Glossary¶
This glossary defines the principal computing and imaging terms used throughout the Holoflow documentation.
| Term | Expansion | Meaning |
|---|---|---|
| CPU | Central processing unit | The general-purpose processor that runs host code and coordinates work submitted to accelerators. |
| CUDA | — | NVIDIA's parallel computing platform and programming model for GPU acceleration. |
| DAG | Directed acyclic graph | A directed graph without cycles, used by Holoflow to represent task and tensor dependencies. |
| DMA | Direct memory access | Hardware-supported data transfer that proceeds without continuous CPU intervention. |
| FFT | Fast Fourier transform | An efficient algorithm for computing a discrete Fourier transform. |
| GPU | Graphics processing unit | A massively parallel processor used to accelerate numerical workloads. |
| GPSC | General-purpose scientific computing | Scientific computation infrastructure intended to support multiple domains and processing methods. |
| HPC | High-performance computing | The use of highly parallel or specialized computing resources for demanding workloads. |
| LDH | Laser Doppler holography | A full-field optical imaging method that uses holographic detection and Doppler analysis to visualize blood flow. |
| PCA | Principal component analysis | A dimensionality-reduction method that projects data onto orthogonal components. |
| RAII | Resource acquisition is initialization | A C++ lifetime-management technique that binds resource ownership to object lifetime. |
| ROI | Region of interest | A selected spatial subset of an image or tensor used for focused processing or measurement. |
| STFT | Short-time Fourier transform | A Fourier analysis performed over successive time windows to describe changing frequency content. |
| GUI | Graphical user interface | A visual interface that allows users to interact with an application through controls, windows, and graphical elements. |