GPU Updates via “Benny”
We just ran across a great synopsis of new GPU product announcements from NVIDIA, AMD and Intel from VMworld last week. Bernhard Tritsch (aka Benny) remarked that the “most remarkable aspect of the individual announcements is that each vendor has found a very unique way to implement GPU-accelerated remoting.”
Benny Summarizes the Following Advantages per Company
According to the author, NVIDIA’s advantages are their extremely powerful GPUs and their well-established GRID concept. Its new Grid v2 is based on high-end GPU cards, a software layer in the hypervisor and a graphics driver for the guests. The graphics resources of the available GPUs can still be assigned to virtual machines in a balanced way. Another new element is a license fee that NVIDIA wants OEM partners to charge customers – to be announced by the OEMs soon.
Benny says that Intel’s selling points are low power consumption and a low price, suitable for low to medium-end remote graphics workstations. This summer Intel began shipping the “Broadwell” APU, fan integrated GPU with 48 cores, demonstrating GPU-accelerated user sessions for the conference.
Finally, he says that AMD is the up and comer in this group that needs to focus on its capability to deliver their multiuser GPU solution as soon as possible. This solution is built around SR-IOV and does not require any software stack on the host side that goes beyond a driver, which sees up to 16 virtual GPUs per physical GPU, with device virtualization and GPU resources managed by hardware. Details about compatible servers and required BIOS specs are forthcoming.
You can read the in-depth article here.
