Re: Timers and timing, was: MySQL Performance 6.0rc1

On Sat, 29 Oct 2005, Poul-Henning Kamp wrote:

> You forgot to tell us what
>
> 	sysctl kern.timecounter.hardware
>
> was on the machine, but I pressume ACPI-fast ?

Yes:

kern.timecounter.hardware: ACPI-fast
kern.timecounter.choice: TSC(-100) ACPI-fast(1000) i8254(0) 
dummy(-1000000)

With TSC on SMP, you see the expected sorts of results:

x 7SMP/clock_gettime_realtime
+ 7SMP/TSC_clock_gettime_realtime
* 7SMP/clock_gettime_poor
+--------------------------------------------------------------------------+
|*        +                                                               x|
|*        +                                                               x|
|*   *    +                                                            x  x|
|*   *    +                                                            x  x|
|*   *    +                                                            x  x|
|**  **  ++                                                           xx xx|
||MA_|   |A|                                                           |A_M|
+--------------------------------------------------------------------------+
     N           Min           Max        Median           Avg        Stddev
x  12          1404          1450          1444     1430.0833     19.327128
+  12           571           623           616     612.91667     13.627901
Difference at 95.0% confidence
         -817.167 +/- 14.1587
         -57.1412% +/- 0.990063%
         (Student's t, pooled s = 16.7221)
*  12           503           566           517     527.08333     26.352362
Difference at 95.0% confidence
         -903 +/- 19.5659
         -63.1432% +/- 1.36817%
         (Student's t, pooled s = 23.1083)

Robert N M Watson