On-Die Dynamic Remapping Cache: Strong and Independent Protection Against Intermittent Faults

As process scaling continues, DRAM is getting more vulnerable to errors. System companies and vendors have introduced ECC protect against growing ECC, however, should be combined with a repair mechanism prevent non-transient faults from repeatedly producing errors overlapping accumulating into severe We propose novel approach memory improve reliability minimal overheads. On-die Dynamic Remapping Cache (DRC) minimizes the overheads by focusing on active faults. Most intermittent (e.g., Variable Retention Time) generate occasionally, number of at any one time significantly lower than total. DRC tracks activity severity run-time uses small cache inside remap This efficiency enables aggressive remapping bit faults, which eliminates fault accumulations improves reliability. Our evaluation shows can provide much stronger protection state-of-the-art schemes no performance degradation negligible chip area overhead.