Adaptable compute acceleration platforms and Sensor interface products Over the years the industry has continued to explore the different options of addressing the ever-increasing need for high performance computing in sensor processing and artificial intelligence applications. Many of these applications require more than a standard CPU to perform the complex algorithms and growing amount of data to be processed. This has given rise to platforms that form a combination of traditional CPUs collaborating with vector processing units like GP GPUs and DSP processors. However, these platforms run into scaling challenges due to inefficiency of the memory bandwidth usage. Also, the GPU/DSP platforms require a relatively large amount of power for their processing. One of the solutions for both problems have been traditional FPGA platforms, which offer highly configurable programmable logic with massive parallel processing capabilities that require a fraction of the power dissipated by GPU/DSP platforms.

In recent years we have seen a large growth of Multi-Processor System on Chip (MPSOC) platforms that further reduce the size, weight and power of sensor processing and artificial intelligence platforms. The MPSOC devices are a combination of multiple CPUs, Realtime CPUs, Graphic processing units, Programmable logic and peripheral interfaces on one chip.

Both FPGA and MPSOC devices are complex products that are challenging to incorporate into a sensor processing platform and require a lot of intimate knowledge and hardware design skills. Using commercial of the shelve hardware that is used by multiple companies for multiple different applications reduces the time to market, design risk and the need for inhouse FPGA and MPSOC experts.

The latest trend in de market is the introduction of the Adaptive compute acceleration platform chips. These devices have an architecture that significantly extends the capabilities of traditional MPSOC and programmable logic devices. The hybrid of programmable logic and vector processing elements paired with general purpose CPUs enable a massive increase in compute power. Furthermore, they can also incorporate specific AI machine learning, advance networking and encryption IP that will enable a new class of sensor processing platforms.  

An FPGA, MPSOC or ACAP platform is adaptable to work with many different types of sensors without changing any of the other peripherals. Therefore it is ideally suited for leveraging the modularity that the FPGA Mezzanine Card (FMC) standard offers. Combining COTS adaptable compute acceleration platforms with COTS FMC sensor interface cards provide solutions for many different applications, whether you need high bandwidth analog to digital converters (ADC) and digital to analog converters (DAC) for radar applications, multigigabit optical or Ethernet interfaces for digital communications, or video interfaces for image processing.

Logic-X has a wide experience and in-depth knowledge of FPGA, MPSOC, highspeed ADC and DAC and optical interfaces and is going to bring to market a new generation of FPGA, MPSOC, ACAP and FMC products that will answer the needs of our customers. It is our goal to provide a mix of Commercial off the shelf products (COTS) and customer specific derivative boards (MCOTS) when the COTS product is does not directly map on the customer application.
Stay tuned and register to our news letter to not miss our product announcements and do not hesitate to reach out to talk to us about your product needs.
 

  Erik Barhorst     17-05-2019 13:36     Reacties ( 0 )

Reacties (0)

Geen reacties gevonden.