Author Archives: Phil

Texas Instruments has renewed their contract with Minimal Metrics to deliver optimized numerical libraries for their next generation of high-performance microprocessors. The Keystone 2, is a revolutionary new product that integrates four cores of an ARM A15 with 8 cores of the  C6678 DSP on the same die. Offload to the DSPs can be accomplished either through the use of OpenCL or a subset of the OpenMP 4 specification, OpenMPACC. For this contract, Minimal Metrics will be providing optimized, hybrid, ARM+DSP-accelerated versions of the following libraries. These libraries are critical elements in the middleware of high-performance numerical simulations. FFTW – Fast Fourier transform BLAS/ATLAS – Vector/vector, matrix/vector and matrix/matrix arithmetic LAPACK – Dense library algebra LIBFLAME – Dense linear algebra Applications are accelerated using new versions of the libraries transparently, without requiring any changes to the source or object code. In this way, HPC applications ported to the ARM gain vast increases in performance by…

Read more

Minimal Metrics has successful completed their work with Texas Instruments to deliver an optimized and complete BLAS (Basic Linear Algebra Subroutines) library. The BLAS are the building blocks of many high performance numerical algorithms – and are known as the keystone to high performance in scientific simulations. The Minimal Metrics team worked closely with the ATLAS group in order to provide a complete and optimized implementation much faster than could be coded by hand. The target platform for the work was the C6678 DSP, aka Keystone, an 8-core DSP with industry leading floating point performance per watt. This library will also be used for the Keystone 2, a revolutionary new product that integrates the ARM A15 CPU with 8 cores of the C6678. The DSP architecture is substantially different than coding for “traditional” RISC/CISC processors. The Keystone series provides a number of architectural features to enhance performance – including hardware-assisted software…

Read more

What’s more expensive than designing a new handset, tablet or mobile device? Verification! With the design tools available today, even a small team can design a device like a smart-phone from a commodity off-the-shelf components. These aren’t physical components mind you, this is silicon IP or intellectual property, and it’s how tremendously successful companies like ARM, Marvell and others sell their technology to Apple, Samsung and LG without ever physically building anything. With today’s tools, you can design the system the same way you might design a flow-chart in Power-point, drag and drop, plug and play… A little LCD screen here, a wireless radio there, memory-controller, flash card, keypad and of course a CPU. But how does one know this system is going to work? The entire system needs to be verified – and the only way to do that is to run it. This isn’t a logical (or functional…

Read more

When GeoSpace Technologies, a global leader in seismic data acquisition systems, wanted to scale up their performance and their capacity, they called Minimal Metrics. Aside from high-frequency trading, no-where is the performance a bigger differentiator than in seismic data processing. While the hardware and algorithms are highly guarded trade secrets, the process is relatively simple to explain. It consists of collecting seismic data from a large number of sensors, removing the impulse (transmitted) signal, correlating that data along temporal and spatial boundaries and then combining that data into a picture of what lies beneath. This picture actually consists of millions of data points and is often used as input into additional processing designed to identify features of interest – like natural gas deposits. Rapid turnaround time to this process is absolutely key due to the capital expenditures required to keep exploration and drilling equipment and personnel in the field. This pipeline is extremely data…

Read more

Minimal Metrics has formed a strategic partnership with Scalable Informatics, makers of some of the fastest high performance storage subsystems on the planet. While having extensive experience in application optimization and high performance computing system design, Minimal Metrics was lacking the deep experience necessary to solve some of our customers’ unique storage challenges. Enter Joe Landman, CEO and Chief Architect of Scalable Informatics. Joe has over 20 years in the design and optimization of high end storage systems – and that experiences shows brilliantly through Scalable Informatics’ product line. With significant customers in the financial sector, health care, simulation science and high performance computing, Joe and Scalable Informatics have the kind of expertise and reputation our customers have grown to expect. Through this partnership with Joe and his team, Minimal Metrics extends their stellar capabilities into the realm of high performance storage and the robust processing of “Big Data“.

The annual Supercomputing 2012 conference is almost upon us. This year, it will be held in Salt Lake City, Utah. Philip Mucci will be attending from Sunday November 11th to the morning of Wednesday, November 14th. In addition to wandering the floor, Philip will be meeting with current and prospective clients. Interested in what Minimal Metrics can do for your productivity? Drop us a line and meet Philip at the show – you’ll be glad you did.

LeanFirm consulting specializes in helping municipalities all over the United States re-engineer and optimize their workflows. What’s a workflow in this case? Consider the case of maintaining an optimal sized inventory of spare vehicle parts. It’s not just a matter of when a part is ordered, but when the decision is approved, when the money is available, when the part is actually delivered or how many times per year the part is needed. While the procedure may be long, the decisions made at each stage are rather clear – with each decision being logged in some form or another by the responsible party. But since none of the information is centralized (remember that most municipal governments are working with data systems that were designed in the early 80’s), getting actionable suggestions for improvements is limited to the observations at each point in the chain. How does one develop real-time and…

Read more

Minimal Metrics has teamed up with the authors of ATLAS to bring their ultra high performance implementation of the BLAS or Basic Linear Algebra Subroutines to Texas Instruments‘ next-generation, multi-core, digital signal processing (DSP) platform. BLAS is used throughout the industry to achieve performance on a number of low-level matrix and vector operations. It is best known as the building block of LAPACK; a programming library invented by Dr. Jack Dongarra in the 70’s to solve problems consisting of simultaneous systems of linear equations. Solving such systems are at the core of many uses of high performance computers today, whether it be in the simulation sciences like weather forecasting, aircraft design or car crash simulation or web technologies such as data mining, web analytics and natural language processing. Ti’s new generation of processors, beginning with the 8-core C6678, a.k.a “Shannon” raise the bar on the performance per-watt for both integer…

Read more

18/18