Oprofile is the way to instrument user and kernel programs on Red Hat Enterprise Linux. AMD complements the base oprofile infrastruction, by offering an easy to use user interface combined with the ability to select common hardware events like cache misses, branch mispredicts and other CPU pipeline stalls. In this talk, AMD will review the hardware architecture of current AMD64 optheron chip technologies and focus on the profile counter capabilities. We will also use AMD's CodeAnalyst software on several usermode and kernel focused benchmarks to demonstrate how to optimize code for Red Hat Enterprise Linux on AMD Servers. Case studies of CPU, Fiber Channel I/O and Network workloads will be analyzed to see where an application might want to tune the operating system or re-think how their application is coded to achieve optimal performance.