Two questions regarding DLIs...

[Steril707]

Have some ideas, but not sure yet if stuff is feasible, so here we go:

 

* how big can a single DLI be? I think I remember something about page boundaries making problems.

* how many commands changing stuff can I get into a line before my display starts behaving strange in let's say Antic 4 and 5 mode?

 

Thanks for any answers in Advance! 

[TGB1718]

Have a look at DLI timing in De-Re-Atari it gives some quite detailed timing information.

 

With regards to how much you can do, if changing colours, then a very limited time before you see screen

corruption, however if your moving a "player" it can be done any time during a DLI.

 

Sometimes you have to experiment and find what's good for you.

[Rybags]

Page boundaries irrelevant really other than added cycle for such things as indexed reads or branches crossing a page boundary.

Size - isn't really an issue, it's more down to time.

Time - a DLI in theory could take the entire screen time if you wanted though nesting situations could cause trouble.

 

But critical stuff like colour and player positions - you have limitations on how much you can do per scanline then you have screen DMA and refresh steals that take more cycles away.

Player positioning does matter - if you move a player too soon it might get omitted.

But yes, experiment and use things like Altirra debug single-step.

 

An important thing for DLIs is to only do what's needed - calculations and stuff should be by main loop or VBlank, DLI is just for stuffing hardware registers for the most part.

Edited January 13, 2023 by Rybags

[TGB1718]

1 hour ago, Rybags said:

An important thing for DLIs is to only do what's needed - calculations and stuff should be by main loop or VBlank, DLI is just for stuffing hardware registers for the most part.

I should have added that, I did all the calculations for screen positions in the main code and the DLI just used the lookup table