Move Any Mountain and B2 (demos)

[tomcat666]

Hi!

I am having problem running a demo called "Move Any Mountain" by "Lazy Bones". The best I could get it is a "flashing screen" running through the memory where the effects should be.
Using basic 1.3 config with 512 or 1Mb chip/512 slow memory (all other configs seem to give GURU). Ofcourse all cycle exact, etc. options turned ON. A500 emulation as close as possible.

Here is the adf if any1 wants to have a look.

http://pc.sux.org/tomcat/MoveAnyMountain.zip

Any info on how to get this going would be appretitated.

The authos said it might need 1Mb chip and he is using a "special" HAM5 mode which doesn't seem to be emulated properly, so this might be the cause ? There is also a demo called "B2" which should have a roto-globus filled, but it doesn't for the same reason... here is the link:

http://pc.sux.org/tomcat/0947-b2.dms

Also there is a HAM 4x4 zoomer which doesn't work properly aparently Even though I never could even get to that part.

All-in-all some strange effects which might need looking into for a proper emulation. I hope it is helpfull also for Toni and other WinUAE people (are there any other ? ).

Best regards,
Tomaz

[Toni Wilen]

Confirmed working on real Amiga? (I am not too interested with random semi-broken disks floating around)

[tomcat666]

Well, the author confirmed they worked. Unfortunetly all I have here is A1200 with broken floppy so I cannot test it. But they *should* work on real HW, but you will have to take the author's word, not mine...

EDIT: A friend of mine also confirmed they work on his real A500 KS 1.3 512chip/512slow...

[Toni Wilen]

"Move Any Mountain" confirmed working.

This is interesting case, it requires undocumented (and not yet emulated) feature, 1M Agnus with 0.5 chip+0.5 slow aliases upper 1M chip ram area to slow ram area. (for example copper pointer set to 0x090000 sees memory at 0xC10000. Note that CPU only sees "bus noise".)

In other words, you can have some kind of 1M chip ram even with 0.5M+0.5M configuration as long as Agnus is ECS model.

This demo requires 1M Agnus and 0.5M chip + 0.5M+ slow. This is the only working configuration

[Little_Sharky]

This is good news,it means it would be implemented in the future builds and then more compatibility problems with amiga software will became less.Nice!

[tomcat666]

NICE ! I am glad you got it working and that there is a new feature that might help with other programs.

Thank you Toni for such a quick "fix" and reply !

[Toni Wilen]

[B2]

HAM6 with 4 or 5 planes implemented, fixes globe effect.

[Move Any Mountain]

This must be a record. Program that requires ECS Agnus but DOES NOT work with 1MB Chip RAM..

[tomcat666]

Originally Posted by Toni Wilen

[B2]

HAM6 with 4 or 5 planes implemented, fixes globe effect.

[Move Any Mountain]

This must be a record. Program that requires ECS Agnus but DOES NOT work with 1MB Chip RAM..

Thank you Toni for the fixes. Looking forward to the next beta to try them out

[Toni Wilen]

"B2 pre-emptive bugs still not fixed" report:

- flashing in dot/glenz vector part (before globe) if cycle-exact enabled (blitter/copper timing problem again most likely)
- some horizontal lines in zoomer (after globe) are not aligned, copper timing error. Really rare problem. May not be fixed until 1.5 is out.

[andreas]

Weird!
Usually those artifacts were gone once CE was *EN*abled.

[hitchhikr]

Sorry for reviving this old thread but it looks like you didn't implement that "hidden" memory mirroring correctly, i need to reset with the fake fast set to 1meg before winuae acknowledges the trick (setting it to 512k afterwards keeps it turned on so it work correctly after that manipulation).

[Toni Wilen]

Originally Posted by hitchhikr

Sorry for reviving this old thread but it looks like you didn't implement that "hidden" memory mirroring correctly, i need to reset with the fake fast set to 1meg before winuae acknowledges the trick (setting it to 512k afterwards keeps it turned on so it work correctly after that manipulation).

It is working as designed. It is only enabled/disabled when memory size (chip or slow) changes to prevent any bad side-effects when reseting emulation normally. I didn't want any bigger changes that can too easily break other things, this feature is extremely rarely needed after all

[hitchhikr]

That's not a very good example of design i'm afraid , if it breaks any compatibility (which i doubt as ther's really few program which directly poke at 0xcxxxxx) any broken piece of software would be also broken on the real machine as well so there's no need to try to be better than the real deal, but there's another reason why i'd like to see this implemented correctly...

Recently there's been a new trend about releasing so called "a500" demos requiring 1meg of chip mem for chunky effects conversions (therefore not working on a500 at all) so if winuae would support that, it would permit to use such trick to make them more compatible with a standard a500 512k/512k configuration (i'm sure more of such demos will be released soon ).

At worst put a checkbox under the fake fast allowing it to be used as fake chip if the ECS Agnus (afair a v1.3 + 512k chip/512k fast was the most common a500 configuration back then) is selected or something like that, so the user can use it when instructed to do so.

Also where is that >512k fake fast coming from ? Which machine had such memory configuration ? (and which software took "advantage" of this ?)

And talking about breaking compatibility:

In the Cryptoburners megademo 1 there's something like this:

move.w $dff012, d0
wait_right_mb:
cmp.w $dff012, d0
bne.b wait_right_mb

They used this to check the right mouse button, it's not the most clever piece of code but it works on amiga machines and not anymore in WinUAE v1.5.2 (to make it work i had to use $dff016 instead).

Cheers,
f.

[Toni Wilen]

Originally Posted by hitchhikr

That's not a very good example of design i'm afraid , if it breaks any compatibility (which i doubt as ther's really few program which directly poke at 0xcxxxxx) any broken piece of software would be also broken on the real machine as well so there's no need to try to be better than the real deal, but there's another reason why i'd like to see this implemented correctly...

It is how I implement undocumented things. The more it is used, the more interested I am supporting it. Only one demo = not interested enough (and this time this is imho really stupid thing to do..) Same with Disposable Hero title screen, didn't really bother with until more programs doing the same thing was found.

(note, new programs doing that won't count! )

Also where is that >512k fake fast coming from ? Which machine had such memory configuration ? (and which software took "advantage" of this ?)

Many third party trap door expansions allowed 1M, 1.5M or even 1.8M (max) configuration.

move.w $dff012, d0
wait_right_mb:
cmp.w $dff012, d0
bne.b wait_right_mb

They used this to check the right mouse button, it's not the most clever piece of code but it works on amiga machines and not anymore in WinUAE v1.5.2 (to make it work i had to use $dff016 instead).

Ok, thats a bug but note that this may not work with many third party mice. (some don't have pullup resistor in button lines)