Evolution of thread-level parallelism in desktop applications

Abstract

As the effective limits of frequency and instruction level parallelism have been reached, the strategy of microprocessor vendors has changed to increase the number of processing cores on a single chip each generation. The implicit expectation is that software developers will write their applications with concurrency in mind to take advantage of this sudden change in direction. In this study we analyze whether software developers for laptop/desktop machines have followed the recent hardware trends by creating software for chip multi-processing. We conduct a study of a wide range of applications on Microsoft Windows 7 and Apple's OS X Snow Leopard, measuring Thread Level Parallelism on a high performance workstation and a low power desktop. In addition, we explore graphics processing units (GPUs) and their impact on chip multi-processing. We compare our findings to a study done 10 years ago which concluded that a second core was sufficient to improve system responsiveness. Our results on today's machines show that, 10 years later, surprisingly 2-3 cores are more than adequate for most applications and that the GPU often remains under-utilized. However, in some application specific domains an 8 core SMT system with a 240 core GPU can be effectively utilized. Overall these studies suggest that many-core architectures are not a natural fit for current desktop/laptop applications.

A study of mobile device utilization

“…However, they also showed that desktop applications leveraged TLP very sparingly. This result was echoed 10 years later by Blake et al [1] with a similar study of TLP of contemporary software and hardware, when multi-core had become the norm rather than the exception in home and office desktops. They reported that 2-3 cores were more than adequate for almost all but a few domain specific applications like Video Authoring.…”

“…High GPU utilization also indicates that some of the parallelism is already offloaded from the CPU to the GPU. All these factors, and the history of the slow pace of exploiting parallelism in desktop environments [1], suggests that having many powerful cores is likely to be overprovisioning.…”

“…However, it sits in the idle state waiting for user input for most of the total running time, which would drag down the average utilization number. To avoid this bias, we use Thread Level Parallelism (TLP) [1,6]. TLP is defined as the machine utilization over the non-idle portions of the benchmark's execution.…”

“…This leads to the question: How much of the computation potential residing in multi-core CPUs is actually being utilized? On the desktop end, Blake et al [1] did a study on Thread Level Parallelism (TLP) on a suite of representative desktop applications. Their work was to measure the core utilization in modern multi-core CPUs, and they suggested that the number of cores that can be profitably used is less than 3 for most commonly used applications.…”

“…The GPU and mobile coprocessors on chip also reduce CPU load. All these factors, and the history of the slow pace of exploiting parallelism in desktop and mobile software environments [1,3], indicate that having many powerful cores is over-provisioning. Further analysis suggests that current mobile applications can benefit from a system with the flexbility to satisfy high performance and good energy-efficiency for different application phases.…”