PCM Conductance Test Data

Using Teradyne Magnum II automatic testing system, device control is achieved through SET/RESET step pulse programming verification scheme, and raw experimental data such as conductivity and random noise are collected at multiple time scales.

Based on this data, a drift compensation algorithm is proposed to achieve 98.08% handwriting recognition accuracy and 33.3 TOPS/W hardware energy efficiency. The dataset can support replication and expansion research in areas such as phase-change memory internal computation, neural network hardware, and multi-level storage device stability.