Nvidia may be known for its graphics chips, but its efforts in supercomputing have taken the spotlight as of late.
Following up to last week’s news that it would team with IBM and the US Department of Energy, Nvidia today has released a new super-charged Tesla GPU accelerator that is its fastest to date.
Designed for a bevy of high performance computing applications, the Tesla K80 dual-GPU crams in twice as many flops and double the memory bandwidth of its predecessor, the Tesla K40 GPU accelerator.
Nvidia says the Tesla K80 has 10 times higher performance than competing CPUs, and that it is especially more capable of handling complex analytics and computationally intensive scientific computing.
The chip’s two graphics processors share the balance of the processing with a combined 4,992 processing cores, 24GB of ultra-fast GDDR5 memory and memory bandwidth of 480GBps, which Nvidia touts as offering a boost in throughput for data scientists crunching petabytes of information.
The company seems to also have implementations ranging from astrophysics to deep machine learning in mind with the new technology, with its more than 280 enterprise HPC and datacenter applications available out of the gate. Nvidia says HP, ASUS, Dell, Quanta, Inspur and Cray will use the K80 GPUs in servers.