Vasily Volkov

EECS Department, University of California, Berkeley

Technical Report No. UCB/EECS-2016-143

August 12, 2016

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2016/EECS-2016-143.pdf

Modern commodity processors such as GPUs may execute up to about a thousand of physical threads per chip to better utilize their numerous execution units and hide execution latencies. Understanding this novel capability, however, is hindered by the overall complexity of the hardware and complexity of typical workloads. In this dissertation, we suggest a better way to understand modern multithreaded performance by considering a family of synthetic workloads, which use the same key hardware capabilities – memory access, arithmetic operations, and multithreading – but are otherwise as simple as possible.

One of our surprising findings is that prior performance models for GPUs fail on these workloads: they mispredict observed throughputs by factors of up to 1.7. We analyze these prior approaches, identify a number of common pitfalls, and discuss the related subtleties in understanding concurrency and Little’s Law. Also, we help to further our understanding by considering a few basic questions, such as on how different latencies compare with each other in terms of latency hiding, and how the number of threads needed to hide latency depends on basic parameters of executed code such as arithmetic intensity. Finally, we outline a performance modeling framework that is free from the found limitations.

As a tangential development, we present a number of novel experimental studies, such as on how mean memory latency depends on memory throughput, how latencies of individual memory accesses are distributed around the mean, and how occupancy varies during execution.

Advisors: James Demmel

---

BibTeX citation:

@phdthesis{Volkov:EECS-2016-143,
    Author= {Volkov, Vasily},
    Title= {Understanding Latency Hiding on GPUs},
    School= {EECS Department, University of California, Berkeley},
    Year= {2016},
    Month= {Aug},
    Url= {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2016/EECS-2016-143.html},
    Number= {UCB/EECS-2016-143},
    Abstract= {Modern commodity processors such as GPUs may execute up to about a thousand of physical threads per chip to better utilize their numerous execution units and hide execution latencies. Understanding this novel capability, however, is hindered by the overall complexity of the hardware and complexity of typical workloads. In this dissertation, we suggest a better way to understand modern multithreaded performance by considering a family of synthetic workloads, which use the same key hardware capabilities – memory access, arithmetic operations, and multithreading – but are otherwise as simple as possible.

One of our surprising findings is that prior performance models for GPUs fail on these workloads: they mispredict observed throughputs by factors of up to 1.7. We analyze these prior approaches, identify a number of common pitfalls, and discuss the related subtleties in understanding concurrency and Little’s Law. Also, we help to further our understanding by considering a few basic questions, such as on how different latencies compare with each other in terms of latency hiding, and how the number of threads needed to hide latency depends on basic parameters of executed code such as arithmetic intensity. Finally, we outline a performance modeling framework that is free from the found limitations.

As a tangential development, we present a number of novel experimental studies, such as on how mean memory latency depends on memory throughput, how latencies of individual memory accesses are distributed around the mean, and how occupancy varies during execution.},
}

EndNote citation:

%0 Thesis
%A Volkov, Vasily 
%T Understanding Latency Hiding on GPUs
%I EECS Department, University of California, Berkeley
%D 2016
%8 August 12
%@ UCB/EECS-2016-143
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2016/EECS-2016-143.html
%F Volkov:EECS-2016-143