Black Box IEEE 1284 Podręcznik Użytkownika Pobierz pdf (Strona 25)

The Black Box Toolkit Page 25 of 98

Finally here are the results for the calibration utility run on the same PC but with “Normal”

priority. This illustration helps show that scheduling priority can have a large affect on

system performance and the accuracy of timing data you collect regardless of the software

you use. In the three screen shots the time to complete 500,000 samples ranges from

10.916 seconds to 12.530 seconds as a result of varying the scheduling priority.

Where a large number of ticks are taken during a sample this can be thought of as the

period the PC is “out to lunch”. In such periods timing accuracy will suffer. If during such

periods you responded to a stimulus it is likely that your response would not be registered

accurately. Hence your response timing would be adversely affected. This effect is clearly

illustrated when playing back video-based stimulus materials. During such presentations

response timing is so bad that experiment generator distributors recommend you don’t rely

on the timing measures taken.

The table below summarises the calibration findings for the example PC but under each of

the three scheduling modes.

Realtime High Normal

Elapsed Time 10.92 secs 10.94 secs 12.53

Mean Sampling Rate 45.81 kHz 45.688 kHz 39.91 kHz

Mean Ticks Per

Sample

78.15 78.35 89.70

Tick SD 8.96 70.03 2309.62

Min Ticks Per

Sample

73 73 73

Max Ticks Per

Sample

1667 35364 1320764

Tick Variance 80.21 4903.72 5334342.04

Pass or Fail? Pass Pass Fail

1 2 ... 20 21 22 23 24 25 26 27 28 29 30 ... 97 98

Brak uwag

Black Box IEEE 1284 Podręcznik Użytkownika Strona 25