Can AmiMasterGear, etc run on A4000?

[Amigaz]

adolescent wrote:
@AMIGAZ

Must be some incompatibility in your system.  Download the source code and start debugging.

"debugging"? me? hey..I'm just a simple marine motor mechanic  :-P

It might be my custom kickrom file that causes it..what do I know..bah  :-?

[Amigaz]

keropi wrote:

by adolescent on 2007/3/24 17:58:36    @AMIGAZ    Must be some incompatibility in your system. Download the source code and start debugging.

and while at it, compile opera web browser for OS3,9!

Understood! I'm the right man for the job.....NOT!!!  :-D

[therail]

Right. I finally got a chance to investigate this thing. i followed adolescent settings under AmiMasterGear, and they indeed make a difference! Running at around 70fps - although unfortunately, not consistantly around that mark. Along with adolescent AmiMasterGear settings, i also done under the settings Options/Video Emulation:

Sprite Priorities=on
Background Priorities=on
Realtime Scroll Refresh=on
Realtime Palette Refresh=on

The above also slows the emulation down a tiny bit, but again not consistantly. On a sidenote, changing the Screenmode to PAL made it go too fast again (around 250fps). Next i changed my CSMK3 settings to see if that helps. Here are the settings i done:


CSMK3 Settings:

RAM: 70ns RAM
SYSTEM: NoMapRom - SuperScaler=off - BranchCache=off - Writebuffer=off

Tried AmiMastergear again and more of the same! it was a bit more consistant this time. But still not fully consistant enough for my liking.

So i ask whether anyone else can get any closer than that. also has anyone with an a4000 setup got it dead on 50/60fps?

[adolescent]

Mine seems to jump around at the beginning, but once in the game it's pretty consistent.  In Sonic the Hedgehog (SMS) it hangs out right at 70-71.

[therail]

adolescent wrote:
Mine seems to jump around at the beginning, but once in the game it's pretty consistent.  In Sonic the Hedgehog (SMS) it hangs out right at 70-71.

yeah it mostly is like that for me. trying them csmk3 helps a little too find. i suppose if can live with that, but i may try a couple of other stuff before i am through (for keropis sake more than mine! ). i suppose it wouldn't make a difference running a bit of patch software that would slow the system down - considering this emu "bangs" the hardware. did you try amimsx? what was that like?

[adolescent]

You could try degrading the system a bit before running the emulator.  Turning off CPU caches would definately make it slower.

AmiMSX seems to run only a bit fast without any changes (just setting the frame from 2 to 1).  There is no FPS counter so it's hard to say if it is running too fast.

[keropi]

are the sources available? perhaps we can ask a software guru to look at them and maybe implement some better timing?

[therail]

yeah keropi that is a good idea but who could do it?! i think these emus are written in an easy programming language so it shouldn't be too hard for an experienced amigan. the sources are on aminet.

i tried turning off cpu cache but it makes the emu too slow. if i put framskip on auto it runs at almost full speed but then is choppy.

[adolescent]

therail wrote:
Sprite Priorities=on
Background Priorities=on
Realtime Scroll Refresh=on
Realtime Palette Refresh=on

I didn't notice any speed decrease with setting this, but it does improve the accuracy.  Especially the sprite and background priorities.

BTW, turning off caches makes it too slow.  :lol:  Edit: I guess you already said that.

[adolescent]

Source is available, but IIRC not the last updated version.

therail wrote:
yeah keropi that is a good idea but who could do it?! i think these emus are written in an easy programming language so it shouldn't be too hard for an experienced amigan. the sources are on aminet.

I don't think assembler qualifies as "easy".   :-)

[Piru]

Quickly checking the source, there seems to be a bug when acking interrupts. intreq should be written twice, or the interrupt can be executed twice. With vertical blanking interrupt that'd mean double speed.

Code such as this:

Code: [Select]


   MOVE.W #$xxxx,$DFF09C
   NOP
   RTE

...should be:

Code: [Select]


  MOVE.W #$xxxx,$DFF09C
  MOVE.W #$xxxx,$DFF09C
  NOP
  RTE

No, I am not interested in fixing and building the app (the source is in m68k assembly, with spanish comments). Just something I spotted.

[therail]

BUMP!

[Cosmos Amiga]

Quickly checking the source, there seems to be a bug when acking interrupts. intreq should be written twice, or the interrupt can be executed twice. With vertical blanking interrupt that'd mean double speed.

Code such as this:

Code: [Select]

   MOVE.W #$xxxx,$DFF09C
   NOP
   RTE
...should be:

Code: [Select]

  MOVE.W #$xxxx,$DFF09C
  MOVE.W #$xxxx,$DFF09C
  NOP
  RTE


I just updated the source : the emulation is now slower on the 040+...

==> https://leblogdecosmos.blogspot.com/2024/06/amimastergear-091.html

It's based on the 0.81 version.

I'm looking for the source of the 0.82 by Gaelan Griffin to do the same tiny upgrades...

[jvmartins]

Yes, AmiEmulators like AmiMasterGear can generally run on an Amiga 4000 at normal speed if configured properly. It largely depends on your specific setup, such as the CPU and RAM configuration of your A4000. Some tweaking might be needed to optimize performance, but it should be feasible without running too fast or too slow if everything is set up correctly.

[Xanxi]

No the tweaking of the emulator parameters discussed about 14 years ago wasn't satisfactory, with a lot of games not working or still working too fast.

Thanks Cosmos for having updated the code with Piru's modification. It seems to work perfectly on my A4000 with any game i can try now, just like on the A1200.
The solution was so simple, until someone with assembly knowlegde spent some time to do it.

[Cosmos Amiga]

The solution was so simple, until someone with assembly knowlegde spent some time to do it.

The 0.82 source reconstruction is possible from his 0.82 .exe but lot of work because many updates compared to the previous 0.81...

If someone know Gaelan Griffin : could you ask him for the 0.82 source ?