zdoom/qzdoom
0
0
Fork 0
mirror of https://github.com/ZDoom/qzdoom.git synced 2024-12-11 05:01:09 +00:00
Code Issues Projects Releases Packages Wiki Activity
qzdoom/snes_spc/fast_dsp/SPC_DSP.cpp
Randy Heit 3bfcc5c09c - Removed lots of spc_* cvars that are no longer meaningful and changed 
spc_amp from a x.4 fixed point number to a normal float.
- Switched SPC playback from the external SNESAPU.DLL to Blargg's LGPL
  snes_spc library. I've compiled it with the fast DSP rather than the
  highly accurate one, since I didn't notice a meaningful difference between
  the two in my limited testing. In short: SPC playback is now built in to
  ZDoom. You don't need to download anything extra to make it work, and it
  also works on Linux as well as Windows (though building with Linux is
  currently untested).
- Fixed: Stereo separation was calculated very wrongly when in 2D sound mode.


SVN r794 (trunk)
2008-03-11 22:17:57 +00:00

703 lines
19 KiB
C++
Raw Blame History

// snes_spc 0.9.0. http://www.slack.net/~ant/
#include "SPC_DSP.h"
#include "blargg_endian.h"
#include <string.h>
/* Copyright (C) 2007 Shay Green. This module is free software; you
can redistribute it and/or modify it under the terms of the GNU Lesser
General Public License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version. This
module is distributed in the hope that it will be useful, but WITHOUT ANY
WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
details. You should have received a copy of the GNU Lesser General Public
License along with this module; if not, write to the Free Software Foundation,
Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */
#include "blargg_source.h"
#ifdef BLARGG_ENABLE_OPTIMIZER
#include BLARGG_ENABLE_OPTIMIZER
#endif
#if INT_MAX < 0x7FFFFFFF
#error "Requires that int type have at least 32 bits"
#endif
// TODO: add to blargg_endian.h
#define GET_LE16SA( addr ) ((BOOST::int16_t) GET_LE16( addr ))
#define GET_LE16A( addr ) GET_LE16( addr )
#define SET_LE16A( addr, data ) SET_LE16( addr, data )
static BOOST::uint8_t const initial_regs [SPC_DSP::register_count] =
{
0x45,0x8B,0x5A,0x9A,0xE4,0x82,0x1B,0x78,0x00,0x00,0xAA,0x96,0x89,0x0E,0xE0,0x80,
0x2A,0x49,0x3D,0xBA,0x14,0xA0,0xAC,0xC5,0x00,0x00,0x51,0xBB,0x9C,0x4E,0x7B,0xFF,
0xF4,0xFD,0x57,0x32,0x37,0xD9,0x42,0x22,0x00,0x00,0x5B,0x3C,0x9F,0x1B,0x87,0x9A,
0x6F,0x27,0xAF,0x7B,0xE5,0x68,0x0A,0xD9,0x00,0x00,0x9A,0xC5,0x9C,0x4E,0x7B,0xFF,
0xEA,0x21,0x78,0x4F,0xDD,0xED,0x24,0x14,0x00,0x00,0x77,0xB1,0xD1,0x36,0xC1,0x67,
0x52,0x57,0x46,0x3D,0x59,0xF4,0x87,0xA4,0x00,0x00,0x7E,0x44,0x9C,0x4E,0x7B,0xFF,
0x75,0xF5,0x06,0x97,0x10,0xC3,0x24,0xBB,0x00,0x00,0x7B,0x7A,0xE0,0x60,0x12,0x0F,
0xF7,0x74,0x1C,0xE5,0x39,0x3D,0x73,0xC1,0x00,0x00,0x7A,0xB3,0xFF,0x4E,0x7B,0xFF
};
// if ( io < -32768 ) io = -32768;
// if ( io > 32767 ) io = 32767;
#define CLAMP16( io )\
{\
if ( (int16_t) io != io )\
io = (io >> 31) ^ 0x7FFF;\
}
// Access global DSP register
#define REG(n) m.regs [r_##n]
// Access voice DSP register
#define VREG(r,n) r [v_##n]
#define WRITE_SAMPLES( l, r, out ) \
{\
out [0] = l;\
out [1] = r;\
out += 2;\
if ( out >= m.out_end )\
{\
check( out == m.out_end );\
check( m.out_end != &m.extra [extra_size] || \
(m.extra <= m.out_begin && m.extra < &m.extra [extra_size]) );\
out = m.extra;\
m.out_end = &m.extra [extra_size];\
}\
}\
void SPC_DSP::set_output( sample_t* out, int size )
{
require( (size & 1) == 0 ); // must be even
if ( !out )
{
out = m.extra;
size = extra_size;
}
m.out_begin = out;
m.out = out;
m.out_end = out + size;
}
// Volume registers and efb are signed! Easy to forget int8_t cast.
