Mega Typhoon WHDLoad

[modrobert]

What is the button or key to continue the game in Mega Typhoon at the "Yes / No" prompt where it reads you have 6 credits after "game over" and it times out after 10 seconds?

This WHDLoad slave:
https://www.whdload.de/games/MegaTyphoon.html

I'm using a regular two button joystick (not CD32 joypad), none of the fire buttons works. 'Y', 'N' and 'Enter' keys on the keyboard fail as well, but you can change selection between "Yes" and "No" using left and right on the joystick.

[Superman]

I just tested it myself and it seems to be a bug of some sort. I tried the ADF version and pressing the fire button to continue works just fine with that but not on Whdload version.

[ransom1122]

There's a problem with the WHDLoad version.. Retroplay needs to check it out

WHDLoad version wont work with v19.0 (Black Screen on loading screen) - I Dropped it to WHDLoad V18.8 and it game booted in WHDLoad

Fire Button on Continue Screen Does not work with WHDLoad with version v18.8 or below.

Fire Button on Continue Screen works with Adf

Quick hack fix i copied over all the game files from Mega Typhoon (1996)(Nordlicht PD)[f 100% ross][h TRSI][t +5 CRX].adf into the WHDLoad Version replacing all files and with WHDLoad V18.8 Game seems to boot and also Fire Button works on Continue Screen..

Ill shoot Retroplay a message about this...

[modrobert]

Originally Posted by Superman

I just tested it myself and it seems to be a bug of some sort. I tried the ADF version and pressing the fire button to continue works just fine with that but not on Whdload version.

Originally Posted by ransom1122

There's a problem with the WHDLoad version.. Retroplay needs to check it out

WHDLoad version wont work with v19.0 (Black Screen on loading screen) - I Dropped it to WHDLoad V18.8 and it game booted in WHDLoad

Fire Button on Continue Screen Does not work with WHDLoad with version v18.8 or below.

Fire Button on Continue Screen works with Adf

Quick hack fix i copied over all the game files from Mega Typhoon (1996)(Nordlicht PD)[f 100% ross][h TRSI][t +5 CRX].adf into the WHDLoad Version replacing all files and with WHDLoad V18.8 Game seems to boot and also Fire Button works on Continue Screen..

Ill shoot Retroplay a message about this...

Aha, so it was a bug, was thinking it was some special key or button combo. Thanks for checking. I will try the "100% ross" version.

EDIT:

I tried replacing the files with the "100% ross" version running WHDLoad V18.9 and fire button now works to continue. Thanks again.

However, after booting my A1200 from the floppy version, besides the nice cracktros (and centerfold), I noticed the game runs smoother without vertical graphics jitter compared to WHDLoad, so sticking with the floppy for now. EDIT: The jitter still exists in this version as well, it's just random.

EDIT2:

After testing the WHDLoad slave v1.3 a bit more with the "100% ross" game files, the vertical graphics jitter sometimes goes away if the game is exited with 'Esc' key and a new game is started ("insert coin"), perhaps something wrong with the graphics initialization.

[modrobert]

The high frequency vertical jitter, roughly one in pixel height, making the surface scroll look a bit fuzzy, can be triggered on all versions; "WHDLoad supported game data", "100% ross booted from floppy" and "100% ross as WHDLoad data" on my A1200 (see sig for details). I found an easier way to fix and trigger it, use 'P' on keyboard for pause during game play, then unpause, you will see the jitter disappear (smooth vertical scroll) and come back sporadically (jittered vertical scroll), once you leave it in certain state (without pausing again) it will continue that way. When seen the jitter is in the slow automatic vertical surface scroll, not about moving the ship around using controls causing fast scroll.

Also, sometimes after "game over" when reaching the "continue?" screen there are big yellow vertical bars over the graphics (sprites perhaps? not sure), menu operation still works but looks corrupted until new game is started. This also happened in all three versions mentioned previously.

I think these could be original game bugs (common with Amiga games), or perhaps related to the A1200 chipset (AGA, assuming this game was designed for ECS/OCS), but mentioning anyway.

[modrobert]

I modified the WHDLoad slave source code and compiled (using Barfly) a fixed v1.3 where the "CONTINUE?" fire button works.

