So what would be the roadmap for shrinking the PS4 APU enough that a handheld device might be plausible?
If they were to do it, you'd think they'd want to launch a year before the PS5 launch, which could be 2019, 2020 or 2021?
Can there be a portable chipset delivering PS4 performance a year before those potential launch dates?
I'd put money on a 2019 PS4 portable (and wireless PSVR2) and a 2020 PS5.
7nm has entered mass production this year, which makes it viable for a console next year. This year, on 14/16nm, AMD have already released a 1.3TF laptop APU with 11CU's, so 1.84TF with 18CU's is very doable on an improved process.
The question is memory. HBM2 and above could easily achieve the requisite bandwidth with pretty low power draw, but it's quite expensive. I recall MrFox stating that GDDR6 would be viable - it's certainly cheaper.
It would still have laptop levels of power draw, rather than mobile, so the form factor and cooling solution is still anyone's guess.
I'd love to see them release the following in 2019:
- Wireless PSVR2, with a new PlayStation camera, which connects via HDMI 2.1 and USB 3.1 type-c. Existing PS4's use a breakout box for connecting said camera.
- Move 2 controllers. Pretty much a split DualShock 4. We saw a Sony patent emerge a little while ago for something like this.
- Stress that the PSVR2, PS camera 2, and Move 2 controllers are all designed with the PS5 in mind.
- PS4Pro Slim. Slimmer, quieter, with a UHD Blu-ray drive. Dedicated HDMI and USB ports for the camera or existing PSVR breakout box.
- PS4 Micro shell. A dock with an HDD and Blu-ray drive. Same dedicated HDMI and USB ports for the camera or existing PSVR breakout box.
- PS4 Portable shell. A screen and storage, with a MicroSD card slot. HDMI out, as well as the above VR-dedicated HDMI and USB ports.
- PS4 processing cartridge. Sold in bundles with either of the above shells.