About Advanced Cluster Systems, Inc.

and its Revolutionary Software Technology

Advanced Cluster Systems (ACS) builds unique parallel programming tools for Multicore, Clusters and Supercomputers. ACS's SET™ (Supercomputing Engine Technology)™ is a software development framework and implementation library for writing high-performance code required to leverage parallelism in hardware. This innovative software framework not only makes parallel programming virtually as simple as sequential programming, but also offers unique and powerful supercomputing technology enhancements to software applications. Leveraging the latest advancements in increasing Internet speed, the SET framework harnesses remote computing clusters and Cloud resources so that applications on desktop computers, laptop computers, tablet computers and mobile devices seamlessly command true high-performance computing solutions*.

Who needs SET, and why do they need it?

HPC (High Performance Computing) application developers (such as ISVs who develop HPC applications to address a market need or a SMB that develops HPC applications unique to its line of business) need SET. Its simplicity, relative to alternatives, minimizes their in-house expertise, the complexity of their code, their learning curve, and their debugging time. Thus the quick time-to-market and workload reduction gives them a strategic advantage and saves them money. In addition, SET opens the door for development of new generation HPC applications that allow the use of clusters and supercomputers as scientific instruments.

Attention, Researchers and ISVs! Real-Time Interactive Supercomputing is Here!

With SET, it is possible to develop with relative ease parallel applications that run on thousands of cores. It also provides an easy way to scale legacy codes. However, SET™ is much more than just an efficient, easy to use parallel programming tool. The SET framework provides SET-based HPC applications with real-time interactive supercomputing capabilities, being able to be interactively steered and controlled much like word processors or spreadsheet applications. In addition, the SET framework can create ad hoc supercomputer-like parallel computing machines out of collections of individual Mac and Linux computers usually found in SMB environments. These parallel computing machines can then run fully interactive SET-based HPC applications. The SET framework is also  capable of utilizing local computing clusters or Cloud resources as well, when available. During tests, the SET™ framework worked fine on 8,000-cores of an IBM iDataPlex cluster.

With SET, researchers and ISVs can now create the next generation interactive supercomputing HPC applications. Moreover, existing HPC applications can be modified to take advantage of the SET framework, adding real-time interactivity capability to them. Please contact us for more information.

How is SET different from other solutions in the market?

SET embodies an approach very different from others, abstracting and implementing commonly-used parallel coding patterns and algorithms. Other solutions may help debug code, not develop it.

Why is SET better than other solutions in the market?

SET helps HPC application developers write good parallel code, avoiding bugs altogether and provides time-to-market and implementation advantages unavailable in other tools. SET has built-in scalability features, and it enables full utilization of all available cores. In addition, the SET framework provides powerful unprecedented supercomputing technology enhancements to software applications, such as real-time interactivity with steering and control of HPC codes and real-time visualization capability, which no other HPC development tool or parallel framework offers.

How does SET allow for faster parallelization and testing?

SET abstracts and implements commonly-used parallel coding patterns and algorithms. Others require the developer to rewrite these patterns over and over themselves, an arduous and time-consuming task for the uninitiated.

Case Studies

During the development phase of SET, the SET parallelization approach (which adopts message passing**) was applied to Wolfram Research Mathematica®1, providing it with supercomputing-style parallelization. The development effort took only 1 man/month, and was accomplished without access to Mathematica's source code. There is no other development system available today that can provide Mathematica with supercomputing-style parallelization in 1 man/month, even with an access to source code. Without SET, such an effort could take a year with many programmers involved. See SEM™ (Supercomputing Engine for Mathematica) for more details.

Also during the development phase of SET, the SET parallelization approach was applied to Apple QuickTime®2 transcoding functionality, providing it with supercomputing-style parallelization. As with Mathematica, the development effort took only 1 man/month, and was accomplished without access to QuickTime source code. Again, there is no other development system available today that is capable of providing QuickTime transcoding functionality with supercomputing-style parallelization in 1 man/month.

SET was utilized in providing supercomputing-style parallelization to Scilab, an open source alternative to Matlab®3. It took us less than 2.5 man/months to accomplish this task. Scilab is a large open source code, far from being an ideal candidate for parallelization because of its non-standard, complicated structure. Yet we have proved that even such tough codes can be parallelized with SET in a record time. See Equalis Thunder and Lightning for more details.

The parallelization approach of SET is explained here. Please do not hesitate to contact us for any question.

 

*SET™ is a patented technology.

**Message passing is an inherent element of all computer clusters. All computer clusters, ranging from homemade Beowulfs to some of the fastest supercomputers in the world, rely on message passing to coordinate the activities of the many nodes they encompass. [From Wikipedia, the free encyclopedia].

 

1. Mathematica® is a registered trademark of Wolfram Research Inc.

2. QuickTime® is a registered trademark of Apple Inc.

3. Matlab® is a registered trademark of Mathworks Inc.

 

 

Copyright © 2014 Advanced Cluster Systems, Inc. All rights reserved.

Supercomputing Engine Technology™and SET™ are trademarks of Advanced Cluster Systems, Inc.