Nvidia has unveiled its latest graphics offering in the form of the Tesla K80 dual-GPU. The new card, which is designed for professional applications, doubles the performance of its K40 predecessor, and packs some impressive specs including a full 24 GB graphics memory.
The first thing to note about the K80 is that it’s most certainly not designed to sit in a gaming rig. While its high-end specs might make it look like the perfect fit for a top-tier desktop PC, it’s actually designed for more serious uses. The card is geared towards difficult computational tasks that make use of high performance computing (HPC) applications, common in fields such as quantum chemistry and astrophysics.
According to Nvidia, the K80 offers 10 times higher performance than the fastest CPUs on the market when it comes to engineering and science applications. In terms of specs, it offers dual GPUs, each packing 12 GB GDDR5 RAM, for a total of 24 GB graphics memory. There are 4,992 CUDA parallel processing cores on board, allowing for 480 GB/s memory bandwidth. It’s also compatible with the company’s GPU Boost technology, which is designed to scale the GPU clock, increasing performance with specific professional applications.
The card is designed to work with more than 280 enterprise, commercial, engineering and scientific HPC apps, as well as data center applications, and is available to purchase right now from selected retailers. Nvidia doesn’t publish pricing information for its professional grade graphics accelerators, but given that its significantly less powerful predecessor launched for well over US$5,000, we’d expect the K80 to carry a pretty hefty price tag.