Re: [opensuse] systemload / cpu's



Per Jessen said the following on 03/20/2010 12:43 PM:


Until you have enough load to use up more cores, there's no need to use
them.

Its not simply "load", its the nature of the load.

Multiple cores make sense for the big servers where there are many
simulates unblocked threads of execution.

For a desktop machine where there is, for the most part, 'just you'
running the foreground task, they offer little advantage.

Yes, you may be running a download in the background, but that's going
to be IO limited, not CPU limited.

In fact most of the time your computer is waiting for you to it a key.

Yes, there are situations where you might need more "cpu power", but
stop and think if the situation is really CPU-limited or IO-limited.

The example Hans gave got the right answer from Per.
The code for those applications is not multi-threaded.
It _could_ be, perhaps, but it isn't. Yes, 'rsync' could be written as
a producer-consumer pair of processes, but it would become more
complicated, more prone to design and logic errors and less 'general'.
Do you want to optimise it for two cores, and them repeat Hans' complain
when you are running a 4-core desktop?

--
HTTP is like being married: you have to be able to handle whatever
you're given, while being very careful what you send back.
--
To unsubscribe, e-mail: opensuse+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: opensuse+help@xxxxxxxxxxxx



Relevant Pages

  • Re: 24-42V to 12V regulator - any advice?
    ... Load current: Anywhere from that drawn by a higher impedance voltmeter to ... maximum foreseeable 3.3 A at some voltage near 5 V. ... P-cores such as ones by Ferroxcube. ... Another inductor option recently coming to my mind is toroids with ...
    (sci.electronics.design)
  • Re: bsnmpd returns incorrect hrProcessorLoad values
    ... I'm using cacti to monitor some servers running FBSD. ... I recently updated the servers to RELENG_8 and bsnmpd is ... returning negative values for the cores' load. ...
    (freebsd-stable)
  • Re: bsnmpd returns incorrect hrProcessorLoad values
    ... I'm using cacti to monitor some servers running FBSD. ... I recently updated the servers to RELENG_8 and bsnmpd is ... returning negative values for the cores' load. ...
    (freebsd-stable)
  • Re: em driver input errors
    ... Core2Quad 6600 and four (to load all cores in CPU) Intel ... (After this incident the switches now control the flood attack) ... the MAC is equal because i use if_laggfor this interfaces for load all cores in CPU ... If this a manageable switch, and supports tools - try to investigate what happens when errors appear. ...
    (freebsd-net)
  • Re: High CPU load when machine is idle (related to PROBLEM: Unusually high load average when idle in
    ... network or any other input running nothing but busybox) should not be ... Maybe load on a server doing absolutely nothing could be 0.00, ... does seem to be a bug somewhere as a load of 0.60 on an idle machine ... These tests have been done on a big desktop machine accessed remotely ...
    (Linux-Kernel)