0

My load average shows 2.0 in 5 and 15 mins.

My CPU is 1x2GHz, I know that 1.0 = 100% but what about GHz? Does it affect that?

Ahmed Hamdy
  • 1

1 Answers1

0

Load average is more complex than simply 1.0 == 100%. As shown here and in many other pages, you'll see that load average is actually relative to the amount of cores you have.

So while a one core system is at capacity at 1.0, a dual core system is at capacity at 2.0. In other words, you need to relate it to the number of available cores when making sense of utilization of your system.

To answer your question about Ghz, the answer is "no". Gigahertz has no bearing on the load average reported by your system.

For more information, see also: What does load average mean in unix/linux?

Community
  • 1
hexparrot
  • 101
  • 3
  • I understand that. I was asking about the "2GHz". Does it affect how the load average results are interpreted? – Ahmed Hamdy May 29 '13 at 15:14
  • @AhmedHamdy No, it does not. Load average is the ratio between the amount of time a particular core spends active vs. sleeping. A faster core will have a lower load average, but `0.5` is still `50%` of a core. – Darth Android May 29 '13 at 15:18
  • The loadavg numbers are more than just the CPU. `0.5` doesn't necessarily mean the CPU or a core has `50%` load. It can also mean the CPU is completely idle and a process is waiting for I/O. – scai May 30 '13 at 08:00
  • According to this page, http://www.linuxjournal.com/article/9001: "In fact, it is precisely the CPU load that is measured, because load averages `do not include any processes or threads waiting on I/O`, networking, databases or anything else not demanding the CPU." – hexparrot May 30 '13 at 14:32
  • Interesting. Most articles will tell you a different thing and also my `man uptime` does: *System load averages is the average number of processes that are either in a runnable or uninterruptable state. [...] A process in uninterruptable state is waiting for some I/O access, eg waiting for disk.* Now when I do a `find / -perm 755 >/dev/null` my load jumps from `0.1` to around `0.7` while `find` only causes about `10%` CPU load (and `top` reports a wait of around `35%`). To me this clearly shows that load avg *does* include I/O. – scai May 31 '13 at 06:22
  • I trust `man` implicitly; thus, I'd concede you're right on this one. – hexparrot May 31 '13 at 15:04