Messages in this thread | | | From | (Larry McVoy) | Subject | Re: Interesting scheduling times - NOT | Date | Thu, 24 Sep 1998 20:42:26 -0600 |
| |
Oliver Xymoron <oxymoron@waste.org>: : > Well, this is a fine theory - and it was what I thought when I started : > measuring things - but it's wrong in practice. If you take a suite of : > tests, lmbench for example, and do a bunch of runs and scatter plot them : > and stare at them you'll see patterns emerging. Now if the pattern was : > that most run times clustered around the min, then my feeling is that : > the min is the right number. Wherever they cluster up is the number I : > wanted because that was the number mostly likely to be seen. : : Again, I agree that generally the average is the number that's : interesting.
Average == mean Median == point where half the values < and half the values are >
An average will /not/ give you what you want. The median will do it. Try it - take a bunch of runs and plot mean, median, max, min and you'll see.
: But earlier you seemed to imply that the minimum is not : generally a meaningful number, because they were way out on the tails of : normal distributions.
Err, I never said that results follow a normal distribution. In fact, I would probably be the person that says they don't - normal, gaussian, whatever, real world results tend not follow them anywhere near as much as the modeling folks would like you to believe.
And I didn't say that the min is meaningless, I said that it isn't always representative. Feel free to take any benchmark and plot a bunch of runs and look at them. After you do that for a while, the median starts to look like a really good choice. At least it did for me.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |