search for: matrix_multiply_out

..._decode_float works in the argument list... I noticed it changes it's arguments depending on whether FIXED_POINT is used. I copied this style for the projection API as well. If this isn't desired, we should make those changes separately. 2) See above. 3) I only zero out initially. For the matrix_multiply_out functions, we need to be able to take a single decoded stream and add weighted versions of it to each of the final output channels. This zero-ing out on stream 0 ensures the block is zero-set so we can incrementally add all the decoded streams to the output channels. 4) See 3), short and float ver...

...s a bad idea. For example, if > you look at the public header, opus_projection_decode() has a pcm > argument of type opus_int16*, so it's a bit confusing for the C file to > define pcm as opus_val16*, even if the two map to the same. > > > 3) I only zero out initially. For the matrix_multiply_out functions, we > > need to be able to take a single decoded stream and add weighted > > versions of it to each of the final output channels. This zero-ing out > > on stream 0 ensures the block is zero-set so we can incrementally add > > all the decoded streams to the output c...

...u look at the public header, opus_projection_decode() has a pcm > > argument of type opus_int16*, so it's a bit confusing for the C file > to > > define pcm as opus_val16*, even if the two map to the same. > > > > > 3) I only zero out initially. For the matrix_multiply_out > > functions, we > > > need to be able to take a single decoded stream and add weighted > > > versions of it to each of the final output channels. This zero-ing > out > > > on stream 0 ensures the block is zero-set so we can incrementally > ad...

...e() has a > pcm > > > argument of type opus_int16*, so it's a bit confusing for the > > C file to > > > define pcm as opus_val16*, even if the two map to the same. > > > > > > > 3) I only zero out initially. For the matrix_multiply_out > > > functions, we > > > > need to be able to take a single decoded stream and add > weighted > > > > versions of it to each of the final output channels. This > > zero-ing out > > > > on stream 0 ensures the...

...ut I still think it's a bad idea. For example, if you look at the public header, opus_projection_decode() has a pcm argument of type opus_int16*, so it's a bit confusing for the C file to define pcm as opus_val16*, even if the two map to the same. > 3) I only zero out initially. For the matrix_multiply_out functions, we > need to be able to take a single decoded stream and add weighted > versions of it to each of the final output channels. This zero-ing out > on stream 0 ensures the block is zero-set so we can incrementally add > all the decoded streams to the output channels. OK, I see....

...pus_int16*, so it's a bit confusing > > for the > > > C file to > > > > define pcm as opus_val16*, even if the two map to the > same. > > > > > > > > > 3) I only zero out initially. For the > matrix_multiply_out > > > > functions, we > > > > > need to be able to take a single decoded stream and > > add weighted > > > > > versions of it to each of the final output channels. > This > > > zero-ing out &...

...pus_int16*, so it's a bit confusing > > for the > > > C file to > > > > define pcm as opus_val16*, even if the two map to the > same. > > > > > > > > > 3) I only zero out initially. For the > matrix_multiply_out > > > > functions, we > > > > > need to be able to take a single decoded stream and > > add weighted > > > > > versions of it to each of the final output channels. > This > > > zero-ing out &...

...example, if > you look at the public header, opus_projection_decode() has a pcm > argument of type opus_int16*, so it's a bit confusing for the C file to > define pcm as opus_val16*, even if the two map to the same. > > > 3) I only zero out initially. For the matrix_multiply_out > functions, we > > need to be able to take a single decoded stream and add weighted > > versions of it to each of the final output channels. This zero-ing out > > on stream 0 ensures the block is zero-set so we can incrementally add > > all the decode...

Hello, Attached is a patch to resolve a memory issue using the Projection API when compiling using a psuedo-stack / limited memory. Please let me any ?s you might have. Cheers, Drew -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.xiph.org/pipermail/opus/attachments/20171120/84e36e1b/attachment-0001.html> -------------- next part --------------

...header, opus_projection_decode() has a pcm > >     argument of type opus_int16*, so it's a bit confusing for the > C file to > >     define pcm as opus_val16*, even if the two map to the same. > > > >     > 3) I only zero out initially. For the matrix_multiply_out > >     functions, we > >     > need to be able to take a single decoded stream and add weighted > >     > versions of it to each of the final output channels. This > zero-ing out > >     > on stream 0 ensures the block is zero-set so we can &...

...gt;     >     argument of type opus_int16*, so it's a bit confusing > for the > >     C file to > >     >     define pcm as opus_val16*, even if the two map to the same. > >     > > >     >     > 3) I only zero out initially. For the matrix_multiply_out > >     >     functions, we > >     >     > need to be able to take a single decoded stream and > add weighted > >     >     > versions of it to each of the final output channels. This > >     zero-ing out > >     >     >...

...;     for the > >     >     C file to > >     >     >     define pcm as opus_val16*, even if the two map > to the same. > >     >     > > >     >     >     > 3) I only zero out initially. For the > matrix_multiply_out > >     >     >     functions, we > >     >     >     > need to be able to take a single decoded > stream and > >     add weighted > >     >     >     > versions of it to each of the final output >...

...t confusing >> > for the >> > > C file to >> > > > define pcm as opus_val16*, even if the two map to the >> same. >> > > > >> > > > > 3) I only zero out initially. For the >> matrix_multiply_out >> > > > functions, we >> > > > > need to be able to take a single decoded stream and >> > add weighted >> > > > > versions of it to each of the final output channels. >> This >> >...

...t confusing > >     for the > >     >     C file to > >     >     >     define pcm as opus_val16*, even if the two map to > the same. > >     >     > > >     >     >     > 3) I only zero out initially. For the > matrix_multiply_out > >     >     >     functions, we > >     >     >     > need to be able to take a single decoded stream and > >     add weighted > >     >     >     > versions of it to each of the final output > channels. This > >   ...

...> > C file to > > > > > define pcm as opus_val16*, even if the two map > > to the same. > > > > > > > > > > > 3) I only zero out initially. For the > > matrix_multiply_out > > > > > functions, we > > > > > > need to be able to take a single decoded > > stream and > > > add weighted > > > > > > versions of it to each of the...

...> >         >     >     >     define pcm as opus_val16*, even if the > two map > >         to the same. > >         >     >     > > >         >     >     >     > 3) I only zero out initially. For the > >         matrix_multiply_out > >         >     >     >     functions, we > >         >     >     >     > need to be able to take a single decoded > >         stream and > >         >     add weighted > >         >     >     >     > versions o...

...> > > define pcm as opus_val16*, even if the >> two map >> > to the same. >> > > > > >> > > > > > 3) I only zero out initially. For the >> > matrix_multiply_out >> > > > > functions, we >> > > > > > need to be able to take a single decoded >> > stream and >> > > add weighted >> > > > >...

...>>>     two map > >>>     >         to the same. > >>>     >         >     >     > > >>>     >         >     >     >     > 3) I only zero out initially. > For the > >>>     >         matrix_multiply_out > >>>     >         >     >     >     functions, we > >>>     >         >     >     >     > need to be able to take a > single decoded > >>>     >         stream and > >>>     >         >     ...