NVIDIA Tesla GPUs Built on Kepler Architecture

By DE Editors

NVIDIA unveiled a new family of Tesla GPUs based on the NVIDIA Kepler GPU computing architecture, which makes GPU-accelerated computing easier and more accessible for a broader range of high performance computing (HPC) scientific and technical applications.
 
The new NVIDIA Tesla K10 and K20 GPUs are computing accelerators built to handle the most complex HPC problems in the world. Designed with a focus on high performance and extreme power efficiency, Kepler is three times as efficient as its predecessor, the NVIDIA Fermi architecture.
 
The SMX streaming multiprocessor was redesigned from the ground up for high performance and energy efficiency. It delivers up to three times more performance per watt than the Fermi streaming multiprocessor,  making it possible to build a supercomputer that delivers one petaflop of computing performance in just 10 server racks. SMXs energy efficiency was achieved by increasing its number of CUDA architecture cores by four times, while reducing the clock speed of each core,  power-gating parts of the GPU when idle and maximizing the GPU area devoted to parallel-processing cores instead of control logic.

The dynamic parallelism capability enables GPU threads to dynamically spawn new threads, allowing the GPU to adapt dynamically to the data. It greatly simplifies parallel programming, enabling GPU acceleration of a broader set of popular algorithms, such as adaptive mesh refinement, fast multipole methods and multigrid methods.

Hyper-Q enables multiple CPU cores to simultaneously use the CUDA architecture cores on a single Kepler GPU. This increases GPU utilization, slashing CPU idle times and advancing programmability. Hyper-Q is ideal for cluster applications that use MPI.
 
The NVIDIA Tesla K10 GPU delivers what the company says is the worlds highest throughput for signal, image and seismic processing applications. Optimized for customers in oil and gas exploration and the defense industry, a single Tesla K10 accelerator board features two GK104 Kepler GPUs that deliver an aggregate performance of 4.58 teraflops of peak single-precision floating point and 320 GB per second memory bandwidth.
 
The Tesla K20 GPU is the new flagship of the Tesla GPU product family,  designed for the most computationally intensive HPC environments. It will be available in the fourth quarter of 2012.
 
The Tesla K20 is based on the GK110 Kepler GPU, which is expected to be incorporated into the new Titan supercomputer at the Oak Ridge National Laboratory, and the Blue Waters system at the National Center for Supercomputing Applications at the University of Illinois.
 
In addition to the Kepler architecture, NVIDIA also released a preview of the CUDA 5 parallel programming platform. Available to more than 20,000 members of NVIDIAs GPU Computing Registered Developer program,  the platform will enable developers to begin exploring ways to take advantage of the new Kepler GPUs, including dynamic parallelism.

For more information, visit NVIDIA.
 
Sources: Press materials received from the company and additional information gleaned from the company’s website.