In a low-key move during Thanksgiving, NVIDIA has added the first graphics chipset to its GeForce 300 series. The GeForce 310 is a direct replacement for the earlier 210 and, like its predecessor, is meant as a basic GPU to upgrade the performance of systems that are limited to slower integrated graphics. It has just a 64-bit memory bus and just 16 processing cores but can still fully accelerate 1080p video in hardware and provide general-purpose computing tasks, whether it's NVIDIA's own CUDA and PhysX or more universal standards like DirectCompute and OpenCL.
In spite of its GeForce 300 series name, the card is believed to be based on current architecture rather than the Fermi design, which will add DirectX 11-level graphics and much stronger general computing performance.
The part normally runs with a 589MHz main clock and carries 512MB of DDR2 memory that runs at 500MHz. Despite its budget focus, a reference board carries DisplayPort, DVI and VGA connectors and will work with HDMI through an adapter for either the DisplayPort or DVI links. NVIDIA hasn't said when it expects cards to be out on the market or detailed full operating system support, though the GeForce 310 like the 210 before it is likely destined for pre-built computers more than as a stand-alone upgrade.