ORBiS TREE · ENGLISH DOCUMENT
Concepts →GPU
A highly parallel processor widely used to train and run modern machine-learning models.
한국어
English
This English page is a curated translation layer linked to the Korean source article. Community changes are currently made on the Korean source, where the full revision history and anonymous edit trail are preserved.
A graphics processing unit, or GPU, contains many parallel compute units and high-bandwidth memory, making it well suited to the matrix and tensor operations used in deep learning.
How it works
Large-model workloads often distribute computation across multiple GPUs. Performance depends not only on arithmetic speed but also on memory capacity, memory bandwidth, interconnects, numerical formats, kernels, and communication overhead.
Why it matters
GPU availability and efficiency strongly affect the economics of model training and inference. Hardware choice must therefore be evaluated together with Quantization, batching, and serving architecture.
Related concepts
SOURCES
Sources
KNOWLEDGE LINKS
Continue from here
BACKLINKS