Abstract
As the overhead of data retrieval becomes forbidding, bringing processor logic to the memory where the data reside becomes more energy-efficient. While traditional CMOS structures are unsuited to the tight integration of logic and memory, emerging spintronic technologies show remarkable versatility. This paper introduces a novel spintronics-based processing-in-memory (PIM) framework called computational RAM (CRAM) to solve data-intensive computing problems.
Original language | English (US) |
---|---|
Pages (from-to) | 42-46 |
Number of pages | 5 |
Journal | IEEE Computer Architecture Letters |
Volume | 17 |
Issue number | 1 |
DOIs | |
State | Published - Jan 1 2018 |
Bibliographical note
Funding Information:This work is supported by DARPA Non-Volatile Logic program, NSF SPX grant no. 1725420, and by C-SPIN, one of the six SRC STARnet Centers, sponsored by MARCO and DARPA. Chowdhury and Harms equally contributed to this work.
Publisher Copyright:
© 2017 IEEE.
Keywords
- CRAM
- MTJ
- STT-MRAM
- processing-in-memory
- spintronics