Peer-Review Record
Compute-in-Memory for Numerical Computations
Micromachines 2022, 13(5), 731; https://doi.org/10.3390/mi13050731
by
Dongyan Zhao 1, Yubo Wang 1, Jin Shao 1, Yanning Chen 1,2, Zhiwang Guo 3,*, Cheng Pan 1, Guangzhi Dong 2, Min Zhou 1, Fengxia Wu 2, Wenhe Wang 1, Keji Zhou 3,* and Xiaoyong Xue 3
Reviewer 1:
Akhilesh Jaiswal
Akhilesh Jaiswal
Reviewer 2:
Rui Yang
Rui Yang
Micromachines 2022, 13(5), 731; https://doi.org/10.3390/mi13050731
Submission received: 25 February 2022
/
Revised: 6 April 2022
/
Accepted: 18 April 2022
/
Published: 2 May 2022
(This article belongs to the Special Issue New Advances in Ionic-Drift Resistive Switching Memory and Neuromorphic Applications)
Round 1
Reviewer 1 Report
The authors try to review use of RRAM CIM for PDE. While the topic is relevant and important. The paper needs significant rework:
- The paper contains many grammatically incorrect statements like -the 3-D vertical 92 ReRAM was emerged in 2009. There are many such statements.
- The paper while provides good overview of the math of PDEs, but fails to critically analyze different RRAM works. The RRAM PDE works are mentioned as if they are the abstract from respective paper, without any mention if drawback, challenges, novelty etc
- The paper also does not do justification to non-idealities of RRAM, and outlook on where such PDEs can be used. How non-idealities can be mitigated, what is the mitigation cost etc
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The authors present a review about computing in memory (CIM) for solving partial differential equations (PDE). Such review is useful given the recent developments in this field, but following issues should be addressed.
- The manuscript introduces the numerical methods, and then some examples of using CIM for PDE solver. But the link between them is missing. It is suggested to provide more details about how CIM array can be adapted to do the numerical calculation, such as weight mapping, sensing techniques, etc., before introducing the examples.
- The CIM based PDE solver should be compared with non-CIM computing techniques, to show the difference in energy efficiency, accuracy, latency, as in Table 2.
- It is suggested to introduce the potential application scenarios for using CIM based numerical computation.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The authors have taken care of my concerns.
