<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Tue, Mar 31, 2015 at 10:44 AM, Richard W.M. Jones <span dir="ltr"><<a href="mailto:rjones@redhat.com" target="_blank">rjones@redhat.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">On Tue, Mar 31, 2015 at 08:32:16AM -0700, Dave Johansen wrote:<br>
> I am not familiar with the low level details of disk I/O but I'm sure that<br>
> they are far more complicated than my basic assumptions, but my concern is<br>
> how can a disk-bound process steal cycles from a CPU-bound one that is not<br>
> access the disk at all. The <a href="http://lwn.net" target="_blank">lwn.net</a> articles that drago01 linked to helped<br>
> shed some light on what's going on, but it sounds like there is still some<br>
> potential work that could be done to help improve the situation.<br>
<br>
</span>When you run the cpu load test program, where do you write the<br>
statistics to?<br></blockquote><div><br></div><div>I had been redirecting it directly to disk.<br></div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
For a fair test you probably want to change the program so it stores<br>
them in preallocated memory and prints them at the end of the test.<br></blockquote></div><br></div><div class="gmail_extra">You're right that is a problem because my "purely CPU bound task" was actually writing to disk every 10 seconds, so I've attached an updated version that pre-allocates a vector and stores the results there so they can be dumped when the users presses Ctrl-C. With this update, the "CPU bound task" should only using CPU and existing memory but I still see the same slow down in the "CPU bound task" when the disk I/O is happening.<br></div></div>