The AMD Zen-Based Hygon Dhyana CPU Support Landing In The Mainline Linux Kernel

[phoronix]

Phoronix: The AMD Zen-Based Hygon Dhyana CPU Support Landing In The Mainline Linux Kernel

Hygon's Dhyana SoC, the facsimile of the AMD Zen microarchitecture as a result of the AMD-Chinese joint venture to begin spinning up domestic x86 chips for the Chinese data center market, will be supported by the next version of the Linux kernel...

http://www.phoronix.com/scan.php?page=news_item&px=AMD-Hygon-Dhyana-420-Queue

[M@GOid]

Too bad this puppy is not easily available outside China. The curiosity to see how it performs on Linux is big:

[edwaleni]

Keep your eyes open for those surreptitious CPU modules being tossed into the package at the behest of the PLA. /sarcasm.

[bachchain]

Too bad this puppy is not easily available outside China. The curiosity to see how it performs on Linux is big:

Why would you expect it to perform any differently than a normal Zen chip?

[chithanh]

Too bad this puppy is not easily available outside China. The curiosity to see how it performs on Linux is big:

The Subor Z+ APU is a different chip than what the article talks about, and its PCI device ID (0x15FF) is not known to amdgpu either. So it won't perform at all on (mainline) Linux.

[M@GOid]

Why would you expect it to perform any differently than a normal Zen chip?

In the test video it was not the same, on Windows.

[M@GOid]

The Subor Z+ APU is a different chip than what the article talks about, and its PCI device ID (0x15FF) is not known to amdgpu either. So it won't perform at all on (mainline) Linux.

Well, that could change fast if it was easily available outside China.

[chithanh]

Well, that could change fast if it was easily available outside China.

As previous Subor game consoles like the D99 are available on aliexpress with worldwide shipping, I expect the same to happen with the Z+.

However it is also a matter of how AMD and their business partners view things. PS4 Linux developers have reported that AMD Linux folks have been hostile to their efforts, and PS4 is easily available outside China.

[bridgman]

PS4 Linux developers have reported that AMD Linux folks have been hostile to their efforts

Calling it "hostile" is a bit unfair... it's just that our contractual arrangements with the semi-custom customers do not allow us to release any information or support any third-party development for their products. That is a far cry from "hostile".

[chithanh]

it's just that our contractual arrangements with the semi-custom customers do not allow us to release any information or support any third-party development for their products.

I don't know any details about the conversation between the PS4 Linux project and AMD developers, but at the 33C3 talk where I was in the audience, what marcan said gave the impression that the unnamed individual from AMD was actively discouraging PS4 Linux efforts building on radeon/amdgpu driver code rather than simply being indifferent.

That is a far cry from "hostile".

Ok so, let me ask you directly: If there were a proposal for upstreaming PS4 support into amdgpu/radeonsi, would AMD try to block this? Same question about the Subor Z+ APU.

BTW, a similar situation happened with the reverse engineered gma500 driver, it duplicates a lot of code in i915. But Intel folks explicitly did not want the code shared with i915 (for presumably non-technical reasons), so a separate driver had to be made.