20 lines (13 loc) · 735 Bytes

CUDA MATLAB Performance Notes

This repository contains examples and notes for using CUDA together with MATLAB.

Hardware example

CPU: Intel Dual Core T4200 2GHz
GPU: NVIDIA GeForce G103M

Observations

GPU programming efficiency is likely affected by the GPU driver and access layer.
Using MATLAB with CUDA, computation times were reduced by about half on Windows 7.
On Linux 2.6, the same approach showed a speedup of approximately 4x.
GPU memory usage during calculations is significantly lower than CPU-only processing.

Notes

These results depend on the specific hardware and software environment.
Modern CUDA and MATLAB versions may produce different performance characteristics.