Computing & Hardware Acceleration

FPGA Boards ///

nnn

Whether to accelerate or increase the performance of various algorithms, for very high-speed data transmission and real-time processing of all kinds of signals, our FPGA platforms are interesting alternatives to complement and even replace traditional processors. Available in industrial standards (PCIe, AMC, VPX) or more specific form factors, with various memory and interface options (FMC connectors, network links, NVMe, …), these boards can be integrated into embedded systems as well as servers or industrial PCs to cover their numerous use cases.

 

With the recent improvements of FPGA programming environments and the availability of functional examples, IPs and Frameworks, the exploitation of all the computing resources of our boards are simplified while guaranteeing the agility and scalability of your projects.

Machine Learning / AI ///

When it comes to reducing power consumption and system footprint of processing nodes (Edge Computing platforms for example), our FPGA-based solutions become obvious, especially given their remarkable performances to accelerate some algorithms like inference tasks for artificial intelligence.

 

We offer FPGA platforms dedicated to Machine Learning, both for embedded systems (Machine Vision for example) and datacenters, whether in the form of PCIe boards (similar to GPU boards generally used for such functions) or more compact modules (SSD form factor) to be integrated within the storage servers for Big Data Analysis and database acceleration applications for example.

Presentation about machine learning technology with scientist touching screen with artificial intelligence (AI), neural network, automation, and data mining words and computer icons

Computational Storage ///

ranks modern supercomputers in computational data center

Traditional data storage architectures based on SSD hard drives interconnected on PCIe and NVMe busses imply numerous limitations, whether in terms of latency, data transfer efficiency or processor availability. Our FPGA modules in U.2 or M.2 format (SSD form factor) are integrated directly on the PCIe or NVMe bus to act as compute nodes closely tied to the data while ensuring CPU Offload.

 

In addition to the possibilities offered by this type of architecture for database acceleration, it is above all an enormous gain on a more global scale with the possibility to perfom smart storage (Erasure Coding, Deduplication) as well as compression and encryption functions in a more efficient way by optimizing data transfers and consumption.

FPGA Servers and systems ///

To improve the TTM or simplify the use of our various FPGA solutions, we also offer pre-integrated solutions, including one or more FPGA boards interconnected and pre-configured around processors and various hardware and software resources.

 

Available in the form of servers (Teraboxes) or 19” chassis of various sizes (from 1U to 4U), these solutions are ready to be integrated into computing racks or embedded systems for rapid implementation. The user just has to focus on the installation of his algorithms and applications to have a turnkey solution for numerous use cases.

Solutions-blocAccueil-cartes-1024x687