QUAMTUM AND NEUROMORPHIC COMPUTING ON FPGA

HLS methodology enables the design of FPGA or Application-Specific Integrated Circuit (ASIC) systems using a C/C++ description, typically accompanied by a testbench. One key advantage of HLS is that designs can be compiled like any other program, making debugging easier than with traditional Hardware Description Languages (HDLs). Another benefit is the reduction of errors: HLS ensures that, once the design is simulated and the desired results are achieved, there are no edge cases with incorrect behavior.

While HLS is useful for algorithmic designs (e.g., sequential mathematical computations), it becomes less effective for highly complex or non-linear systems. In such cases, the classic FPGA/ASIC design flow using VHDL is more appropriate. VHDL is the preferred hardware description language for institutions like the European Space Agency and is the most widely used in Europe.

HDL-based development has traditionally been tied to vendor-specific tools, often creating inefficiencies in the industry—such as outdated development environments, limited integration with high-level languages like Python, or high costs in multidisciplinary projects (hardware + driver + PCB development).

These challenges have led to the emergence of open-source initiatives to streamline workflows. At ITCL, we stay continuously up to date to remain at the forefront of these technologies

Continuous Integration (CI) refers to the use of automated, high-quality testing procedures that run with each code change. CI ensures that newly introduced code is error-free and does not break existing functionality.

FPGA designs are rarely standalone. Typically, they need to interface with a complete computing system. This requires deep architectural knowledge and the ability to write software that connects the FPGA design to the target application. Two approaches are used: bare-metal, with no operating system, and embedded Linux, where the design runs within a Linux OS. The team is experienced in both, selecting the most suitable option depending on the project’s needs.

Quantum algorithms such as HHL (Harrow-Hassidim-Lloyd) and Variational Quantum Linear Solver (VQLS) efficiently solve systems of linear equations. These problems are foundational across data processing and physical system simulations, where extracting properties of solutions or feeding them into other algorithms is essential.

A key productivity driver in industry, combinatorial optimization includes problems like warehouse storage optimization or delivery route planning. Quantum approaches such as QAOA (Quantum Approximate Optimization Algorithm) and Quantum Annealing offer approximate methods that may outperform classical algorithms in finding optimal or near-optimal solutions.

QML has seen rapid growth in recent years, introducing novel algorithms, models, and tools. Quantum systems can work within exponentially large data spaces, offering new opportunities to handle and extract insights from large-scale data more efficiently and effectively.

A quantum-inspired technique with growing popularity, tensor networks offer efficient information compression and operation via tensor structures. ITCL has extensive experience in this area, applying tensor networks to simulate quantum systems, extract dataset properties, and optimize artificial intelligence models by reducing computational complexity.