This tool aims to detect Silent Data Corruption (SDC) ocurring in supercomputers and large datacenters. The current implementation is in C but we will be adding a Python version soon. We also are planning to cover GPU memory as well. The way this tool works is rather simple. At the begining of the execution two hexadecimal paterns will be selected randomly. Then it allocates a large buffer of the available memory. Then, it will iterate doing the three following actions: write, sleep and read. When reading it will check whether any of the read patterns is different from the one written, if it is the case it will log it as a detected SDC. The next iteration the second pattern will be used for writing and it will continue swtiching between both paterns until the end of the execution.
leobago / sdcfinder Goto Github PK
View Code? Open in Web Editor NEWLicense: BSD 3-Clause "New" or "Revised" License