Reinforcement learning based self-adaptive voltage-swing adjustment of 2.5D I/Os for many-core microprocessor and memory communication.
Hantao HuangSai Manoj Pudukotai DinakarraoDongjun XuHao YuZhigang HaoPublished in: ICCAD (2014)
Keyphrases
- reinforcement learning
- external memory
- main memory
- memory subsystem
- state space
- power system
- high speed
- cost model
- electrical power
- memory requirements
- communication networks
- model free
- learning algorithm
- disk accesses
- function approximation
- optimal policy
- high voltage
- database systems
- transmission line
- random access
- temporal difference
- action selection
- design methodology
- block size
- memory space
- reinforcement learning algorithms
- memory efficient
- memory management
- control parameters
- index structure
- special purpose hardware