Optimize synth code by pre-computing the volume scaling for each note.
Scaling is now calculated once per MIDI event at the most, instead of once per sample. Increase voice ramping speed, increase number of active voices a little. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@9870 a1c6a512-1295-4272-9138-f99709370657
This commit is contained in:
Stepan Moskovchenko
parent
7f1d214801
commit
28b5afd05a
4 changed files with 35 additions and 27 deletions
|
|
@ -127,6 +127,7 @@ struct SynthObject
|
|||
int note, vol, ch, isUsed;
|
||||
int curRate, curOffset, targetOffset;
|
||||
int curPoint;
|
||||
signed short int volscale;
|
||||
};
|
||||
|
||||
struct SynthObject voices[MAX_VOICES] IBSS_ATTR;
|
||||
|
|
|
|||
|
|
@ -16,6 +16,7 @@
|
|||
*
|
||||
****************************************************************************/
|
||||
|
||||
void setVolScale(int a);
|
||||
|
||||
extern struct plugin_api * rb;
|
||||
|
||||
|
|
@ -24,8 +25,14 @@ long tempo=375000;
|
|||
|
||||
inline void setVol(int ch, int vol)
|
||||
{
|
||||
// printf("\nvolume[%d] %d ==> %d", ch, chVol[ch], vol);
|
||||
int a=0;
|
||||
chVol[ch]=vol;
|
||||
|
||||
/* If channel volume changes, we need to recalculate the volume scale */
|
||||
/* factor for all voices active on this channel */
|
||||
for(a=0; a<MAX_VOICES; a++)
|
||||
if(voices[a].ch == ch)
|
||||
setVolScale(a);
|
||||
}
|
||||
|
||||
inline void setPan(int ch, int pan)
|
||||
|
|
@ -111,25 +118,13 @@ long pitchTbl[] ICONST_ATTR={
|
|||
72901,72934,72967,72999,73032,73065,73098,73131,73164,73197,73230,73264,
|
||||
73297,73330,73363,73396,73429,73462,73495,73528
|
||||
};
|
||||
/*
|
||||
void findDelta(struct SynthObject * so, int ch, int note)
|
||||
{
|
||||
|
||||
struct GWaveform * wf = patchSet[chPat[ch]]->waveforms[patchSet[chPat[ch]]->noteTable[note]];
|
||||
so->wf=wf; // \|/ was 10
|
||||
so->delta = (((gustable[note]<<10) / (wf->rootFreq)) * wf->sampRate / (SAMPLE_RATE));
|
||||
so->delta = (so->delta * pitchTbl[chPW[ch]])>> 16;
|
||||
}
|
||||
*/
|
||||
|
||||
|
||||
void findDelta(struct SynthObject * so, int ch, int note)
|
||||
{
|
||||
|
||||
struct GWaveform * wf = patchSet[chPat[ch]]->waveforms[patchSet[chPat[ch]]->noteTable[note]];
|
||||
so->wf=wf; // \|/ was 10
|
||||
|
||||
unsigned long delta= 0 ;
|
||||
so->wf=wf;
|
||||
unsigned long delta= 0; /* More percision- extra bit - not so off-key as before */
|
||||
|
||||
delta = (((gustable[note]<<FRACTSIZE) / (wf->rootFreq)) * wf->sampRate / (SAMPLE_RATE));
|
||||
delta = (delta * pitchTbl[chPW[ch]])>> 16;
|
||||
|
|
@ -139,7 +134,6 @@ void findDelta(struct SynthObject * so, int ch, int note)
|
|||
|
||||
inline void setPW(int ch, int msb, int lsb)
|
||||
{
|
||||
// printf("\npitchw[%d] %d ==> %d", ch, chPW[ch], msb);
|
||||
chPW[ch] = msb<<2|lsb>>5;
|
||||
|
||||
int a=0;
|
||||
|
|
@ -152,16 +146,26 @@ inline void setPW(int ch, int msb, int lsb)
|
|||
}
|
||||
}
|
||||
|
||||
|
||||
/* Sets the volume scaling by channel volume and note volume */
|
||||
/* This way we can do the multiplication/indexing once per */
|
||||
/* MIDI event at the most, instead of once per sample. */
|
||||
void setVolScale(int a)
|
||||
{
|
||||
struct SynthObject * so = &voices[a];
|
||||
so->volscale = ((signed short int)so->vol*(signed short int)chVol[so->ch]);
|
||||
}
|
||||
|
||||
void pressNote(int ch, int note, int vol)
|
||||
{
|
||||
static int lastKill = 0;
|
||||
//Silences all channels but one, for easy debugging, for me.
