For an operating system that required unmatched hardware resources, Windows Vista failed to translate the horsepower available into sheer speed. The same is not the case for Windows 7, as the evolution from Vista brings new levels of performance and reliability. The latest iteration of the Windows client, released to manufacturing on July 22, 2009, and already available for download to certain audiences, including MSDN and TechNet subscribers, delivered a superior user experience in terms of performance and reliability since early in Beta stage. An integral part of the work done to kick Windows up a notch is related to PerfTrack.

“PerfTrack is the feedback and monitoring system inside of Windows 7 that performs measurements on, well, all things related to the overall performance of the OS, especially as it relates to system responsiveness to user actions. So, when you click on something (an icon, a folder name, etc...), how long does it take for the user to receive an expected reaction from the system? What are the bottlenecks that lead to a poor experience (user-observable latency) when using some feature in Windows? Is the root problem in the design of the feature itself or with the underlying OS? Enter PerfTrack,” Microsoft revealed via Channel 9.

At the bottom of this article, you will be able to find and watch an embedded video detailing the work done to boost performance and reliability in Windows 7. Development Manager David Fields and his colleague, Group Program Manager Bill Karagounis, explain the work that went into Windows 7 and the way they used PerfTrack in order to deliver a superior experience compared to Windows Vista.

“PerfTrack is part of the effort of what we've been doing in terms of performance and it allows us to do a couple of things. One is to understand how the system is performing in the filed. We've always been able to run benchmarks and benchmarks are of course not always representative of what users are doing. It's a very simple way of instrumenting code and operations within the operating system, with a start and stop event, and then take those and skim them back. That allows us to get a response time curve of what the population looks like,” Fields explained.