This 2005 paper was an update to the original Perceived Performance paper, necessitated by the advent of Hypervisors and Virtual Machines.

This is the original paper I wrote in 2003 that coined the term "Perceived Performance", what folks today call "Perceived End-User Performance".  This paper kicked off a mini industry in how we measure performance from the end-user perspective, a major change from the traditional ways in which we measured performance of remote systems.

A paper explaining how Windows performance counters and tool like the Task Manager lie about performance when you run Windows inside a Virtual Machine.

A very old paper looking at how individual threads of a process have affinity in the Windows thread scheduler and how that helps performance due to on-chip caches.