Mission Elevator sound (yet again?)

[Toni Wilen]

It needs to be proper test, not something that I can immediately criticize for not proving practically anything useful at all.

Because game bug is stupid way of starting the audio and waiting for audio interrupt (start audio, do something, clear interrupt, poll for wait interrupt. When correct way is clear the interrupt request before starting anything that can cause interrupts), it is very timing sensitive (was bitplane dma active between start and clear, or other audio channel, or sprite, or copper, or was cpu executing other interrupt exception at that time and so on..). It can look like working perfectly when in reality it can still happen with extremely low probability. (when exactly right combination of DMA channels and CPU state happens)

Also game "version" needs to included and matched. chipset (a1000 or OCS or ECS all of them can have minor differences in cycle usage if program does weird things).

[amilo3438]

Originally Posted by Toni Wilen

It needs to be proper test, not something that I can immediately criticize for not proving practically anything useful at all.

I wonder how the "proper test" should look like ?
(At least, I am not sure how it should be done.)

What I wonder, for the beginning, does the same happens on a real machine with the test I am using? If yes, then we could close these thread imedeitly, and if not there must be something wrong in the versions after the winuae v2.3.0! (as the same also does not happen in the v2.3.0, at least I could not provoke it)

[DamienD]

Originally Posted by amilo3438

I wonder how the "proper test" should look like ?
(At least, I am not sure how it should be done.)

Well, since this is an old game, the floppy disk version only runs with:

Chipset = OCS
Kickstart = 1.2
RAM = 512KB Chip

Someone with a real Amiga who has the above specs needs to test

[amilo3438]

Existing game versions:

Mission Elevator (1987)(Softgold).adf (CRC 027280B0)
Mission Elevator (1987)(Softgold)[a].adf (CRC 671147C4)
Mission Elevator (1987)(Softgold)[a2].adf (CRC D0E5E378)
(sps1282)missionelevator.ipf (CRC A7372BE2)


Working on WinUAE quickstart configurations:

A500, OCS Agnus, KS1.2, 512KB Chip RAM
A500, OCS Agnus, KS1.3, 512KB Chip RAM
A500, ECS Agnus, KS1.3, 512KB Chip RAM
A500, ECS Agnus, KS1.3, 1MB Chip RAM (Note: or with KS2.04 / KS3.1) (EDIT: and NOT A500+)


Game tested to work on A500 WinUAE quickstart configurations by using simple test explained in post #60 !
(works "fine" on v2.3.0 and starts with "the sound problem" from v2.3.1)

Originally Posted by DamienD

Someone with a real Amiga who has the above specs needs to test

[prowler]

Originally Posted by amilo3438

Yes, I would like to test them.

You can download them here.

[amilo3438]

Originally Posted by prowler

You can download them here.

Thanks for betas!


Game version used: Mission Elevator (1987)(Softgold).adf (CRC 027280B0)
WinUAE quickstart configurations used: A500, OCS Agnus, KS1.2, 512KB Chip RAM

Test results: (by using simple test explained in post #60)

winuae231b1 = the problem appeared 1x in 5 min
winuae231b6 = the problem appeared 1x in 5 min
winuae231b7 = the problem appeared 0x in 5 min (pure luck I guess)
winuae231b8 = the problem appeared 3x in 1 min -> http://eab.abime.net/showpost.php?p=723330&postcount=55
winuae231b9 = the problem appeared 3x in 1 min
winuae231b13 = the problem appeared 3x in 1 min

Obviously the problem starts to appear quite often from the beta 8 version! (winuae problem !?)

Also, beta 1 and beta 6 shows us that the game problem still exists, but it doest appear often.
(I guess this is similar as on a real hardware, i.e. see yt-video on post #41)


EDIT: Several observations that may or may not be helpful:

Regarding the beta 7:

(1) If CPU Frequency is changed from 7MHz (A500) to 14MHz (A1200) or 28MHz (2x) the problem still does not appear in beta 7 !
(but if emu is started in "cycle-exact=disabled" mode the problem will appear very soon !)

(2) If emu is started with "sound=disabled" the game will stop after the title screen !
(and will not continue when the sound is enabled again... it also stops the game if sound is disabled during game play)


Regarding the beta 8:

(1) Changing the CPU Frequency does not change anything - the problem is still there !

(2) Same as on beta 7 !

[Toni Wilen]

I did some logic analyzer checks and found new undocumented feature: audio DMA start interrupt (possibly other audio interrupts too) is delayed by 2 cycles. (for example, audio channel 1 DMA position is $13, INTREQR read access at cycle $14: request bit is not set. Cycle $15: interrupt request bit is finally set)

Extra 2 cycles seems to be _exactly_ enough to "fix" the sound problem.

EDIT: It seems to be fix the problem, 5 minute test game worked fine.

[amilo3438]

Great news, thanks for the effort!


EDIT:

I was tested the game again on a few days older WinUAE 3.4.1 (Public Beta 0, 2017.01.10):

@1st testing = the problem appeared 3x in 1 min
(so, no need for more testing on this version)


And on the todays fresh WinUAE 3.4.1 (Public Beta 0, 2017.01.13):

@1st testing = the problem appeared 3x in 5 min
@2nd testing = the problem appeared 1x in 5 min
@3rd testing = the problem appeared 1x in 5 min

Obviously, it behaves now much better than before!
Also, there still exist probability that the problem might happen (but now not too often), what seems close to before mentioned yt-video (where it also happened).
Therefore, it would not be good if is not happening ! (IMHO)

Does the problem appear same frequently on a real hardware, cant tell, didnt test it yet, maybe one day.
(because my A500 is packed in a box, and replaced with WinUAE)

EDIT 2:
Above results refers to test done according to post #60 !
During the gameplay the problem appeared only once in some 15 min and 20 min of playing !

[Toni Wilen]

Some more tweaks done.

[amilo3438]

Originally Posted by Toni Wilen

Some more tweaks done.

In todays WinUAE 3.4.1 (Public Beta 0, 2017.01.14) it works even better !
During the game testing the problem does not appeared, not in my (simple & fast) test nor in the gameplay! (at least I didnt manage to provoke it)

[DrBong]

Originally Posted by DrBong

It's funny how this thread should pop up about this game (courtesy of who else but you Damien! ). I recently bought another version of this game from the U.S. to go with the Euro version I supplied for the WHD patch (good to see that WHD updates haven't broken the slave BTW!). Thanks for the reminder, I must see if it's supported.....

Originally Posted by Asman

Give me sign if your version is not supported then I will add it.

I see that you recently updated the WHD slave for Mission Elevator - nice work Adam! I have at least one version of this game that is unsupported. Are you still interested in adding support for it?

[Asman]

Originally Posted by DrBong

I see that you recently updated the WHD slave for Mission Elevator - nice work Adam! I have at least one version of this game that is unsupported. Are you still interested in adding support for it?

Sure, I need only this unsupported image(s).

[DrBong]

@Adam
Good stuff, I'll see about making some disk images on the weekend and uploading them to the Zone.