A.D.A. Amiga Demoscene Archive

  Welcome guest! Please register a new account or log in

  

  

  

log in with SceneID

  

Demos Amiga Demoscene Archive Forum / Coding / Cinter - a 4k synth for all Amigas
 Page:  ««  1  2  
Author Message
h0ffman
Member
#1 - Posted: 26 Feb 2020 11:59
Reply Quote
not sure about your startup / system take over code here, but if you're running a 060 based config there is a good chance the VBR is being moved. If the music runs on an interrupt and doesn't know the VBR address, then it might not be triggering. A quick way to test that is to poke $dff180 with different colours at the start and end of the inerrupt code. when running if the background colour doesnt change then the interrupt probably isn't running.

another thing to check is that Audio DMA is enabled, although I have a feeling the replay may switch that on.
Jazzcat
Member
#2 - Posted: 28 Feb 2020 16:58
Reply Quote
flynn_nrg:
I'm using the CurtCool's automatic song, converted the python script and the output of the converter looked all ok.

Do you have any ideas of what could be causing this?


Unfortunately I can't help you on that, but one thing I know for sure is that Automatic is not Curt Cool's song :-)
Blueberry
Member
#3 - Posted: 2 Mar 2020 21:22
Reply Quote
Hmm, that's weird. If you have assembled the example as is, using the supplied startup code (with COMPATIBILITY set to 1), then a moved VBR should not cause any problems.

Are you using the manual DMA feature?

Which assembler are you using, and how are you running the result?
flynn_nrg
Member
#4 - Posted: 29 Mar 2020 17:25
Reply Quote
Sorry for the late reply.

I was including the files in the wrong place, so after swapping the .cinter4 files and the .raw one everything is working fine now.

Thanks for everything. Awesome synth!
deetsay
Member
#5 - Posted: 7 May 2020 12:12
Reply Quote
Hi Blueberry!

And thanks for this powerful music engine!

A while ago I made a "JIT-compiler" / "self-modifying code" version of the Cinter sample generator (forked the original in github).

https://github.com/deetsay/Cinter/blob/master/player/Cinter4JIT.S

This dropped the calculation time for Curt Cool's Crashing Bitplanes soundtrack from 45 seconds to 25 seconds on the A500. Or something like that. Having the different sample generators behind separate calls also allowed to start the tune before all samples were ready, making it effectively start after only about 15 seconds of calculation.

Speed gains are highly dependent on how exotic numbers are used in the song's parameters :-). But it seems that 0 and 1 are very common, making all the MULUs in the default calculator somewhat ineffective....

It's not pretty and not well commented or documented at all. Almost just a raw version of what was needed to get the intro going nicely.

Please let me know:

a) if there's any specific advice regarding comments and documentation that you'd need me to pay attention to before you'd consider bringing something like this into the main branch, or

b) if - for whatever reason - you're not interested in including this in the main branch. In that case I probably won't be bothered to improve it much, but maybe I might.
Blueberry
Member
#6 - Posted: 10 May 2020 16:51
Reply Quote
Wow, that's really cool. And very useful for 64k intros where space is plenty.

I am definitely interested in including this into the main Cinter codebase. It could be controlled by a define, similarly to the CINTER_MANUAL_DMA feature.

I will take a closer look at the code and provide some more comments soon...
 Page:  ««  1  2  

  Please register a new account or log in to comment

  

  

  

 

A.D.A. Amiga Demoscene Archive, Version 3.0