Re: a faster way to gettimeofday? rdtsc strangeness

[Jamie Lokier]

H. Peter Anvin wrote:
> I would like to see the error bars on your measurement.

I just ran it on my laptop 20 times in a row and it printed exactly the
same values every time:

	Kernel reports clock frequency:  366.601 MHz
	Measured clock frequency:        366.584 MHz

Well we need more precision than that.  So here goes:

	Kernel reports clock frequency:  366.601000 MHz

	Measured clock frequency:        366.584116 MHz
	Measured clock frequency:        366.584071 MHz
	Measured clock frequency:        366.584108 MHz
	Measured clock frequency:        366.584116 MHz
	Measured clock frequency:        366.584090 MHz
	Measured clock frequency:        366.583968 MHz
	Measured clock frequency:        366.584069 MHz
	Measured clock frequency:        366.584106 MHz
	Measured clock frequency:        366.584172 MHz
	Measured clock frequency:        366.584093 MHz
	Measured clock frequency:        366.584077 MHz
	Measured clock frequency:        366.583939 MHz
	Measured clock frequency:        366.584083 MHz
	Measured clock frequency:        366.584118 MHz
	Measured clock frequency:        366.584062 MHz
	Measured clock frequency:        366.584085 MHz
	Measured clock frequency:        366.584066 MHz
	Measured clock frequency:        366.610417 MHz <- anomaly :-)
	Measured clock frequency:        366.584109 MHz
	Measured clock frequency:        366.584062 MHz
	Measured clock frequency:        366.582293 MHz
	Measured clock frequency:        366.584103 MHz
	Measured clock frequency:        366.584093 MHz
	Measured clock frequency:        366.584107 MHz
	Measured clock frequency:        366.584059 MHz
	Measured clock frequency:        366.583985 MHz
	Measured clock frequency:        366.584133 MHz
	Measured clock frequency:        366.584092 MHz
	Measured clock frequency:        366.584087 MHz
	Measured clock frequency:        366.555754 MHz <- another deviant!
	Measured clock frequency:        366.584044 MHz
	Measured clock frequency:        366.584109 MHz
	Measured clock frequency:        366.584084 MHz
	Measured clock frequency:        366.583608 MHz
	Measured clock frequency:        366.584112 MHz
	Measured clock frequency:        366.584134 MHz
	Measured clock frequency:        366.584049 MHz
	Measured clock frequency:        366.584090 MHz
	Measured clock frequency:        366.584155 MHz
	Measured clock frequency:        366.584006 MHz
	Measured clock frequency:        366.584182 MHz
	Measured clock frequency:        366.584009 MHz

It takes the median of 1000 samples of the TSC time taken to do a
rdtsc/gettimeofday/rdtsc measurement, then uses that as the threshold
for deciding which of the subsequent 1000000 measurements are accepted.
Then linear regression through the accepted points.

I see a couple of results there which suggests a probable fault in the
filtering algorithm.  Perhaps it should simply use the smallest TSC time
taken as the threshold.

Sorry, I'd need permission to post the code.

enjoy,
-- Jamie