New memory technologies promise denser and cheaper main memory, and may one day displace DRAM. However, many of them experience permanent failures due to wear far more quickly than DRAM. DRAM mechanisms that handle permanent failures rely on very low failure rates and, if directly applied to these new memories, are extremely inefficient.
In this talk, I will present recent work on tolerating failures in memories that wear out, a.k.a. wearable memories. The proposed mechanisms span multiple levels of the stack, including managed runtimes, operating systems and hardware, and are inspired by the triad reduce, reuse, and recycle. The mechanisms allow efficient use of wearable memories and include waste reduction via finer memory granularity disabling, error correction mechanisms tailored to the technology, and alternative uses for working bits in regions that have failed.
Karin Strauss is a researcher in Computer Architecture at Microsoft Research and a Computer Science and Engineering affiliate faculty at University of Washington. Her research focuses on challenges brought by silicon scaling, and she has done work on cache coherent chip multiprocessors, hardware support for detection and avoidance of concurrency bugs in shared-memory multithreaded software, and challenges associated to using emerging memory technologies as main memory.