TY - JOUR
T1 - Low-Energy Shared-Current Write Schemes for Voltage-Controlled Spin-Orbit-Torque Memory
AU - Lee, Albert
AU - Alam, Irina
AU - Yang, Jiyue
AU - Wu, Di
AU - Pamarti, Sudhakar
AU - Gupta, Puneet
AU - Wang, Kang L.
N1 - Publisher Copyright:
© 1963-2012 IEEE.
PY - 2023/2/1
Y1 - 2023/2/1
N2 - Voltage-controlled (VC) spin-orbit-torque (SOT) magnetic random access memory (MRAM) is being considered as the next-generation magnetic memory with potential to achieve superior speed, power, and write error rates over existing MRAM technologies. By placing multiple VC devices on a single SOT bus, VC-SOT MRAM can also enable compact structures, in which multiple devices can be addressed individually yet programmed via a shared current. In this work, we propose two implementations of shared-current write: the horizontal shared current write (HSCW), which reduces the average SOT current per bit by the number of bits on the SOT bus, and the vertical shared current write (VSCW), which can further leverage data dependency for increased performance. We simulate the efficiency of the HSCW and VSCW using a Landau-Lifshitz-Gilbert (LLG)-based VC-SOT model and a 28-nm CMOS technology and show that HSCW and VSCW can achieve an energy saving of 74% and 40%-87%, respectively, in a 32-bit setting. Analysis of data patterns in common workloads finds that 40% of data share more than 85% common bits, for which VSCW can leverage for further improved performance.
AB - Voltage-controlled (VC) spin-orbit-torque (SOT) magnetic random access memory (MRAM) is being considered as the next-generation magnetic memory with potential to achieve superior speed, power, and write error rates over existing MRAM technologies. By placing multiple VC devices on a single SOT bus, VC-SOT MRAM can also enable compact structures, in which multiple devices can be addressed individually yet programmed via a shared current. In this work, we propose two implementations of shared-current write: the horizontal shared current write (HSCW), which reduces the average SOT current per bit by the number of bits on the SOT bus, and the vertical shared current write (VSCW), which can further leverage data dependency for increased performance. We simulate the efficiency of the HSCW and VSCW using a Landau-Lifshitz-Gilbert (LLG)-based VC-SOT model and a 28-nm CMOS technology and show that HSCW and VSCW can achieve an energy saving of 74% and 40%-87%, respectively, in a 32-bit setting. Analysis of data patterns in common workloads finds that 40% of data share more than 85% common bits, for which VSCW can leverage for further improved performance.
UR - http://www.scopus.com/inward/record.url?scp=85147211825&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85147211825&partnerID=8YFLogxK
U2 - 10.1109/TED.2022.3228831
DO - 10.1109/TED.2022.3228831
M3 - Article
AN - SCOPUS:85147211825
SN - 0018-9383
VL - 70
SP - 478
EP - 484
JO - IEEE Transactions on Electron Devices
JF - IEEE Transactions on Electron Devices
IS - 2
ER -