Data parallelism is a form of parallel computing that focuses on distributing data across multiple nodes to be operated on in parallel. It can be applied to regular data structures like arrays and matrices, resulting in a speedup of the process compared to sequential execution. The performance of data parallel programming models depends on the locality of data references.
UC Berkeley
Fall 2022
This course deepens students' understanding of computer architecture and the translation of high-level programs into machine language. Emphasis is on C and assembly language programming, computer organization, parallelism, CPU design, and warehouse-scale computing. Prerequisites include CS61A and CS61B or equivalent C-based programming experience.
No concepts data
+ 51 more concepts