Microsoft handheld Xbox console rumors spring up again as AMD Van Gogh APU appears in Linux kernel code with a potential graphics boost

[Redaktion]

Linux kernel code has revealed some details about AMD's Van Gogh Ryzen 5000 ULV mobile APU series. Apparently, the processors will offer support for 256-bit DDR5 memory, which is a step up from Renoir APUs. Unsurprisingly, the appearance of an AMD APU with greater potential graphics processing has led to renewed Microsoft handheld/Xbox portable console rumors.

https://www.notebookcheck.net/Microsoft-handheld-Xbox-console-rumors-spring-up-again-as-AMD-Van-Gogh-APU-appears-in-Linux-kernel-code-with-a-potential-graphics-boost.528206.0.html

[neblogai]

Lets hope it comes to laptops (and is actually used by laptop manufacturers), not just in MS console. An ultra portable laptop with this chip would be perfect for me, if placed in a laptop with it's top 18W SPL, so that iGPU can boost properly.

[Moravid]

Van Gogh was leaked from an AMD CPU and APU roadmap, custom SoCs don't show up there so highly unlikely it's for an Xbox handheld

[neblogai]

It depends- it can happen that laptop manufacturers are not interested in actually using it, but MS is. Also- I think now we have seen (3?) codenames for it (Aerith, Van Gogh, and now this 'Chachani')- which should indicate intention for multiple platforms.

[Novasean]

AMD can't produce enough chips for the Current demand for the AMD based Laptops currently  "available".  If you add Dell XPS and more mainstream lines.. Every laptop order will say "ships in 12+ weeks" like Lenovo's do now.

[ZODD]

I'll believe it when I see it, but would love to have one if true.
Will be pleasantly surprised if the dive into the handheld market but not holding my breath over this rumor

[Wereweeb]

Seriously doubt a console will come out of this. Also, doesn't 256-bit imply quad-channel memory? Why not put that in the article so it's easier for others to understand the significance and tradeoffs this will have?

[opelit]

"Also, doesn't 256-bit imply quad-channel memory?"

Actually, OCTA-channel. Its LPDDR5. Till they will only allow for 64bits memory dies. But its still Low Power.

If previous leaks are correct, 4 CU and 4 cores even with such wide memory should be fine for the 9W. Especially its RDNA2.

[Spunjji]

If previous leaks are correct, 4 CU and 4 cores even with such wide memory should be fine for the 9W. Especially its RDNA2.

Do 4 CUs actually need 256bit LPDDR5, though? that doesn't seem to make much sense when 8 CU Vega made do with 128bit LPDDR4.

This doesn't really add up, to me, unless it's at least 8CU of RDNA 2 (or more).

[opelit]

I have the same feeling.
But maybe RDNA2 needs the bandwidth to works good compared to Vega. And without the huge "game cache" which normal big Radeons have (the 96MB cache they have) they applied such wide memory to patch that hole.

But imo, It should be better to use the die space to put a little (let say 16MB for GPU/CPU) cache instead forcing to use 4 chips of 64bit memory or even 8 chips of 32bit memory. And that alone suggest it's not a chip for a mobile console as... simply there is no space for so many die chips.