Virtu testing
We used version 8.15.10.2321 for the Intel HD Graphics and, keeping consistency with results further on in the review, Catalyst 10.12 for the discrete Radeon HD 6950 2GB. Virtu 1.1.101.16618 was used, too.
What you're looking at is the performance of a Radeon HD 6950 2GB card when run connected directly to the monitor and without any Virtu software installed, the same card with Virtu installed and the display connected to the HD 6950, and the display connected to the Z68's IGP DVI port and the HD 6950's data transmitted over the Sandy Bridge chip's frame-buffer. Got that?
We see that there's no difference in performance when comparing the card connected directly or via Virtu. Running it through the IGP incurs a slight performance loss as the output is mapped on through the Sandy Bridge graphics.
The performance loss is most acute when running Metro 2033; there's clearly something wrong when moving data from the HD 6950, over the PCIe bus, and to the IGP.
But the performance wouldn't be worth much if the IGP's QuickSync mojo couldn't be used at the same time; that's what Virtu is all about.
There's a very minor loss in performance when running the IGP through the Radeon's outputs, and we could certainly live with the one-second deficit.