// Prefixes are to avoid accidental use of locals with same names.
// Interleved gauss table (to improve cache coherency)
// interleved_gauss [i] = gauss [(i & 1) * 256 + 255 - (i >> 1 & 0xFF)]
static short const interleved_gauss [512] =
{
370,1305, 366,1305, 362,1304, 358,1304, 354,1304, 351,1304, 347,1304, 343,1303,
339,1303, 336,1303, 332,1302, 328,1302, 325,1301, 321,1300, 318,1300, 314,1299,
311,1298, 307,1297, 304,1297, 300,1296, 297,1295, 293,1294, 290,1293, 286,1292,
283,1291, 280,1290, 276,1288, 273,1287, 270,1286, 267,1284, 263,1283, 260,1282,
257,1280, 254,1279, 251,1277, 248,1275, 245,1274, 242,1272, 239,1270, 236,1269,
233,1267, 230,1265, 227,1263, 224,1261, 221,1259, 218,1257, 215,1255, 212,1253,
210,1251, 207,1248, 204,1246, 201,1244, 199,1241, 196,1239, 193,1237, 191,1234,
188,1232, 186,1229, 183,1227, 180,1224, 178,1221, 175,1219, 173,1216, 171,1213,
168,1210, 166,1207, 163,1205, 161,1202, 159,1199, 156,1196, 154,1193, 152,1190,
150,1186, 147,1183, 145,1180, 143,1177, 141,1174, 139,1170, 137,1167, 134,1164,
132,1160, 130,1157, 128,1153, 126,1150, 124,1146, 122,1143, 120,1139, 118,1136,
117,1132, 115,1128, 113,1125, 111,1121, 109,1117, 107,1113, 106,1109, 104,1106,
102,1102, 100,1098, 99,1094, 97,1090, 95,1086, 94,1082, 92,1078, 90,1074,
89,1070, 87,1066, 86,1061, 84,1057, 83,1053, 81,1049, 80,1045, 78,1040,
77,1036, 76,1032, 74,1027, 73,1023, 71,1019, 70,1014, 69,1010, 67,1005,
66,1001, 65, 997, 64, 992, 62, 988, 61, 983, 60, 978, 59, 974, 58, 969,
56, 965, 55, 960, 54, 955, 53, 951, 52, 946, 51, 941, 50, 937, 49, 932,
48, 927, 47, 923, 46, 918, 45, 913, 44, 908, 43, 904, 42, 899, 41, 894,
40, 889, 39, 884, 38, 880, 37, 875, 36, 870, 36, 865, 35, 860, 34, 855,
33, 851, 32, 846, 32, 841, 31, 836, 30, 831, 29, 826, 29, 821, 28, 816,
27, 811, 27, 806, 26, 802, 25, 797, 24, 792, 24, 787, 23, 782, 23, 777,
22, 772, 21, 767, 21, 762, 20, 757, 20, 752, 19, 747, 19, 742, 18, 737,
17, 732, 17, 728, 16, 723, 16, 718, 15, 713, 15, 708, 15, 703, 14, 698,
14, 693, 13, 688, 13, 683, 12, 678, 12, 674, 11, 669, 11, 664, 11, 659,
10, 654, 10, 649, 10, 644, 9, 640, 9, 635, 9, 630, 8, 625, 8, 620,
8, 615, 7, 611, 7, 606, 7, 601, 6, 596, 6, 592, 6, 587, 6, 582,
5, 577, 5, 573, 5, 568, 5, 563, 4, 559, 4, 554, 4, 550, 4, 545,
4, 540, 3, 536, 3, 531, 3, 527, 3, 522, 3, 517, 2, 513, 2, 508,
2, 504, 2, 499, 2, 495, 2, 491, 2, 486, 1, 482, 1, 477, 1, 473,
1, 469, 1, 464, 1, 460, 1, 456, 1, 451, 1, 447, 1, 443, 1, 439,
0, 434, 0, 430, 0, 426, 0, 422, 0, 418, 0, 414, 0, 410, 0, 405,
0, 401, 0, 397, 0, 393, 0, 389, 0, 385, 0, 381, 0, 378, 0, 374,
};
//// Counters
#define RATE( rate, div )\
(rate >= div ? rate / div * 8 - 1 : rate - 1)
static unsigned const counter_mask [32] =
{
RATE( 2,2), RATE(2048,4), RATE(1536,3),
RATE(1280,5), RATE(1024,4), RATE( 768,3),
RATE( 640,5), RATE( 512,4), RATE( 384,3),
RATE( 320,5), RATE( 256,4), RATE( 192,3),
RATE( 160,5), RATE( 128,4), RATE( 96,3),
RATE( 80,5), RATE( 64,4), RATE( 48,3),
RATE( 40,5), RATE( 32,4), RATE( 24,3),
RATE( 20,5), RATE( 16,4), RATE( 12,3),
RATE( 10,5), RATE( 8,4), RATE( 6,3),
