Free Trial

Safari Books Online is a digital library providing on-demand subscription access to thousands of learning resources.

  • Create BookmarkCreate Bookmark
  • Create Note or TagCreate Note or Tag
  • PrintPrint
Share this Page URL

Chapter 2. Coding for Performance > Defining Performance

Defining Performance

There are two common metrics for performance:

  • Items per unit time. This might be transactions per second, jobs per hour, or some other combination of completed tasks and units of time. Essentially, this is a measure of bandwidth. It places the emphasis on the ability of the system to complete tasks rather than on the duration of each individual task. Many benchmarks are essentially a measure of bandwidth. If you examine the SPEC Java Application Server benchmark (SPEC jAppServer[1]), you’ll find that final results are reported as transactions per second. Another example is the linpack benchmark used as a basis for the TOP500[2] list of supercomputers. The metric that is used to form the TOP500 list is the peak number of floating-point operations per second.



  • Time per item. This is a measure of the time to complete a single task. It is basically a measure of latency or response time. Fewer benchmarks specifically target latency. The most obvious example of a latency-driven benchmark is the SPEC CPU benchmark suite, which has a speed metric as well as a rate metric.


You are currently reading a PREVIEW of this book.


Get instant access to over $1 million worth of books and videos.


Start a Free 10-Day Trial

  • Safari Books Online
  • Create BookmarkCreate Bookmark
  • Create Note or TagCreate Note or Tag
  • PrintPrint