|Download Help (Windows Only)|
The GPU Analysis Toolkit allows you to communicate with a graphics processing unit (GPU) from LabVIEW. A GPU can compute large, computationally challenging problems while the CPU works as the primary processor. This functionality results in co-processing for faster performance than if the application uses only the CPU. Large problem sizes and parallel computations can decrease performance because of CPU limitations in both the number of cores and the amount of cache per core; however, GPUs do not have these same limitations. Use the GPU Analysis Toolkit for designing the block diagram to offload large FFT or BLAS operations to a GPU device and to upload the results back to the CPU.
|Note To use the GPU Analysis Toolkit, you must have the NVIDIA CUDA Toolkit and an NVIDIA display driver. Refer to the readme_GPUAnalysis.html file, located in the labview\readme directory, for more information about these requirements.|
The following figure shows the algorithm flow when offloading FFT operations in a multichannel FFT application. In this figure, blue signifies a process on the CPU and green signifies a process on the GPU.