In a candid interview, Microchip Technology, a global leader in embedded solutions, delved into its latest advancements with the introduction of the PIC64-HPSC and PIC64GX microprocessors (MPUs). These new 64-bit products are poised to revolutionize various industry segments, including aerospace, defense, industrial, automotive, and communications equipment, with a particular focus on enhancing intelligent edge applications. The conversation with Kevin So, Product Marketing Director of Microchip’s Communications Business Unit, sheds light on how these MPUs are strategically designed to meet the growing demands of the commercial space sector, particularly in Low Earth Orbit (LEO) constellations, while also offering robust capabilities in AI/ML processing, security, and fault tolerance.
With the introduction of the PIC64-HPSC and PIC64GX MPUs, how is Microchip positioned in the embedded solutions market, and which industry segments is the company strategically targeting?
Microchip is a worldwide leader in the 8-, 16- and 32- bit embedded solutions market. PIC64 expands our portfolio to include 64-bit products with feature sets that are aligned to what’s increasingly needed in the A&D, industrial, automotive and communications equipment markets. with an emphasis on enabling the intelligent edge.
How will the PIC64-HPSC family impact the commercial space sector, particularly in Low Earth Orbit (LEO) constellations?
The global space economy is growing by leaps and bounds, and the commercial space sector is really a key component of that growth. When talking to our partners in this sector, we continuously hear the need for more compute, more autonomy, more data connectivity and, increasingly, security. Given mission costs, cost to launch and development timeframes, the solutions also must be built to withstand the rigors of space, whether that be in Low Earth Orbit, on the Moon or beyond. That’s exactly what PIC64-HPSC delivers. And we take that a step further and provide multiple PIC64-HPSC versions to address multiple mission profiles with the same hardware and software footprint. This is what we refer to as “Scalable Mission Assurance.”
What are the primary features of the PIC64-HPSC MPUs which enable efficient AI/ML processing for autonomous space missions?
Any system that leverages AI/ML can be thought of as composed of two high level operations: 1) data ingestion and 2) data processing. For data ingestion, PIC64-HPSC offers groundbreaking support for Time-Sensitive Networking (TSN) features on port speeds up to 10Gbps. This is augmented by RDMA over Converged Ethernet (RoCEv2) which is a data center technique that allows data to be written directly into host memory – bypassing overhead and latency associated with the operating system.
From a data processing perspective, PIC64-HPSC leverages high-performance 64-bit RISC-V cores that not only support RISC-V Vector processing but additional vector extensions built into the ISA that provides an additional 6x increase in throughput versus RVV.
Lastly, we can’t forget about software and the AI/ML frameworks themselves. The PIC64-HPSC will support industry standard AI frameworks such as TensorFlow Lite and Pytorch. System developers working with PIC64-HPSC can choose multiple programming options from hand-tuned assembly, all the way to vector-optimized libraries.
How does the integration of RISC-V CPUs with vector-processing instruction extensions in the PIC64-HPSC MPUs enhance computational capabilities for space applications?
Sensor and image data is often organized with set bit widths, for instance 8-bits. Vector processing instructions allow multiple pieces of data to be processed at the same time- for instance, in matrix multiplications. So, at the end of the day it comes down to efficiency and throughput.
What are the high-speed network connectivity options available in the PIC64-HPSC MPUs, and how do they enhance performance?
As on Earth, bandwidth requirements in space applications continue to grow. That’s why with PIC64-HPSC we’ve integrated up to 20×10 GbE ports. Each port is augmented with support for IEEE’s upcoming TSN profile for Aerospace Onboard Ethernet Communications (P802.1DP). This allows us to move more data faster, with more determinism, more reliability and with lower latency.
PIC64-HPSC also supports PCIe® and Compute Express Link® (CXL) to allow us to connect to a wide range of space-based peripherals including SSDs and AI accelerators.
What advanced security features are included in the PIC64-HPSC MPUs, and how do they protect space missions from cybersecurity threats?
Traditionally, security has not been a top requirement in space applications. However, we see that quickly changing. Many countries and enterprises now see space as a key part of their critical infrastructure and as a result are driving new requirements into the MPUs used in space. PIC64-HPSC delivers on these requirements by integrating what we consider multi-layer defense-grade security which begins in our manufacturing facilities and extends throughout the entire lifecycle of the product. Secure-boot, anti-tamper and data encryption are all table-stakes features these days. But we’ve taken those features and integrated post-quantum cryptographic techniques to ensure that our design and our customer’s designs are secure both now and in the fast-approaching quantum computing era.
Please explain the high fault-tolerance capabilities of the PIC64-HPSC MPUs and how do they ensure reliable operation in space environments?
Fault-tolerance is of critical importance in any space mission. With PIC64-HPSC, we’ve gone well beyond traditional ECC and SECDED type error corrections for our integrated memories. We support dual-core lockstep in our application cores to allow early detection of faults, we’ve integrated WorldGuard which provides end-to-end hardware partitioning and fault-isolation, and we’ve also developed new innovations to protect our NVM and DDR memory interfaces. Lastly, TSN provides fault-tolerance and reliability for our ethernet communications links.