|
||||
/* Silences all channels but one, for easy debugging, for me. */
|
||||
/*
|
||||
if(ch == 0) return;
|
||||
if(ch == 1) return;
|
||||
if(ch == 2) return;
|
||||
if(ch == 3) return;
|
||||
// if(ch == 4) return;
|
||||
if(ch == 4) return;
|
||||
if(ch == 5) return;
|
||||
if(ch == 6) return;
|
||||
if(ch == 7) return;
|
||||
|
|
@ -203,6 +207,7 @@ void pressNote(int ch, int note, int vol)
|
|||
voices[a].state=STATE_ATTACK;
|
||||
voices[a].decay=255;
|
||||
|
||||
setVolScale(a);
|
||||
|
||||
voices[a].loopState=STATE_NONLOOPING;
|
||||
voices[a].loopDir = LOOPDIR_FORWARD;
|
||||
|
|
@ -304,9 +309,7 @@ void sendEvent(struct Event * ev)
|
|||
|
||||
if((ev->status & 0xF0) == MIDI_PRGM)
|
||||
{
|
||||
if((ev->status & 0x0F) == 9)
|
||||
printf("\nNOT PATCHING: Someone tried patching Channel 9 onto something?");
|
||||
else
|
||||
if((ev->status & 0x0F) != 9)
|
||||
setPatch(ev->status & 0x0F, ev->d1);
|
||||
}
|
||||
}
|
||||
|
|
@ -351,7 +354,7 @@ int tick(void)
|
|||
if(e->d1 == 0x51)
|
||||
{
|
||||
tempo = (((short)e->evData[0])<<16)|(((short)e->evData[1])<<8)|(e->evData[2]);
|
||||
printf("\nMeta-Event: Tempo Set = %d", tempo);
|
||||
/* printf("\nMeta-Event: Tempo Set = %d", tempo); */
|
||||
bpm=mf->div*1000000/tempo;
|
||||
numberOfSamples=SAMPLE_RATE/bpm;
|
||||
|
||||
|
|
|
|||
|
|
@ -255,7 +255,6 @@ inline void stopVoice(struct SynthObject * so)
|
|||
|
||||
}
|
||||
|
||||
|
||||
signed short int synthVoice(struct SynthObject * so)
|
||||
{
|
||||
struct GWaveform * wf;
|
||||
|
|
@ -362,16 +361,21 @@ signed short int synthVoice(struct SynthObject * so)
|
|||
|
||||
if(so->state == STATE_RAMPDOWN)
|
||||
{
|
||||
so->decay--;
|
||||
if(so->decay == 0)
|
||||
so->decay-=5;
|
||||
if(so->decay < 5)
|
||||
so->isUsed=0;
|
||||
s = (s * so->decay) >> 8;
|
||||
}
|
||||
|
||||
return s*((signed short int)so->vol*(signed short int)chVol[so->ch])>>14;
|
||||
/* Scaling by channel volume and note volume is done in sequencer.c */
|
||||
/* That saves us some multiplication and pointer operations */
|
||||
return s*so->volscale>>14;
|
||||
}
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
inline void synthSample(int * mixL, int * mixR)
|
||||
{
|
||||
register int dL=0;
|
||||
|
|
|
|||
|
|
@ -22,7 +22,7 @@ PLUGIN_HEADER
|
|||
|
||||
#define FRACTSIZE 10
|
||||
#define SAMPLE_RATE 22050 // 44100 22050 11025
|
||||
#define MAX_VOICES 12 // Note: 24 midi channels is the minimum general midi
|
||||
#define MAX_VOICES 13 // Note: 24 midi channels is the minimum general midi
|
||||
// spec implementation
|
||||
#define BUF_SIZE 512
|
||||
#define NBUF 2
|
||||
|
|
|
|||
Loading…
Reference in a new issue