why porting ? kinda waste of time and just have fun only

[powerpan]

but don't know why rockbox codec always missing LSBs
when our player plays DSD, it use 32bits output for DSD64/DSD128 playback

Which codec are you referring to?  Most of our decoders use 32 bit output and then truncate to 16 bit in the driver for efficiency reasons.

most lossless codec like flac/ape/wavpack, including wav
and WAV's id3v2 doesn't support unicode

You mean they don't decode or they decode with lower precision?

DSP_SET_SAMPLE_DEPTH in codec, like flac will return 29bits only. where is the 3 bits? who knows...

anyway, hardware is like a piece of cake. depends how you think of rockbox, let it be or make it a "finished" product.

[saratoga]

Ah yes, that makes sense.  These are 32 bit processors.  Returning 32 bits without rounding would require 64 bit operations, which are extremely expensive on portable devices.  And there is little point since you cannot make actual audio with > 24 bits of precision anyway. 

[powerpan]

Ah yes, that makes sense.  These are 32 bit processors.  Returning 32 bits without rounding would require 64 bit operations, which are extremely expensive on portable devices.  And there is little point since you cannot make actual audio with > 24 bits of precision anyway.

no need to round in fact, why for wav? flac doesn't need rounding....

[saratoga]

The WAV decoder doesn't round as far as I know, but the DSP engine will have to since it operates at 32 bit precision.

[powerpan]

but when we chase the bit depth, it does output 29bits instead of 32bits

[gevaerts]

This thread is much more of a waste of time than any port (successful or otherwise) has ever been. Please stop it

[saratoga]

but when we chase the bit depth, it does output 29bits instead of 32bits

Iirc the DSP engine expects everything to have 3 bits of headroom so that there is no clipping during processing.  See PCM_OUTPUT_DEPTH.

Why does this matter to you?