Dear Ron,
I suggest you to read this document:
https://cwiki.apache.org/confluence/display/CLOUDSTACK/GPU+and+vGPU+support+for+CloudStack+Guest+VMs.
I have tested pass-through mode with XenServer 6.2 and can give probably
a bit more information if you need. The reference document is good
indeed, so you probably do not need my comments :)
Regards,
Vadim.
On 2015-10-26 22:47, Ron Wheeler wrote:
The GPU section on page
http://docs.cloudstack.apache.org/projects/cloudstack-administration/en/4.5/service_offerings.html
[1] seems to be lacking some information (at least as far as I can
tell).
1) What does the GPU do? It says to select a card that I want to use.
No hint about the basis on which I should make the selection or if
there are any restrictions on the choice (based on hardware?)
2) It mentions support for 2 NVIDIA cards and says that if you want to
use another card "make the createServiceOffering API call". How is this
done on a UI screen?
3) The Type selection is worded in an odd way with 3 selections
explicitly named and a "so on" thrown in. Then it goes on to talk about
a passthrough vGPU type in a separate paragraph.
Perhaps there should be links to external web sites to clarify the
definitions of these cards and types.
If someone can explain what this section actually means, I will try to
write it in a clearer fashion.
Ron
Links:
------
[1]
http://docs.cloudstack.apache.org/projects/cloudstack-administration/en/4.5/service_offerings.html