mb-audit failure for Linux version

[tomcw]

In cea2, "mb-audit: a Mockingboard test suite for emulators", @audetto asked:

---

Interesting, and surprising when I run it on the linux version I got a failure.
Turns out that the issues was me running bigger batches of CPU code (16 ms vs 1 ms).

So the question I have now is:

if we know that 1ms is ok and 16ms is too long

https://github.com/AppleWin/AppleWin/blob/master/source/Windows/AppleWin.cpp#L132

when the MB is updated here https://github.com/AppleWin/AppleWin/blob/master/source/Windows/AppleWin.cpp#L217

Is there a maximum?
Is it possible that 1ms is too long as well?
Should the update really be done for each cpu instruction (like the video)?

[tomcw]

What are the failure codes you get?

In mb-audit there is a test (T6522_B) with a big gap (~$9000 cycles) between reads of the 6522 timer.
https://github.com/tomcw/mb-audit/blob/0c02f0e7f3e437c1aa62c7d80bdf4c80c4d4aa9e/chip-6522.a#L1695
But I know this isn't necessarily the same thing you are describing.

[audetto]

This is the error I got

mb-audit v0.5-beta, 2021
                    1  2  3  4  5  6  7
               $00:          ?  C
               $80:          ?  C
                SP:            ABV
65C02 detected
Slot #5 :Mockingboard failed test:SSI263
Test: 41:05:00
Expected:00 Actual:01

Now, the fix was simple, I just have to run cpu cycles in batches of 1ms rather than the 16.6ms I was doing before.
What is curious is

• batches of 15ms still work, 16ms or more is too long
• this problem did not manifest earlier (but I probably do not test enough)
• the real card does the equivalent of MB_PeriodicUpdate constantly rather than in batches (of 1 or 16 ms), can one write a test that fails with AW's 1ms period?

[audetto]

And this is what I get if I remove MB_PeriodicUpdate altogether

mb-audit v0.5-beta, 2021
                    1  2  3  4  5  6  7
               $00:          ?  ?
               $80:          ?  ?
                SP:
65C02 detected
Multi-card tests passed
Slot #5 :Press CTRL+RESET (ESC to skip)
Slot #4 :Press CTRL+RESET (ESC to skip)
Reset tests passed
Slot #5 :Tone: B:456
Slot #4 :Tone: B:456
All tests passed (total=$0419)

Should this be a failure?

[tomcw]

Well the ? just means that a 6522 was detected at either $00 or $80... but mb-audit was unable to match what it detected to any of the known cards.

And there's a clue here...

Slot #5 :Tone: B:456
Slot #4 :Tone: B:456

This means that it only found an AY-3-8913 at $80, but not at $00.
NB. If it found an AY8913 at only $00, it would display Slot #5 :Tone: A:123.
And if it found an AY8913 at $00 and at $80, it would display Slot #5 :Tone: A:123,B:456.

AY8913 detection is done by writing to the AY8913's registers, and then attempting to read them back.