Posts

Welcome to the new Architecture Reading Group

by at

Objective
Recent advances in Computer Architecture – particularly the transition to Multicore Architectures as a vehicle for continued performance growth – have had a profound impact on most fields of computing from programming languages and compilers, to operating systems and high performance computing.
The goal of this reading group is to discuss the most recent advances in Computer Architecture and their impact on other fields of computing. To that end, we will discuss recent high-impact publications from top conferences in architecture and related fields. Suggestions for reading material are encouraged from all students and faculty involved.
Students and faculty interested in architecture are invited to attend. Even if your research is not in Computer Architecture, but you are interested in how changes in this field will impact your area, you are more than welcome to attend.

Logistics:
To get the most out of this reading group, everyone involved is encouraged to read the assigned papers before the meeting. One participant will be responsible for leading the discussion at each meeting. This means preparing 2-3 slides to introduce the topic (1 slide) and highlight some discussion points for the meeting (strengths and weaknesses, interesting ideas, etc).

Cooperative Boosting: Needy Versus Greedy Power Management, Indrani Paul et al, ISCA 2013

by at

Abstract
This paper examines the interaction between thermal management techniques and power boosting in a state-of-the-art heterogeneous processor consisting of a set of CPU and GPU cores. We show that for classes of applications that utilize both the CPU and the GPU, modern boost algorithms that greedily seek to convert thermal headroom into performance can interact with thermal coupling effects between the CPU and the GPU to degrade performance. We first examine the causes of this behavior and explain the interaction between thermal coupling, performance coupling, and workload behavior. Then we propose a dynamic power-management approach called cooperative boosting (CB) to allocate power dynamically between CPU and GPU in a manner that balances thermal coupling against the needs of performance coupling to optimize performance under a given thermal constraint. Through real hardware-based measurements, we evaluate CB against a state-of-the-practice boost algorithm and show that overall application performance and power savings increase by 10% and 8% (up to 52% and 34%), respectively, resulting in average energy efficiency improvement of 25% (up to 76%) over a wide range of benchmarks.

ISCA13_Indrani_Paul (1)