FUSE: Fusing STT-MRAM into GPUs to Alleviate Off-Chip Memory Access Overheads.
Jie ZhangMyoungsoo JungMahmut T. KandemirPublished in: HPCA (2019)
Keyphrases
- memory access
- random access memory
- memory bandwidth
- data access
- shared memory
- main memory
- parallel programming
- memory management
- processing units
- external memory
- data fusion
- response time
- access patterns
- parallel processing
- high volume
- instruction set
- general purpose
- design considerations
- computational power
- operating system
- cache misses
- website
- parallel computation
- data management
- nearest neighbor