Graphics titans AMD and Nvidia have released new products built to accelerate supercomputers.
AMD launched the FirePro S10000 12GB Edition, the “industry’s first supercomputing server graphics card with 12GB of memory”.
Set for a Spring launch in 2014, the FirePro supports PCI Express 3.0 and uses AMD’s Graphics Core Next architecture.
AMD stated that the card is designed for use cases such as compute and visualisation servers, such as those used in medicine, science and design sectors, double and single precision, which enable features such as computational fluid dynamics alongside video and image enhancements, and ultra high-end workstations, such as those used in computer-aided engineering.
Following AMD’s announcement, Nvidia has today announced the Tesla K40, claimed to be the “world’s fastest accelerator for supercomputing and big data analytics”.
Like AMD’s FirePro, the K40 has 12GB of memory, alongside 1.4 TFLOPS, 2880 cores and 288 GB/s speeds. Nvidia’s GPU Boost technology is said to give the card up to 25 per cent of extra performance on applications.
Nvidia also announced the latest version of its graphics architecture platform, CUDA 6.
CUDA 6 is said to “dramatically simplify” parallel programming for developers and software engineers, by utilising unified memory.