kode54
/
cog
2
1
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Align all use of Accelerate vDSP functions 

vDSP functions expect their input and output pointers to be aligned to
an even four values. Correct this by aligning all pointers. The
allocated buffers used for one parameter should already be aligned
somewhat, but align the incremented positions used on some of them so
that the vDSP functions don't misbehave. Also align the volume scaler
input by doing scalar math until the pointer is aligned prior to calling
vDSP_vsmul. Also, change 16-bit and 32-bit scale to use vsdiv instead of
vsmul with a really small number already divided into one.

Fixes the test vectors that were sent in extrapolating incorrectly due
to their final blocks having uneven sample counts, resulting in
unaligned pointers.

Signed-off-by: Christopher Snowhill <kode54@gmail.com>
CQTexperiment
Christopher Snowhill 2022-02-15 22:53:09 -08:00
parent 627aeda8b1
commit 4906c38827
1 changed files with 34 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

50
Audio/Chain/ConverterNode.m
View File

@ -78,6 +78,16 @@ void PrintStreamDesc(AudioStreamBasicDescription *inDesc) {
void scale_by_volume(float *buffer, size_t count, float volume) {
if(volume != 1.0) {
size_t unaligned = (uintptr_t)buffer & 15;
if(unaligned) {
size_t count3 = unaligned >> 2;
while(count3 > 0) {
*buffer++ *= volume;
count3--;
count--;
}
}
vDSP_vsmul(buffer, 1, &volume, buffer, 1, count);
}
}
@ -564,18 +574,20 @@ tryagain:
float gain = 1.0;
if(bitsPerSample == 1) {
samplesRead = bytesReadFromInput / inputFormat.mBytesPerPacket;
convert_dsd_to_f32(inputBuffer + bytesReadFromInput, inputBuffer, samplesRead, inputFormat.mChannelsPerFrame, dsd2pcm);
memmove(inputBuffer, inputBuffer + bytesReadFromInput, samplesRead * inputFormat.mChannelsPerFrame * sizeof(float));
size_t buffer_adder = (bytesReadFromInput + 15) & ~15;
convert_dsd_to_f32(inputBuffer + buffer_adder, inputBuffer, samplesRead, inputFormat.mChannelsPerFrame, dsd2pcm);
memmove(inputBuffer, inputBuffer + buffer_adder, samplesRead * inputFormat.mChannelsPerFrame * sizeof(float));
bitsPerSample = 32;
bytesReadFromInput = samplesRead * inputFormat.mChannelsPerFrame * sizeof(float);
isFloat = YES;
} else if(bitsPerSample <= 8) {
samplesRead = bytesReadFromInput;
size_t buffer_adder = (bytesReadFromInput + 1) & ~1;
if(!isUnsigned)
convert_s8_to_s16(inputBuffer + bytesReadFromInput, inputBuffer, samplesRead);
convert_s8_to_s16(inputBuffer + buffer_adder, inputBuffer, samplesRead);
else
convert_u8_to_s16(inputBuffer + bytesReadFromInput, inputBuffer, samplesRead);
memmove(inputBuffer, inputBuffer + bytesReadFromInput, samplesRead * 2);
convert_u8_to_s16(inputBuffer + buffer_adder, inputBuffer, samplesRead);
memmove(inputBuffer, inputBuffer + buffer_adder, samplesRead * 2);
bitsPerSample = 16;
bytesReadFromInput = samplesRead * 2;
isUnsigned = NO;
@ -584,8 +596,9 @@ tryagain:
samplesRead = bytesReadFromInput / 2;
if(isUnsigned)
convert_u16_to_s16(inputBuffer, samplesRead);
convert_s16_to_hdcd_input(inputBuffer + bytesReadFromInput, inputBuffer, samplesRead);
