Accelerating Sparse Attention with a Reconfigurable Non-volatile Processing-In-Memory Architecture.
Qilin ZhengShiyu LiYitu WangZiru LiYiran ChenHai Helen LiPublished in: DAC (2023)
Keyphrases
- processing elements
- memory management
- hardware implementation
- main memory
- reconfigurable hardware
- real time
- low cost
- parallel architecture
- functional units
- distributed processing
- associative memory
- compute intensive
- memory hierarchy
- heterogeneous computing
- computational power
- random access
- computation intensive
- dynamic reconfiguration
- memory access
- general purpose processors
- information processing
- general purpose
- data storage
- systolic array
- software architecture
- massively parallel
- field programmable gate array
- focus of attention
- parallel computers
- parallel processors
- reconfigurable architecture
- hardware architecture
- management system
- visual attention
- compressive sensing
- storage devices
- sparse data
- image processing algorithms
- computing systems
- sparse representation
- operating system
- data processing
- high dimensional