Basically I make use of the existing _bsr_controller_type() call in 'ReadJoyPad.s' to establish if we got a joystick or not, and then load separate WHDLoad resload_Patch() list depending of which type of joystick is detected.

cmp.b   #0,(controller_joypad_1) ; this is true for joystick, not joypad (CD32)

I would be happy if someone can test if using a joypad still work with this slave because I don't have a CD32 joypad to test with.

This slave expects the WHDLoad source version of the game where the selectable trainers work (not the 100% Ross version).

JOTD,

If you happen to stumble across this, then feel free to make a new official release of the slave, then maybe the supplied source code will help (or not, up to you).

The attached archive includes slave binary and source code.

[DanyPPC]

Tested on real Amiga 1200/030 OS 3.2 and joypad KTRL CD32, but no success.

The buttons don't work on Continue screen.

[modrobert]

Originally Posted by DanyPPC

Tested on real Amiga 1200/030 OS 3.2 and joypad KTRL CD32, but no success.

The buttons don't work on Continue screen.

Aha, OK, so the fire button didn't work with joypad in original v1.3 slave either I suppose, because haven't made any changes to that part. I was assuming it only fails when using regular two button joystick which is fixed now.

Alright, will see if I can fix the joypad problem as well.

[ransom1122]

WHDLoad V18.8

v1.3 + mega_typhoon_fix.lha (Slave File only)
CD32 PAD and normal pad works for Continue

WHDLoad V19.0
v1.3 + mega_typhoon_fix.lha (Slave File only)
CD32 PAD Does not work and normal Pad Works for Continue

[modrobert]

Originally Posted by ransom1122

WHDLoad V18.8

v1.3 + mega_typhoon_fix.lha (Slave File only)
CD32 PAD and normal pad works for Continue

WHDLoad V19.0
v1.3 + mega_typhoon_fix.lha (Slave File only)
CD32 PAD Does not work and normal Pad Works for Continue

Interesting, so it worked for you then.

I use WHDLoad v18.9 dev version which is the latest stable at: https://www.whdload.de/

The development support files (e.g. includes and kick13.s) are also from v18.9 dev package, do you think I should use v19.0 instead?

I don't know why WHDLoad V19.0 fails, but perhaps it doesn't make sense to support that just yet since it is Beta.


@DanyPPC

Which version of WHDLoad version did you use? Also, did you replace the MegaTyphoon.slave file with the one attached in this thread (mega_typhoon_fix.lha)? As mentioned previously, this slave expects the WHDLoad source version of the game where the selectable trainers work (not the 100% Ross version).

[ransom1122]

I used winuae with xb2 controller confined as cd32 pad, and yes results differed with different whdload version

[DanyPPC]

Originally Posted by modrobert

@DanyPPC

Which version of WHDLoad version did you use? Also, did you replace the MegaTyphoon.slave file with the one attached in this thread (mega_typhoon_fix.lha)? As mentioned previously, this slave expects the WHDLoad source version of the game where the selectable trainers work (not the 100% Ross version).

WHDLoad official 18.9
Replaced megatyphoon.slave with the one fixed in your archive
No, I don't use the 100% Ross version. Absolutely.

Thanks for your work.

[modrobert]

Originally Posted by DanyPPC

WHDLoad official 18.9
Replaced megatyphoon.slave with the one fixed in your archive
No, I don't use the 100% Ross version. Absolutely.

Thanks for your work.

OK, just so I get this right, does your joypad KTRL CD32 work everywhere except at the "CONTINUE?" in the game with the fixed slave?

I have a theory that the "CONTINUE?" screen needs it's own "vbl_hook patch" because the joypad data doesn't seem to update there (also, my HRTmon can't be hotkey triggered/entered at that screen which makes me think it's something related to interrupts).

[DanyPPC]

KTRL CD32 pad works good with MegaTyphoon, directions + fire button 1 and fire button 2 for the second arm.

Issues only for buttons on Continue screen, directions work for select YES or NO.

[modrobert]

Originally Posted by DanyPPC

KTRL CD32 pad works good with MegaTyphoon, directions + fire button 1 and fire button 2 for the second arm.

