Accelerating ML Recommendation With Over 1, 000 RISC-V/Tensor Processors on Esperanto's ET-SoC-1 Chip.
David R. DitzelPublished in: IEEE Micro (2022)
Keyphrases
- low power consumption
- low power
- instruction set
- signal processor
- low cost
- memory subsystem
- power consumption
- maximum likelihood
- high speed
- embedded systems
- multithreading
- single chip
- level parallelism
- high order
- collaborative filtering
- recommender systems
- application specific
- parallel processing
- parallel algorithm
- higher order
- hardware and software
- real time
- recommendation systems
- floating point
- memory access
- tensor factorization
- computer architecture
- high density
- application specific integrated circuits
- parallel processors
- physical design
- parallel computing
- image sensor
- dimensionality reduction
- user preferences
- cmos technology
- personalized recommendation
- circuit design
- recommendation algorithms
- high performance computing
- single processor
- vlsi implementation
- field programmable gate array
- high end
- general purpose
- distributed memory