Topics

• MCU Working Group
• IoTMark™
• Machine Learning Benchmarks Progress Report
• Press Coverage Roundup

MCU Working Group Reactivated

Ten years ago EEMBC introduced CoreMark® as a replacement for the classic Dhrystone and MIPS performance metrics. Its real-world behavior stood in stark contrast to the synthetic nature of its predecessors, and it rapidly became the MCU standard. Five years later, CoreMark-PRO was introduced to address the next generation of MCUs and CPUs. It added more tests (like XML parsing and matrix reduction), used larger datasets, and introduced both process- and thread-parallelism.

Today's MCUs and CPUs are even more advanced than their ancestors, with higher frequencies, richer instruction sets, more advanced architectures, and a range of multiprocessing capabilities from dedicated IP and on-die DPSs to new asymmetric core strategies. As with any good benchmark, CoreMark needs to continue evolving to address this changing competitive landscape. Since the introduction of CoreMark-PRO, several new companies have popped up and have begun to sample new and exciting silicon.

CoreMark-X is still in the definition phase, as EEMBC members work to qualitatively aggregate all of these new features into a compelling benchmark. However, it is still far from complete, and could benefit from broader participation. EEMBC would like to invite these new companies (and old ones with new products) to join this effort and be a part of CoreMark’s evolution.

For more infomation about participating in the MCU working group, please email EEMBC president Peter Torelli at peter.torelli@eembc.org

IoTMark™ Internet of Things Benchmark Now Available for Bluetooth Low Energy Edge Nodes

EEMBC’s IoTMark™-BLE benchmark is now available for licensing. Part of the IoTMark benchmarking suite for measuring the combined energy consumption of an edge node’s sensor interface, processor, and radio interface, IoTMark-BLE focuses on Bluetooth Low Energy (BLE) devices. Learn more . . .

Machine Learning Benchmarks Progress Report Now Available

A new progress report is available that provides the EEMBC community and its followers with an overview of how the Machine Learning Benchmark Suite is being developed, including the motivation for the suite and the choices that have been made so far about how and what it will measure.

Although the development of the Machine Learning Benchmark Suite is well underway, new working group members are still welcome and will have the opportunity to influence decisions on a number of open issues. These include score reporting, model format, model optimization, test dataset, accuracy validation set, and target accuracy. The Machine Learning Working Group will be meeting throughout the remainder of 2018 to resolve these issues, and EEMBC welcomes your participation.

To request your copy of the Machine Learning Benchmarks Progress Report and to find out more about the Machine Learning Working Group, please email EEMBC president Peter Torelli at peter.torelli@eembc.org.

Hot off the Press

Over the last five months, EEMBC has made announcements regarding the formation of its Machine Learning Working Group and the availability of its ADASMark™, SecureMark™-TLS, and IoTMark™-BLE benchmarking suites. The industry press has taken notice, providing EEMBC with outstanding editorial coverage in a number of high-profile publications, including EE Times and Autonomous Vehicle Technology. The highlight reel includes:

• May 17, EE Times — Rick Merritt explores the Machine Learning Working Group and how the effort spun out of ADASMark™ development
• July 25, Autonomous Vehicle Technology — Kevin Jost provides an in-depth look at ADASMark™
• August 16, EE Times — Rick Merritt discusses two new EEMBC benchmarking suites that address key IoT functions: SecureMark-TLS™ and IoTMark™-BLE
• August 31, EE Times — Junko Yoshida chats with EEMBC president and CTO Peter Torelli about ADASMark™

Follow us on Twitter.

Visit us on Linkedin.