UltraSoC works with PDF Solutions
to prevent in-life product failures
COMPANIES AIM TO PREVENT CHIP FAILURES IN THE FIELD. NEIL TYLER REPORTS
UltraSoC is collaborating with PDF Solutions, combining comprehensive data analytics with
advanced machine learning (ML) techniques, with the aim of predicting and preventing chip
failures in the eld.
The solution will combine in-life information from UltraSoC’s hardware-based behavioural
monitors with PDF Solutions’ end-to-end machine learning and analytics platform to identify
chips that are likely to fail in the eld.
It is the rst solution that’s able to provide such a comprehensive view of historical data from
semiconductor manufacturing, test, assembly, supply chain traceability and in- eld data within a
common semantic data model.
Over 100 leading semiconductor companies worldwide use PDF Solutions’ Exensio Software
Platform to monitor, diagnose, and identify manufacturing issues to improve key performance
metrics from the factory oor to test operations and assembly.
UltraSoC’s embedded analytics and monitoring technology will be able to deliver the nal
piece of the data analytics puzzle by providing data on the in-life digital behaviour of the chip or
system.
UltraSoC monitoring observes functional behaviour trends over a period of time to construct a
comprehensive picture of potential problems with the device while in use.
Combining in- eld monitoring data, manufacturing data, and the appropriate AI powered
by machine learning, holds real potential and will offer chip makers and OEMs a complete
predictive analytics platform for their SoCs. The ML-driven analytics framework can be used to
automatically generate alerts, actions and system reports.
Commenting Dennis Ciplickas, VP of Advanced Solutions at PDF Solutions, said, “Connecting
to UltraSoC’s in-life monitors and data will enable us to extend our analytics and ML offerings to
support a total preventive maintenance solution for semiconductor devices.”
AI multicore processor for embedded sensor applications
ETA Compute has begun shipping silicon for
its ECM3532, its AI multicore processor for
embedded sensor applications.
The multicore device features the
company’s patented Continuous Voltage
Frequency Scaling (CVFS) and delivers power
consumption of microwatts for many sensing
applications.
Eta Compute’s ECM3532 is a Neural
Sensor Processor (NSP) for always-on image
and sensor applications.
“Our Neural Sensor Platform is a complete
software and hardware platform that delivers
more processing at the lowest power pro les
in the industry. This essentially eliminates
battery capacity as a barrier to thousands of
IoT consumer and industrial applications,”
said Ted Tewksbury, CEO of Eta Compute.
The ECM3532 family brings AI to edge
devices and transforms sensor data into
actionable information for voice, activity,
gesture, sound, image, temperature,
pressure, and bio-metrics applications,
among others. The platform addresses
issues for the most important problems in
edge computing: longer battery life, shorter
response time, increased security and higher
accuracy.
The company’s standalone AI platform
includes a multicore processor that includes:
ash memory, SRAM, I/O, peripherals and
a machine learning software development
platform. The patented CVFS, according
to the company, substantially increases
performance and ef ciency for edge devices.
The self-timed CVFS architecture
automatically and continuously adjusts
internal clock rate and supply voltage
to maximize energy ef ciency for the
given workload. The ECM3532 multicore
NSP combines an MCU and a DSP, both
with CVFS, to optimise execution for
the best ef ciency making it a suitable
solution for IoT sensor nodes.
Socionext
introduces
TSN IP
Socionext has developed a Time-Sensitive
Network (TSN) IP for FPGA and ASIC
implementation.
The IP, which has been developed
to provide true deterministic Ethernet
for industrial applications, is compliant
with the next-generation Ethernet TSN
(communication standard IEEE 802.1
Subset) and its evaluation environment.
Features include support for a 2-port
daisy chain topology suitable for connecting
industrial equipment, 1 Gbps high-speed
operation, low latency less than 400 ns,
and low jitter less than 0.1 μs.
The IP enables a range of industrial
applications such as motion controllers,
which require faster response control as
well as TSN support, and remote I/O,
often used in network communications
to enable control of secured bandwidth
and low latency. TSN also seamlessly
connects and interoperates networks
for IT (Information Technology) and OT
(Operational Technology) making it suitable
in supporting the development of smart
factories.
Socionext will provide an FPGA
evaluation board and start-up manual for
IP implementation, as well as Linux opensource
driver software. This will allow users
to quickly evaluate and develop industrial
applications equipped with TSN.
Socionext, with its extensive experience
in developing devices for industrial
applications, will be able to ensure a
simple transition from FPGA to ASIC and
help customers develop their own custom
ASICs for optimum functionalities and
performance with the IP.
www.newelectronics.co.uk 25 February 2020 9
/www.newelectronics.co.uk