data parallelism
nounUK /ˈdeɪtə ˈpærəlɛlɪzəm/
US /ˈdeɪtə ˈpærəlɛlɪzəm/
Data parallelism is when a computer splits the same job into pieces and does them all at once.
The program uses data parallelism to process a million photos at the same time.
OTHER MEANINGS
2. A programming model where the same operation is applied to different parts of a dataset simultaneously.
In data parallelism, each processor handles a different chunk of the array.
3. A machine learning training strategy that splits a batch of data across multiple devices.
They trained the model with data parallelism across eight GPUs.
4. A hardware design approach where multiple processing units work on separate data streams.
The chip's data parallelism lets it crunch many numbers per clock cycle.