Talk:Thmp3
From ProjectWiki
(Difference between revisions)
Line 1: | Line 1: | ||
+ | ==will this work with r24== | ||
will this work with r24? - Echo | will this work with r24? - Echo | ||
Line 5: | Line 6: | ||
Have you tried it with r24? -Eris | Have you tried it with r24? -Eris | ||
− | notes on helix updates | + | ==notes on helix updates== |
* if remove lrlr interleave can write directly to the pcm output buffer, but at end of song can a frame be less than 1152 samples?? | * if remove lrlr interleave can write directly to the pcm output buffer, but at end of song can a frame be less than 1152 samples?? | ||
* maybe add round robin support to helix directly?? (would need to pass cpos and bsize to helix somehow) | * maybe add round robin support to helix directly?? (would need to pass cpos and bsize to helix somehow) | ||
* find registers to use by placing dummy code? (best to find proper c/arm register usage and calling convention docs) | * find registers to use by placing dummy code? (best to find proper c/arm register usage and calling convention docs) | ||
* on current deinterleave function might be more efficient to calculate if/where it overflows first and use 2 loops if necessary (would save from the 'if pos >' code each frame) | * on current deinterleave function might be more efficient to calculate if/where it overflows first and use 2 loops if necessary (would save from the 'if pos >' code each frame) |
Revision as of 14:36, 2 February 2009
will this work with r24
will this work with r24? - Echo
Not tried it with > r23 of devkitarm. You may need to compile the arm9 side with -Os option to shrink code as much as possible.
Have you tried it with r24? -Eris
notes on helix updates
- if remove lrlr interleave can write directly to the pcm output buffer, but at end of song can a frame be less than 1152 samples??
- maybe add round robin support to helix directly?? (would need to pass cpos and bsize to helix somehow)
- find registers to use by placing dummy code? (best to find proper c/arm register usage and calling convention docs)
- on current deinterleave function might be more efficient to calculate if/where it overflows first and use 2 loops if necessary (would save from the 'if pos >' code each frame)