An efficient method for evaluating RRAM crossbar array performance

• An efficient reduction method to reduce the size of RRAM array up to 1024 times.
• Compatible with both 1R and 1S1R structure in DC simulation of reading and writing.
• Compatible with random resistance distributions.
• Error analysis shows a good accuracy.

An efficient method is proposed in this paper to mitigate computational burden in resistive random access memory (RRAM) array simulation. In the worst case scenario, a 4 Mb RRAM array with line resistance is greatly reduced using this method. For 1S1R-RRAM array structures, static and statistical parameters in both reading and writing processes are simulated. Error analysis is performed to prove the reliability of the algorithm when line resistance is extremely small compared with the junction resistance. Results show that high precision is maintained even if the size of RRAM array is reduced by one thousand times, which indicates significant improvements in both computational efficiency and memory requirements.