Abstract

We are in an era of highly data-intensive applications, and the existing memory technologies are inadequate to meet their challenges. Non-Volatile Memories (NVMs) have emerged as a cost-effective alternative to the conventional SRAM based Last Level Caches (LLC) and DRAM-based main memories; however, they suffer from limited write endurance. Applications having non-uniform writes will cause heavily written blocks to fail faster than lightly written blocks, thereby reducing the lifetime of NVMs. Most of the modern processors use split organization in the first level cache and unified organization in the subsequent cache levels. Our proposed approach, ViSC (Virtually Split Cache) explores the write variation across the data and instruction blocks by virtually splitting unified LLC for wear-leveling. The logical mapping of LLC ways into instruction and data is interchanged periodically to distribute the writes uniformly. Our experimental results show that ViSC reduces the write variations significantly and improves the lifetime of NVMs by 1.94, 2.06, and 1.72 times for unicore, dual-core, and quad-core, respectively, by incurring negligible power and area overheads.

Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call

Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.