“If a man can write a better book, preach a better sermon, or make a better mouse trap than his neighbor, though he build his house in the woods, the world will make a beaten path to his door.” The sentence, usually shortened to emphasize the better mouse trap, is by Ralph Waldo Emerson (1803-1882), an American poet and philosopher. What Emerson said so long ago applies even today as inventive people work to build a better computer mouse such as wireless or optical.

In the Research and Development Labs at Intel, we are applying Emerson’s philosophy to building smarter computers which are more natural and easier to use—a lot less strict and a lot more adaptive to humans. We’re building smarter computer technologies that can rapidly and transparently analyze large complex datasets and explore different outcomes, and ultimately help people do what they want to do more easily. These new capabilities or “uses” are driving a whole new set of computer and architecture requirements.

The seven papers in this issue of Intel Technology Journal (Volume 9, Issue 2) focus on Compute-Intensive, Highly Parallel Applications and Uses. They review the exploratory work into complex and large “workloads” that can ultimately run efficiently on future computers. Generally, these are characterized as compute-intensive, highly parallel workloads requiring new levels of intelligence, performance, and sophistication in both hardware and software that does not exist today. And we look at how the performance scalability and uses on parallel architectures of such applications can help to best architect the next generation of computers.

The first paper is on ray tracing, a technique used in photo-realistic imagery such as in the creation of computer games and special digital effects in movies. Ray tracing can be an important workload to establish requirements for new architecture that will one day run efficiently on mainstream computers.

The second and third papers are on computer vision. The vast accumulation of digital data requires new classes of applications. We are investigating computing platforms that can deliver enough performance for these future workloads to enable their use in mass-market applications. Computer Vision (CV) is one such workload. In the second paper we introduce and characterize some of the most common CV algorithms and applications. We chose a complete video surveillance application as a representative case study for a complex CV workload. The third paper looks at Intel’s Open Source Computer Vision Library and describes using OpenCV for “learning-based vision,” where objects such as faces, or patterns such as roads, are learned and recognized.

The fourth and fifth papers look at data mining, or the ability to extract knowledge, acquire models, and draw meaningful conclusions from a dataset. The fourth paper examines data mining applied to bioinformatics. Bioinformatics is the recording, annotation, storage, analysis, and search/retrieval of gene sequences, protein sequences, and structural information. In this paper, we report on the performance scalability analysis of six bioinformatics applications on a 16-way Intel^® Xeon™ multiprocessor system. The fifth paper looks at large-scale data-mining problems based on tree-based models. Tree-based models, in the context of large-scale data-mining problems, provide many challenges for a computing platform. The balance between complexity and accuracy is studied for different parameter sets and its performance impact is discussed.

The sixth paper looks at optimization algorithms using the Interior Point Method (IPM). IPM has become a dominant choice for solving large optimization problems for many scientific, engineering, and commercial applications. In this paper we describe a parallel IPM for solving optimization problems.

The seventh paper examines future IT enterprise platform requirements based on usages and deployment models. We present the needs of various vertical industries (e.g. retail, manufacturing, financial) and discuss the business usage and the technology deployment trends across these industries. We describe how the emerging models are different in their characteristics from those prevalent today, and, using several real-world examples, explain the platform implications.

These papers look at new, intelligent, large, and sophisticated workloads that can analyze large complex datasets and explore different outcomes, and ultimately help people do what they want to do more easily on computers. We also look at vertical industries and how their needs will steer platform definitions. These capabilities or “uses” are driving a whole new set of computer and architecture requirements. Let’s experiment together on future usage models impacting future computer platforms to build tomorrow’s smarter computers.