GPUs for Signal Processing Algorithms in MATLAB

Accelerating signal processing algorithms with GPUs and MATLAB

Several MATLAB® toolboxes for signal processing and communications contain highly optimized GPU functions that run on NVIDIA GPUs to reduce computation time. Although execution speed varies by application, users have achieved speedups of 30x for wireless communication system simulations.

Signal processing and communications algorithms contain structurally parallel data flows that involve iterative, computationally intensive, and time-consuming mathematical operations. NVIDIA GPUs contain thousands of highly specialized cores that operate in parallel to reduce execution time of these algorithms and accelerate simulation.

Examples of GPU accelerated applications

Examples of GPU accelerated communications systems applications: bit error rate (BER) simulation (top) and clutter simulation in airborne radar system (bottom).

Existing signal processing and communications algorithms can run on NVIDIA GPUs with minimal code changes.

GPU-Enabled Functionality

See also: MATLAB GPU computing, low-pass filter, high-pass filter, research with MATLAB