RCGTEST: Bad RCG tests marring results
Several badly formed or improperly setup tests mar results of RCG tests, making verification of software more difficult.
Investigation of these results started with this ticket: #398 (closed)
None of these problems seem to depend on version.
-
RCG Core: iop loopback time for 16 bit model is always zero, which means either 'instantaneous' (same cycle) detection, or no detection. Sometimes other loopback times are intermittently zero. Perhaps threshold for detection is too low?
-
DACKILL Timer: DKT SEI Trip Timer and DKT SUS Trip Timer fail intermittently with very low values, zero or one. This test is often successful and can usually be passed by hand-running it.
-
Standard Filter Module 30Hz test fails about 50% of the time. Normal error must be too close to the threhold. Results might be improved by restarting models just before the test.
-
DAC order tests sometimes fail with output from alternate channels not in evidence. This is I think caused by certain models not starting properly before a test.