If you use virtualisation software you'll be only too aware that, while modern CPU include extensions allowing virtual operating systems to address the hardware directly, modern graphics systems often go to waste under such systems. Thankfully, Nvidia is aware of this too – and is working on a solution.
The company's latest range of Quadro FX professional-grade graphics hardware will, according to
BetaNews, include full support for 3D acceleration functionality even on a virtualised operation system – something not previously possible.
The technology, which the company has called SLI Multi-OS, is a combination of clever driver tricks and firmware hooks into Intel's Virtualization [sic] Technology built in to most of the company's current range of CPUs to allow virtual machines to address the hardware directly – benefiting from the speed boosts offered over the purely software graphics implementations they have been limited to before.
While support for SLI Multi-OS is currently limited to a single virtualisation provider – Parallels – the company has pledged to work to introduce support into other products, including Microsoft's VirtualPC and the VMware range of software.
While making use of expensive professional graphics hardware via multiple virtual machines is smart, at least one vendor has opted to use the technology from a different angle: an up-coming range of graphics workstations from Hewlett Packard is set to use SLI Multi-OS to allow multiple physical workstations to offload tasks onto a single Quadro FX card – effectively sharing its power between them. With the top-end Quadro FX 5800 card – which features 4GB of addressable RAM on-board – retailing at over £2,500, there are significant savings to be made.
Are you looking forward to the day when the SLI Multi-OS technology filters down to the gaming cards and your VirtualBox XP install will play games at a decent pace under Linux, or is the system nothing more than a high-end curiosity? Share your thoughts over in
the forums.
Want to comment? Please log in.