Abstract

The Compiler Lease of Cache Memory (CLAM) project has previously explored the effects of lease-based replacement policies for benchmarks (programs) run on single-core with single-level and single-core with two-level cache architectural config- urations. This work builds and expands on this prior work by continuing the same line of inquiry for a dual-core with two-level cache architectural configuration. The lease-based replacement policies studied are: Compiler Assigned Reference Leasing (CARL), Phased Reference Leasing (PRL), Scope Hooked Eviction Leasing (SHEL), and Cross-Scope Hooked Eviction Leasing (C-SHEL). These are compared to the per- formance of the traditional Pseudo Least Recently Used (PLRU) replacement policy. Samples are collected from PolyBenchC benchmark runs using PLRU on a single- core with two-level cache configuration. These are then used to generate leases for the dual-core with two-level cache configuration. On the latter, the benchmarks run in pairs and in multi-tasking mode, i.e., one on each core with private L1 instruction and data caches, but with shared L2 cache and main memory. The number of misses seen

Publication Date

8-14-2024

Document Type

Thesis

Student Type

Graduate

Degree Name

Electrical Engineering (MS)

Department, Program, or Center

Electrical and Microelectronic Engineering, Department of

College

Kate Gleason College of Engineering

Advisor

Dorin Patru

Advisor/Committee Member

Chen Ding

Advisor/Committee Member

Mark Indovina

Comments

This thesis has been embargoed. The full-text will be available on or around 8/26/2025.

Campus

RIT – Main Campus

Available for download on Tuesday, August 26, 2025

Share

COinS