Save Big on Coursera Plus. 7,000+ courses at $160 off. Limited Time Only!
Explore a cutting-edge approach to dynamic memory management in this 15-minute conference talk from MPLR 2023. Delve into the challenges of minimizing memory fragmentation and the current gaps in understanding workload-allocator interactions. Learn about a novel trace-based simulation methodology that uses two-dimensional rectangular bin packing to represent these interactions. Discover how this approach relates to maximum resident set size (RSS) and a newly defined fragmentation metric. Gain insights into the performance of modern allocators across single- and multi-threaded workloads, and understand the statistical significance of the findings. Ideal for computer scientists, software engineers, and researchers interested in advanced memory management techniques and their impact on system performance.