Re: [PATCH v2 35/36] cpu_map: Add Denverton CPU model
On Tue, Nov 26, 2024 at 11:54:45 +0000, Daniel P. Berrangé wrote:
On Fri, Nov 22, 2024 at 10:33:15PM +0100, Jiri Denemark wrote:
> Signed-off-by: Jiri Denemark <jdenemar(a)redhat.com>
> ---
> diff --git a/src/cpu_map/index.xml b/src/cpu_map/index.xml
> index 3d407856f2..8ab5e66a2f 100644
> --- a/src/cpu_map/index.xml
> +++ b/src/cpu_map/index.xml
> @@ -110,6 +110,10 @@
> <include filename='x86_GraniteRapids-v1.xml'/>
> <include filename='x86_SierraForest.xml'/>
> <include filename='x86_SierraForest-v1.xml'/>
> + <include filename='x86_Denverton.xml'/>
> + <include filename='x86_Denverton-v1.xml'/>
> + <include filename='x86_Denverton-v2.xml'/>
> + <include filename='x86_Denverton-v3.xml'/>
> </group>
Fun times.
Currently my Alder Lake laptop has "host model" expanded to SnowRidge.
With this series, it now expands to Denverton.
Neither of these are right. This series merely swaps wrongness.
I believe this kind of thing is unavoidable given that Intel isn't adding
QEMU CPU models for laptop class CPUs, only Xeon class.
Yeah, I don't think we can do much about it given that there's no model
matching the CPU you have so the algorithm is just choosing the closest
one according to a list of features. We can achieve good results by
adding a real CPU signature to a specific model, but there's no right
CPU model to choose from in your case.
Jirka