Nvidia announced its upcoming Tesla K40 GPU accelerator, Monday, at the International Conference for High Performance Computing, Networking, Storage and Analysis convention in Denver, Co.
Delivering 1.43 teraflops of processing power from its 2,880 CUDA cores and 12GB of GDDR5 memory, the Tesla K40 is set for a 2014 release at a price point of $5499. This surpasses all other accelerators on two common measures of computational performance.
"GPU accelerators have gone mainstream in the HPC and supercomputing industries, enabling engineers and researchers to consistently drive innovation and scientific discovery," said Nvidia General Manager of Tesla Accelerated Computing Products, Sumit Gupta. "With the breakthrough performance and higher memory capacity of the Tesla K40 GPU, enterprise customers can quickly crunch through massive volumes of data generated by their big data analytics applications."
Requirements for power and cooling will not change, therefore being in line with the K20X's 235W. Feature-wise, the K40 will offer GPU boost through selectable clock-speeds. The K40 cards can be set at 810MHz and 875MHz to its 745MHz.
Texas Advanced Computing Center (TACC) at the University of Texas at Austin will be one of the first Nvidia partners to deploy the graphics accelerator. The company and university plans on building a remote visualization and data analysis system scheduled to go online January called "Maverick."
"The Tesla K40 GPU accelerators will help researchers crutch through massive volumes of big data and gain new insights through large-scale, sophisticated visualizations," said Visualization Director at TACC, Kelly Gaither. "With NVIDIA GPUs, Maverick will provide researchers powerful interactive capabilities to advance their most complex scientific challenges."
Availability for the device will also come from a variety of server manufacturers including Appro, ASUS, Bull, Cray, Dell, and Eurotech among others.
Nvidia also announced a new long-term collaboration with IBM. The purpose of the joint venture is for a "supercharge the corporate data center" through development of GPU-accelerated enterprise software applications using IBM's Power processors.