Cache-Aware Roofline Model: Performance, Power and Energy-Efficiency Limits of Multi-cores

Speaker:  Leonel Sousa – INESC-ID, Rua Alves Redol, 9, Portugal
Topic(s):  Hardware, Power and Energy , Architecture, Embedded Systems and Electronics, Robotics , Web, Mobile and Multimedia Technologies , Applied Computing

Abstract

In this talk, a set of fundamental Cache-aware Roofline Models (CARMs) are presented, which describe the performance, power, energy and energy-efficiency upper-bounds of parallel architectures, namely multi-core CPU and GPU architectures. These models evaluate how key micro-architectural aspects, such as accessing different functional units or different memory hierarchy levels, affect the attainable performance, power and energy-efficiency. The proposed models are validated on several CPU and GPU architectures with a very high accuracy, achieved by relying on hardware counters and a set of custom performance/power monitoring tools. Besides, the experimental results show the ability of the proposed CARMs to provide more intuitive and useful guidelines than the state-of-the-art approaches, especially when optimizing and characterizing real-world applications from different benchmark suites. The performance CARM was also recently integrated by Intel as a fully supported feature into the Intel Advisor, and it is described as “an incredibly useful diagnosis tool (...) that developers can use to guide them (in the application optimization process), ensuring that they can squeeze the maximum performance out of their code with minimal time and effort''.

About this Lecture

Number of Slides:  60 - 150
Duration:  60 - 150 minutes
Languages Available:  English
Last Updated: 

Request this Lecture

To request this particular lecture, please complete this online form.

Request a Tour

To request a tour with this speaker, please complete this online form.

All requests will be sent to ACM headquarters for review.