Issues only for buttons on Continue screen, directions work for select YES or NO.

Thanks, I'll try to fix that.

[jotd]

whdload 19 has direct CD32 joypad read API. I haven't used it yet, but I'm preparing a big migration for dozens of slaves to use the new API and not some copy/paste code.

the API takes timing into account: it's not going to poll the joypad too often.

[modrobert]

Originally Posted by jotd

whdload 19 has direct CD32 joypad read API. I haven't used it yet, but I'm preparing a big migration for dozens of slaves to use the new API and not some copy/paste code.

the API takes timing into account: it's not going to poll the joypad too often.

That sounds good, what interrupt is WHDLoad 19 depending on for the CD32 joypad protocol?


I think the problem with the "CONTINUE ?" screen is this INTREQ vbl loop:

$009A1412 MOVE.W  #$20,($DFF09C)
$009A141A BTST    #5,($DFF01F)
$009A1422 BEQ.B   $9A141A
$009A1424 MOVE.W  #$20,($DFF09C)
$009A142C RTS

My theory is that this screws up the vblank hook which is supposed to update the joypad buttons state in a timely manner, it could be some other interrupt problem specific to the "CONTINUE ?" screen as well, not sure.

During the "CONTINUE ?" screen I noticed that the "read_buttons" sub routine in the slave is called, so I added this as the first line of "read_buttons":

bsr _read_joysticks_buttons

This way it always updates buttons regardless of vblank interrupts, hopefully it works, but since I don't have a CD32 joypad, can't test it. For some reason this interrupt problem doesn't seem to occur during emulation since it worked for ransom1122, so can't really use an emulator to test either.

I have updated the archive (mega_typhoon_fix2.lha) in the previous post in this thread.

Please test and let me know if it works.

[DanyPPC]

This time works !

[ransom1122]

mega_typhoon_fix2.lha - feedback

WinUAE - Quickstart A1200 - Cycle Exact Enabled (Default)

WHDLoad V18.8
v1.3 + (Slave File only)
Cycle Exact Enabled (Default) - Black Screen on Boot
Cycle Exact Disabled - Boots
WHDLoad Trainers work
CD32 PAD and normal pad works for Continue

WHDLoad V19.0
v1.3 + (Slave File only)
Cycle Exact Enabled (Default) - Black Screen on Boot
Cycle Exact Disabled - Boots
WHDLoad Trainers work
CD32 PAD and normal pad works for Continue

Looks like results are exact now with different versions of whdload.

Only problem is it only runs when cycle exact is disabled?

[modrobert]

Originally Posted by DanyPPC

This time works !

Great!

Originally Posted by ransom1122

mega_typhoon_fix2.lha - feedback

WinUAE - Quickstart A1200 - Cycle Exact Enabled (Default)

WHDLoad V18.8
v1.3 + (Slave File only)
Cycle Exact Enabled (Default) - Black Screen on Boot
Cycle Exact Disabled - Boots
WHDLoad Trainers work
CD32 PAD and normal pad works for Continue

WHDLoad V19.0
v1.3 + (Slave File only)
Cycle Exact Enabled (Default) - Black Screen on Boot
Cycle Exact Disabled - Boots
WHDLoad Trainers work
CD32 PAD and normal pad works for Continue

Looks like results are exact now with different versions of whdload.

Only problem is it only runs when cycle exact is disabled?

There is a black screen when I launch the game on my real A1200 for roughly 10 seconds before the game starts, perhaps that part is fast when "cycle exact" is disabled in WinUAE (even faster with JIT mode enabled), so try waiting a bit longer with "cycle exact" enabled.

Also found this...

Originally Posted by Toni Wilen

As was already said, only 68000 (and 68010) is cycle-accurately emulated. Later models can't be emulated cycle-accurately because internal instruction sequencer timing is not known.

For most whdload games approximate speed or memory cycle-exact is fine (memory cycle exact = CPU is internally fastest possible but any "slow" access = chip ram/custom registers/CIA accesses are slowed down like on real hardware). Usually even fastest possible work.

btw, don't attempt to to use CPU speed slider to "fix" specific games. Its CPU speed change effect can and will change between versions.