Weather forecasting. Tracking billions of tweets. Mitigating big-time financial risks. These burly applications gobble up computing power like an NFL linebacker plowing through a box of Ding Dongs.
So when we tore the lid off the supercharged Tesla K40 GPU accelerator – the highest-performance GPU ever built – our customers quickly put it to work tackling some of their toughest problems. And their initial results show this GPU is even better than advertised.
Improving Weather Forecasting
Weather prediction is one of the toughest computing jobs around.
The U.S. National Weather Service’s supercomputers run an application called WRF to forecast weather conditions throughout the country.
Dr. Bormin Huang, from the University of Wisconsin-Madison, uses K40 GPUs to accelerate WRF to deliver more accurate, higher resolution forecasts.
His latest findings show that, at baseline, the Tesla K40 runs WRF nearly 30 percent faster than the Tesla K20.
But when he turns on the new GPU Boost feature, which converts power headroom into a user-controlled performance boost, it’s 46 percent faster.
Accelerating WRF allows him to run more detailed, higher resolution weather models for extended periods of time, leading to more precise and accurate forecasting.
Tracking Twitter Trends – Analyzing a Billion Tweets in Real Time
A staggering 2.5 exabytes of data is created every day on the internet (source: http://en.wikipedia.org/wiki/Big_data). Top social media sites contribute to this massive volume of information, and can be a treasure trove of info on the latest trends.
But in the real-time world of Twitter, it’s a major challenge to extract useful data, and then analyze it before today’s trends become yesterday’s news.
Enter Map-D, a start-up out of the Massachusetts Institute of Technology.
Without a budget to build a large CPU server farm, Map-D founders turned to GPUs to create a cutting-edge visualization system that can track more than a billion tweets worldwide and deliver real-time analysis of virtually any Twitter query one can think of – an impressive database application hosted on a single server equipped with eight Tesla K40 GPUs.
With an aggregate GPU memory of 96GB, it’s large enough to store 1 billion tweets in memory for ultra-fast analysis. And with GPUs, it’s over 70 times faster than with CPUs alone, or fast enough to let you know what’s happening in the Twitterverse at any given moment in time.
Want to know what people thought of the Oscar nominees? Curious who are the fan favorites to win at the 2014 Sochi Olympics?
Yep, they can track all that. In real time, even as it changes from moment to moment.
Reduced Financial Risk
Monte Carlo algorithms – used to analyze the valuation of complex financial instruments – are among the most important in the financial services industry.
Software vendor Xcelerit uses GPU accelerators to deliver unprecedented performance for applications built with Monte Carlo algorithms.
Once again, the Tesla K40 dramatically accelerates the application – compared to the Tesla K20X, it’s consistently 20-30 percent faster.
Test Drive a Tesla K40 Yourself
We’ve made it easy – and free – for you to test drive a K40. Simply register for the test drive program at www.nvidia.com/GPUTestDrive.
The site is pre-loaded with a number of different applications – from AMBER to GROMACS – allowing you to see how the Tesla K40 can supercharge your work.
If you take the test drive, or have results from your own Tesla K40, share them with us in the comment section below.