Nvidia delivers Tesla K40 GPU to speed up high-performance computing

New GPU accelerator targets big data analytics and large-scale scientific workloads

Nvidia has added a new model to its Tesla line of GPU cards, claiming it is the world's highest-performance accelerator ever built for boosting scientific, engineering, high-performance computing (HPC) and enterprise applications.

Available today, the Tesla K40 has double the memory and up to 40 percent higher performance than its predecessor, the Tesla K20X. This means the adapter offers 10 times the performance of today's CPUs for handling the kind of calculations involved in big data analytics and large-scale scientific workloads, Nvidia said.

Based on Nvidia's Kepler GPU architecture, the Tesla K40 boasts 2,880 CUDA processing cores combined with 12GB of fast GDDR5 memory. This configuration enables the Tesla K40 to deliver 4.29 teraflops floating point performance at single precision and 1.43 teraflops at double precision, Nvidia claimed.

Sumit Gupta, general manager of Tesla Accelerated Computing products at Nvidia, said GPU accelerators have now become mainstream in the HPC and supercomputing industries.

"With the breakthrough performance and higher memory capacity of the Tesla K40 GPU, enterprise acustomers can quickly crunch through massive volumes of data generated by their big data analytics applications," he said.

The Tesla K40 also adds another capability, Nvidia's GPU Boost technology. This enables the GPU to ramp up its clock speed for extra performance until the adapter card hits its predetermined power limit, analogous to the Turbo Boost seen in Intel processors.

As well as being a standalone product available through Nvidia reseller partners, the Tesla K40 will ship inside servers from Bull, Cray, Dell, Eurotech, HP, IBM, SGI, Supermicro, Tyan and others.

Meanwhile, Nvidia said it is partnering with IBM to deliver GPU-accelerated versions of IBM enterprise applications on its Power Systems portfolio.

The collaboration aims to enable IBM customers to more rapidly process, secure and analyse massive volumes of streaming data, and builds on the firm's announcement in August of the OpenPower Consortium to drive broader data centre adoption of its Power architecture.

"This partnership will bring supercomputer performance to the corporate data centre, expanding the use of GPU accelerators well beyond the traditional supercomputing and technical computing markets," said Ian Buck, vice president of Accelerated Computing at Nvidia.