Pervasive

May 2010 - Posts

• Is Intel On To Something?

  If you ask anyone on the DataRush team, they could tell you more than you wanted to know about how we are an innovator in massively parallel processing.  They could tell you about use cases and fast runtimes and how DataRush is the bees knees for data-intensive applications. But what else would you expect to hear?  Instead of telling you how great DataRush really is, we went to Intel, a technology leader, and asked to demonstrate the speed of our parallel processing engine on their hardware.  Next thing we know, we’re testing on a brand new Intel Xeon processor 7500 series and the results made our team extremely enthusiastic.  In fact, our results were superior to those of clusters:  nearly 2 terabytes an hour!  Intel must have like it because they just released a success story about our accomplishment.     Not that we’re bragging, but here’s some of what Intel had to say: Better performance on simpler infrastructure.  The MalStone benchmarks were developed to measure performance on cloud infrastructures, and these single-server results are competitive with or superior to those of compute clusters. Scalability to many cores. As shown in the chart, “MalStone B Benchmark Scalability with Pervasive DataRush,” the solution scales well as the number of execution cores increase.  Simpler implementation.  The Pervasive DataRush solution architecture is designed explicitly to be applicable to virtually any type of application, abstracting multicore scalability away from the rest of the program logic.    Together, Intel and Pervasive DataRush, are tackling massive data challenges on a small footprint.  Processing two terabytes an hour opens up a door to endless possibilities that is being recognized by technology leaders, developers, and end users everywhere.  Don’t take our word for it – but maybe Intel is on to something…   Posted May 17 2010, 04:01 PM by livey with 3 comment(s) Filed under: Pervasive DataRush, intel, Scalability, data-intensive, MalStone B, 2 terabytes an hour

• What if genetic tests could be done in minutes instead of days?

  For many medical tests, time is an inherent part of the equation.  In most cases, medical results are delayed by compute time.  A good example of this is genetic testing.  Research finds that a patient’s genetics are an important factor in the efficacy of treatment.  Two examples announced recently illustrate this. Plavix, the second biggest selling drug after cholesterol-lowering Lipitor, is intended to prevent blood clots that can cause heart attacks and strokes in patients with advanced cardiovascular disease. It's also commonly prescribed to patients treated with devices called stents to prop open diseased coronary arteries. Some 2.5 million to 3 million Plavix prescriptions are written in the U.S. every month. But a genetic variation in a significant minority of patients can prevent the drug from working, or can limit its effectiveness, increasing a patient's risk for a potentially life-threatening heart attack.  Unfortunately a patient typically waits at least two days, and often longer, for the genetic results to come back to determine if Plavix is right for them. An even more widely used example is determining which of two popular dieting regimes will be more effective.  Many dieters choose a low-fat approach; others choose a low-carb diet.  What if individuals could easily test for their genetic makeup to determine the best diet for themselves to obtain fast weight loss?  Presently, testing for this is time-consuming and expensive.  But what if patients could see results in minutes instead of waiting weeks on all types of medical testing?  Leveraging all the available power of today's multicore hardware at the same time can result in dramatically faster speed.  Pervasive's DataRush platform attacks analytical applications with fully parallel processing and result in significantly shorter wait times.  Faster results equals better health, deeper understanding, and improved outcomes.  Why wait?  Posted May 04 2010, 09:57 AM by shochschild with 1 comment(s) Filed under: multicore hardware, genetic testing, genetic test, long compute time