Compact assembler loop for loading data: is address -1 ok?

[lucienEn]

Hi,

 

I noticed a lot of sample code (including generated code from 2600 builder) ignores the first byte of the data since they use bne to exit the loop. Or instead they wait for 0 in the data block but that's unpredictable length and wastes 1 byte. I found I could just use the start address -1. It works fine so far but any downsides with this below (or use bmi but that's limited to 127 height):

 

    ldx #10
Kernel1_Loop
    sta WSYNC
    sta HMOVE
    lda colubk_data-1, x      ; this will load data in range colubk_data .. colubk_data+9
    sta COLUBK
    dex
    bne Kernel1_Loop

 

Lucien

Edited January 8 by lucienEn

[Thomas Jentzsch]

-1 is fine

[lucienEn]

Great, thanks!

[+splendidnut]

Since this is being used in a display kernel, don't forget to watch out for page-crossing issues (extra cycle penalty throwing off timing).

[lucienEn]

19 hours ago, splendidnut said:

Since this is being used in a display kernel, don't forget to watch out for page-crossing issues (extra cycle penalty throwing off timing).

I wonder how this syntax works looking at 2600 builder:

 

   if >. != >[.+(playfield_lines)]
      align 256
   endif
 

So maybe the '>' looks at the high byte and compares that to high byte of the playfield lines? What is the [.+?...] meaning?

[+splendidnut]

Here are the meanings behind the symbols:

'>' says look at the high byte

'.' is the current address in the assembly file

'[ ]' work just like parenthesis... they give an alternative due to 6502 assembly typically using ( ) to imply indirect addressing mode when used in an instruction.

'[.+(playfield_lines)]'  is current address '.' + 'playfield_lines' label address or EQU value 

 

I believe I got everything correct, please let me know if I messed up anything.

 

Looking at the example code, it's not entirely clear what playfield_lines refers to.  If I had to take a guess, 'playfield_lines' refers to the number of lines in (size of) the data table that immediately follows the if...endif construct.

 

If that is true:

   - The example code would make sure the table is within a page... and if it crosses a page boundary, it would align it to the next one (force it to start at the beginning of the next one).

   - You would still need to handle the case of -1... easiest thing would be to put a byte of padding right before the data table (after the if..endif construct)

[lucienEn]

Thanks! Yes that code comes from the 2600 builder web site. I want to use that construct as well for my auto-code generation.

 

Downside is that it might waste a lot of bytes but I think auto-code could adjust for that and group them better.

Edited January 10 by lucienEn