Re: About new backend for GPU compute ROCm in qemu

["Huang, Honglei" <honghuan@amd.com>]

On 9/7/2026 9:05 PM, Dmitry Osipenko wrote:
> On 8/17/26 06:19, Huang, Honglei wrote:
>> Hi Michael, Alex, Dmitry, Akihiko,
>>
>> I'm bringing AMD GPU compute ROCm based on virtio. I posted a ROCm over
>> virtio
>> implementation to virglrenderer nine months ago (MR !1568 [1]). The ROCm
>> side has
>> been supportted by ROCm offical.
>>
>> Current implementation is a virtio gpu context type capset handled inside
>> virglrenderer, sharing the display path. That's an awkward fit, many
>> compute GPUs have no display engine at all.
> 
> The headless context type is totally fine with virtio-gpu. For example,
> ChoromeOS wirh crosvm disables display entirely even on guest side by
> building virtio-gpu kernel driver without KMS support. This is something
> to improve in QEMU to make it work without GL context.
> 

Thanks for the inforamtion.
It's great to know that headless contexts are fine and that ChromeOS 
already has a working model for this. The changes to the ROCM native
context from QEMU are not much, maybe we can leverage the changes made 
from ChromeOS to assist ROCm.

Regaeds,
Honglei