Optimizing GPU Cache Policies for MI Workloads.
Johnathan AlsopXianwei ZhangTsung Tai YehBradford M. BeckmannMatthew D. SinclairSrikant BharadwajAlexandru DutuAnthony GutierrezOnur KayiranMichael LeBeaneBrandon PotterSooraj PuthoorPublished in: IISWC (2019)
Keyphrases
- mutual information
- access patterns
- miss ratio
- database workloads
- prefetching
- query processing
- database systems
- data access
- parallel processing
- online transaction processing
- real time
- buffer pool
- memory hierarchy
- gpu implementation
- computer systems
- graphics hardware
- multithreading
- optimal policy
- response time
- hit rate
- parallel computing
- main memory
- parallel computation
- data transfer
- cache misses
- parallel implementation
- database