going to care about the size of a device they have to place in line of sight, and likely don’t want to move repeatedly for losing calibration or optimal placement?

Such a box doesn't need to be as small as a mobile phone or portable USB speaker. It shouldn't need a fancy, tiny, extra expensive cooling solution. What's wrong with the current breakout box that means it, or something of its size, woudln't work for PSVR2?

 

[anexanhume]

Such a box doesn't need to be as small as a mobile phone or portable USB speaker. It shouldn't need a fancy, tiny, extra expensive cooling solution. What's wrong with the current breakout box that means it, or something of its size, woudln't work for PSVR2?

The problem lies with the dissipation of the main die in the SoC/SiP. Fancy tiny and expensive is pretty much up the average VR user’s alley.

For example, some iPad SoCs place the memory off package so they can fit a heatspreader directly on top of the die.

 

[Shifty Geezer]

For example, some iPad SoCs place the memory off package so they can fit a heatspreader directly on top of the die.

To fit in an iPad. Again, what is wrong with the current BOB design that means it's not a suitable form factor for a PSVR2? How small and powerful is a PSVR2 BOB going to have to be?

 

[anexanhume]

To fit in an iPad. Again, what is wrong with the current BOB design that means it's not a suitable form factor for a PSVR2? How small and powerful is a PSVR2 BOB going to have to be?

Very simply, they may be convinced the form factor is important and that it would be undesirable without these mitigations.

If it is in the headset, the argument becomes obvious, no?

 

[MrFox]

Very simply, they may be convinced the form factor is important and that it would be undesirable without these mitigations.

If it is in the headset, the argument becomes obvious, no?

The break out box is there primarily as a splitter to TV, the small amount of processing is rather secondary.

 

[anexanhume]

The break out box is there primarily as a splitter to TV, the small amount of processing is rather secondary.

And it becomes a lot more complicated when it’s suddenly responsible for tracking player movements so it can determine the optimal beam forming to ensure a reliable wireless connection. The comparatively simple gen 1 box still had a giant heatsink in it.

The GaAs or GaN transmit amp in there is probably a few watts on its own.

 

[MrFox]

And it becomes a lot more complicated when it’s suddenly responsible for tracking player movements so it can determine the optimal beam forming to ensure a reliable wireless connection. The comparatively simple gen 1 box still had a giant heatsink in it.

The good reasons I can think of to process data locally is for bandwidth reduction or extremely latency-sensitive feedback loops. Compression for wireless could be a big one. But if you need a big heatsink it means you also have a big battery on the waist? Might as well stay wired amd much lower cost. Wireless VR is a big problem to solve.

 

[Deleted member 13524]

Sounds like you’re assuming it has to be in the headset. The breakout box for PSVR has a processor with RAM and a large aluminum heatsink.

https://www.ifixit.com/Teardown/PlayStation+VR+Teardown/69341

No, I'm assuming PSVR2 will have the same amount of external-to-console dedicated processors as Vive, Rift or any Windows Mixed Reality headset: none.

If PSVR2 is wireless as other patents suggest, it will need an antenna(s) as well. You can’t really integrate the BOB inside the PS5 if you need LOS to your headset to broadcast, nor would you want to bake that cost into the PS5.

And none of that needs a heatsink, much less an exquisite setup like the one shown in the patent.

New Haven't Sony said they're not doing another handheld?

Famitsu in the meanwhile updated the news article clarifying that Sony didn't have anything to announce about a new handheld, not that they weren't making any new handhelds.
https://www.famitsu.com/news/201809/20164436.html

Can you ask about deployment of portable machines such as successor to PlayStation Vita?

Oda At the moment, I do not have a plan of presentation about the new model of the portable machine (it is the plan at the present moment only). Playstation Vita continues production until 2019 in Japan, and shipment will be completed thereafter.

Detailed articles including other answers will be delivered soon.

[At 14:30 on September 20, 2018, I received an additional contact from the manufacturer and added it to my invitation]
"About the new model of portable devices, we do not have plans for presentation at the moment."

Unfortunately, eurogamer (or any western publication I found) failed to follow up on this.
If anything, Oda's statements all but confirm the existence of a handheld in the works.

Don't you dare shatter my dreams of a PS4 Go, Shifty!

 

[MrFox]

Anyway the upcoming ps4 portable, which I am sure exists, cannot be considered a successor to the Vita.
So the journalists reporting there are no vita successor are not wrong.

 

[anexanhume]

No, I'm assuming PSVR2 will have the same amount of external-to-console dedicated processors as Vive, Rift or any Windows Mixed Reality headset: none.



And none of that needs a heatsink, much less an exquisite setup like the one shown in the patent.



Famitsu in the meanwhile updated the news article clarifying that Sony didn't have anything to announce about a new handheld, not that they weren't making any new handhelds.
https://www.famitsu.com/news/201809/20164436.html



Unfortunately, eurogamer (or any western publication I found) failed to follow up on this.
If anything, Oda's statements all but confirm the existence of a handheld in the works.

Don't you dare shatter my dreams of a PS4 Go, Shifty!

Again, the breakout box already has RAM and a heat sink. The power dissipation is only going to go up with increased signal bandwidth and now a beamformed stream to the headset. It doesn’t matter if there’s any additional signal processing in the box, nor have I stated as such. The above alone is enough to justify the solution presented.

 

[Deleted member 13524]

Again, the breakout box already has RAM and a heat sink.

The SoC in the breakout box is overkill as it only exists for audio processing according to Sony. Even the reprojection is done on the Liverpool/Neo GPU.
Regardless, the box in PSVR is an over-engineered solution probably made to somehow compensate for the low-end CPU in 2013's PS4, and designed well before any current-gen VR game even existed.

The Rift, the Vive or the WMR headsets don't need a breakout box for signal processing.

The power dissipation is only going to go up with increased signal bandwidth and now a beamformed stream to the headset. It doesn’t matter if there’s any additional signal processing in the box, nor have I stated as such. The above alone is enough to justify the solution presented.

There's no need for a fan in the currently existing 60GHz transmitters/receivers like the Vive Wireless Adapter or the TPLink.
Why would future systems need heat dissipation and why would they even need a heatsink that dissipates a chip on the other side of the PCB?