With the PostMark mail server testing, KVM took close to a 20% performance hit while VirtualBox was down by about 25% and Xen's disk performance was almost halved.
The fastest virtualization method for the SQLite test was Oracle's VirtualBox and it was even faster than the bare metal results. However, this is due to a problem with VirtualBox as it is likely not enforcing the sync/fsync requests by SQLite. If you will recall from a year or two ago when there were similar QEMU/KVM SQLite benchmarks, it too would outperform the host. It was tracked down to being a sync/fsync issue that could potentially put the virtual machine's data at risk since it was not passing these requests down. That issue was since cleared up, but it seems VirtualBox may be having this same issue/bug.
Well, to not much surprise, Linux KVM was the fastest means of virtualization when looking at the compute performance (single and multi-threaded) and disk workloads. The Kernel-based Virtual Machine was performing very well with hardware-assisted virtualization on the Core i7 "Sandy Bridge" with Intel VT-x. KVM offered the fastest performance in all of the tests except for SQLite, where VirtualBox was the fastest but that is due to a bug. VirtualBox regularly was the second fastest virtualization method tested while Xen was generally in a distant last for its HVM virtualization.
The advantage of VirtualBox over KVM, however, is its current support for 2D/3D acceleration within guest virtual machines that is then redirected to the host's graphics card. This will hopefully change in the future for QEMU/KVM thanks to SPICE and other work, but for now it lacks this support. Xen is also capable of PCI pass-through and taking control of graphics cards. VMware also offers 2D/3D acceleration in a great way by leveraging Gallium3D. The next Linux virtualization benchmarks on Phoronix are seeing where the VirtualBox 3D performance is at for Linux guests. There are also AMD Bulldozer virtualization benchmarks around the corner.