Using SimPoints in Diverse Simulation Environments
Erez Perelman, Michael Van Biesbrouck, Timothy Sherwood et al. · eScholarship (California Digital Library) · 2002
Modern architecture research relies heavily on detailed pipeline\nsimulation. Simulating the full execution of an industry standard benchmark\ncan take weeks to months to complete. To address this issue researchers have\nrecently proposed using Simulation Points (SimPoints) to increase the\nefficiency and accuracy of simulation. SimPoints are a small subset of\nexecution intervals that are representative of a program's phases, so that when\ncombined they accurately represent the complete execution of the program.\nThese SimPoints are found using clustering techniques based on the structure of\nthe executed code. The prior work on SimPoints assumed a somewhat advanced\nsimulation environment that supported check-points, had perfect warm-up\nability, and focused only on simulating single threaded architectures. In this\nresearch we explore what happens when you break these assumptions and provide\ntechniques that are capable of dealing with more complex simulation\nenvironments. We focus on environments that have the ability to fast-forward,\nbut to not checkpoint, and we extend the use of SimPoints to multi-threaded\narchitectures. We present results comparing our improved SimPoint techniques\nwith past SimPoint work and in addition provide a comparison to Random\nSampling.Pre-2018 CSE ID: CS2002-0727
Cited by 0 works.