memmove(inputBuffer, inputBuffer + bytesReadFromInput, samplesRead * 4);
size_t buffer_adder = (bytesReadFromInput + 3) & ~3;
convert_s16_to_hdcd_input(inputBuffer + buffer_adder, inputBuffer, samplesRead);
memmove(inputBuffer, inputBuffer + buffer_adder, samplesRead * 4);
hdcd_process_stereo((hdcd_state_stereo_t *)hdcd_decoder, inputBuffer, (int)(samplesRead / 2));
if(((hdcd_state_stereo_t *)hdcd_decoder)->channel[0].sustain &&
((hdcd_state_stereo_t *)hdcd_decoder)->channel[1].sustain) {
@ -599,20 +612,23 @@ tryagain:
samplesRead = bytesReadFromInput / 2;
if(isUnsigned)
convert_u16_to_s16(inputBuffer, samplesRead);
vDSP_vflt16((const short *)inputBuffer, 1, (float *)(inputBuffer + bytesReadFromInput), 1, samplesRead);
scale_by_volume((float *)(inputBuffer + bytesReadFromInput), samplesRead, (float)(1.0 / (1ULL << 15)));
memmove(inputBuffer, inputBuffer + bytesReadFromInput, samplesRead * sizeof(float));
size_t buffer_adder = (bytesReadFromInput + 15) & ~15; // vDSP functions expect aligned to four elements
vDSP_vflt16((const short *)inputBuffer, 1, (float *)(inputBuffer + buffer_adder), 1, samplesRead);
float scale = 1ULL << 15;
vDSP_vsdiv((const float *)(inputBuffer + buffer_adder), 1, &scale, (float *)(inputBuffer + buffer_adder), 1, samplesRead);
memmove(inputBuffer, inputBuffer + buffer_adder, samplesRead * sizeof(float));
bitsPerSample = 32;
bytesReadFromInput = samplesRead * sizeof(float);
isUnsigned = NO;
isFloat = YES;
} else if(bitsPerSample <= 24) {
samplesRead = bytesReadFromInput / 3;
size_t buffer_adder = (bytesReadFromInput + 3) & ~3;
if(isUnsigned)
convert_u24_to_s32(inputBuffer + bytesReadFromInput, inputBuffer, samplesRead);
convert_u24_to_s32(inputBuffer + buffer_adder, inputBuffer, samplesRead);
else
convert_s24_to_s32(inputBuffer + bytesReadFromInput, inputBuffer, samplesRead);
memmove(inputBuffer, inputBuffer + bytesReadFromInput, samplesRead * 4);
convert_s24_to_s32(inputBuffer + buffer_adder, inputBuffer, samplesRead);
memmove(inputBuffer, inputBuffer + buffer_adder, samplesRead * 4);
bitsPerSample = 32;
bytesReadFromInput = samplesRead * 4;
isUnsigned = NO;
@ -621,9 +637,11 @@ tryagain:
samplesRead = bytesReadFromInput / 4;
if(isUnsigned)
convert_u32_to_s32(inputBuffer, samplesRead);
vDSP_vflt32((const int *)inputBuffer, 1, (float *)(inputBuffer + bytesReadFromInput), 1, samplesRead);
scale_by_volume((float *)(inputBuffer + bytesReadFromInput), samplesRead, gain * (1.0 / (1ULL << 31)));
memmove(inputBuffer, inputBuffer + bytesReadFromInput, samplesRead * sizeof(float));
size_t buffer_adder = (bytesReadFromInput + 31) & ~31; // vDSP functions expect aligned to four elements
vDSP_vflt32((const int *)inputBuffer, 1, (float *)(inputBuffer + buffer_adder), 1, samplesRead);
float scale = (1ULL << 31) / gain;
vDSP_vsdiv((const float *)(inputBuffer + buffer_adder), 1, &scale, (float *)(inputBuffer + buffer_adder), 1, samplesRead);
memmove(inputBuffer, inputBuffer + buffer_adder, samplesRead * sizeof(float));
bitsPerSample = 32;
bytesReadFromInput = samplesRead * sizeof(float);
isUnsigned = NO;