High-Performance Embedded Computing
What Is High-Performance Embedded Computing (HPEC)?
High-performance embedded computing delivers high levels of computing for mission-critical applications that need to perform in rugged and small environments. Embedded systems are integral for computing information between multiprocessors.
HPEC brings supercomputing performance once only possible in the data center to dense, ruggedized systems deployable in today’s most rigorous defense applications. The sheer amount of data produced by Electro-Optical and Infrared (EO/IR), Electronic Warfare (EW), radar, and Signals Intelligence (SIGINT) sensors for analysis demands multiple processing elements and I/O interconnected via high-throughput, low-latency switched fabrics. Even small, unmanned vehicles utilize sensors generating large volumes of data that require high performance computing capability at the edge of the network. This drives difficult size, weight, and power requirements to go along with the environmental requirements of a rugged vehicle.
One way to deal with the SWAP-C demands is by utilizing strategies and technology deployed by data centers in order to bring high-performance computing closer to the edge of the network. This includes shifting towards heterogeneous compute workloads.
Improved SWAP-C with Heterogeneous Workloads
The latest trends in heterogeneous computing are increasing efficiency and allowing more data to be processed with less power to allow high-level computing in smaller environments and devices. Rather than take a common, brute-force approach to all problems on a single type of processor, heterogeneous computing uses the best silicon for the job. By combining general-purpose processors with FPGAs, network processors, and GPUs, the most efficient solution can be used.
Advances in heterogeneous computing have driven multi-card solutions to single-card solutions and are now enabling single-chip solutions. A single package with a CPU, FPGA, and GPU not only minimizes space, but allows all devices to talk to each other over high speed interconnects and shared memory.
Open Standards Are Enabling HPEC
Industry standards like OpenVPX and SOSA provide guidelines and architectures allowing complex HPEC systems to be built in compact, ruggedized form factors. COTS products from a variety of vendors and are interoperable as a result of the standards. These architectures lend themselves to heterogeneous workloads and as multi-processor solutions move from the board level to the chip level, workloads that once took multiple slots in a VPX chassis can now be done in one. This means increased processing density in a variety of processing-intensive applications including EO/IR, radar, and SIGINT.
New Wave DV’s High-Performance Embedded Computing Options
New Wave DV stays at the forefront of high-performance computing and provides solutions in ruggedized, embedded form factors. Our latest line of ruggedized XMC cards allows customers to upgrade any VPX single board computer (SBC) with a series of co-processing XMCs. This allows for a variety of heterogeneous workloads in a single VPX slot.
The New Wave DV V1153 and V1160 make a large number of optical or copper high-speed serial interfaces accessible to a host CPU.
- V1153 provides a Xilinx UltraScale FPGA co-processor
- V1160 features an industry-leading NVIDIA Mellanox Connect X-5 network interface
Together, these cards provide great flexibility and bring a large number of heterogeneous computing possibilities to a SBC of your choosing. This means industry-leading 100G Ethernet or InfiniBand support alongside a flexible FPGA fabric in a single VPX slot. Contact us today if you’d like to discuss solutions for your high-performance embedded computing application.