PA_ prefixed functions seem to be from palibds, which is a Nintendo DS homebrew development library.
DS being a constrained device it is logical that whoever wrote this code packed multiple byte-sized info into a bigger 16-bit short and using bit shifting to extract a specific byte
You’re right, this is some homebrew I wrote for my DS when I was in highschool.
But it was nothing as clever or logical as that! I think I had read somewhere that floats were really expensive on the DS ARM9 CPU, as was division (?). So to allow fractional values for object positions, etc, you would store them in an integer at 256x their true value, then shift it by 8 to the right to ‘divide’ by 256 and get the proper value back.
Not sure whether this was actually true, but either way this would’ve been completely cancelled out by calling that ‘lowestXInObj’ function 20 times per frame - it loops through every object in the scene!
Context please, what is this bit shifting shenanigans that im witnessing right now?
PA_prefixed functions seem to be from palibds, which is a Nintendo DS homebrew development library.DS being a constrained device it is logical that whoever wrote this code packed multiple byte-sized info into a bigger 16-bit short and using bit shifting to extract a specific byte
You’re right, this is some homebrew I wrote for my DS when I was in highschool.
But it was nothing as clever or logical as that! I think I had read somewhere that floats were really expensive on the DS ARM9 CPU, as was division (?). So to allow fractional values for object positions, etc, you would store them in an integer at 256x their true value, then shift it by 8 to the right to ‘divide’ by 256 and get the proper value back.
Not sure whether this was actually true, but either way this would’ve been completely cancelled out by calling that ‘lowestXInObj’ function 20 times per frame - it loops through every object in the scene!
Should’ve used fixed point integers!