RATE( 5,5), RATE( 4,4), RATE( 3,3),
RATE( 2,4),
RATE( 1,4)
};
#undef RATE
inline void SPC_DSP::init_counter()
{
// counters start out with this synchronization
m.counters [0] = 1;
m.counters [1] = 0;
m.counters [2] = -0x20u;
m.counters [3] = 0x0B;
int n = 2;
for ( int i = 1; i < 32; i++ )
{
m.counter_select [i] = &m.counters [n];
if ( !--n )
n = 3;
}
m.counter_select [ 0] = &m.counters [0];
m.counter_select [30] = &m.counters [2];
}
inline void SPC_DSP::run_counter( int i )
{
int n = m.counters [i];
if ( !(n-- & 7) )
n -= 6 - i;
m.counters [i] = n;
}
#define READ_COUNTER( rate )\
(*m.counter_select [rate] & counter_mask [rate])
//// Emulation
void SPC_DSP::run( int clock_count )
{
int new_phase = m.phase + clock_count;
int count = new_phase >> 5;
m.phase = new_phase & 31;
if ( !count )
return;
uint8_t* const ram = m.ram;
uint8_t const* const dir = &ram [REG(dir) * 0x100];
int const slow_gaussian = (REG(pmon) >> 1) | REG(non);
int const noise_rate = REG(flg) & 0x1F;
// Global volume
int mvoll = (int8_t) REG(mvoll);
int mvolr = (int8_t) REG(mvolr);
if ( mvoll * mvolr < m.surround_threshold )
mvoll = -mvoll; // eliminate surround
do
{
// KON/KOFF reading
if ( (m.every_other_sample ^= 1) != 0 )
{
m.new_kon &= ~m.kon;
m.kon = m.new_kon;
m.t_koff = REG(koff);
}
run_counter( 1 );
run_counter( 2 );
run_counter( 3 );
// Noise
if ( !READ_COUNTER( noise_rate ) )
{
int feedback = (m.noise << 13) ^ (m.noise << 14);
m.noise = (feedback & 0x4000) ^ (m.noise >> 1);
}
// Voices
int pmon_input = 0;
int main_out_l = 0;
int main_out_r = 0;
int echo_out_l = 0;
int echo_out_r = 0;
voice_t* v = m.voices;
uint8_t* v_regs = m.regs;
int vbit = 1;
do
{
#define SAMPLE_PTR(i) GET_LE16A( &dir [VREG(v_regs,srcn) * 4 + i * 2] )
int brr_header = ram [v->brr_addr];
int kon_delay = v->kon_delay;
// Pitch
int pitch = GET_LE16A( &VREG(v_regs,pitchl) ) & 0x3FFF;
if ( REG(pmon) & vbit )
pitch += ((pmon_input >> 5) * pitch) >> 10;
// KON phases
if ( --kon_delay >= 0 )
{
v->kon_delay = kon_delay;
// Get ready to start BRR decoding on next sample
if ( kon_delay == 4 )
{
v->brr_addr = SAMPLE_PTR( 0 );
v->brr_offset = 1;
v->buf_pos = v->buf;
brr_header = 0; // header is ignored on this sample
}
// Envelope is never run during KON
v->env = 0;
v->hidden_env = 0;
// Disable BRR decoding until last three samples
v->interp_pos = (kon_delay & 3 ? 0x4000 : 0);
// Pitch is never added during KON
pitch = 0;
}
int env = v->env;
// Gaussian interpolation
{
int output = 0;
VREG(v_regs,envx) = (uint8_t) (env >> 4);
if ( env )
{
// Make pointers into gaussian based on fractional position between samples
int offset = (unsigned) v->interp_pos >> 3 & 0x1FE;
short const* fwd = interleved_gauss + offset;
short const* rev = interleved_gauss + 510 - offset; // mirror left half of gaussian
int const* in = &v->buf_pos [(unsigned) v->interp_pos >> 12];
if ( !(slow_gaussian & vbit) ) // 99%
{
// Faster approximation when exact sample value isn't necessary for pitch mod
output = (fwd [0] * in [0] +
fwd [1] * in [1] +
rev [1] * in [2] +
rev [0] * in [3]) >> 11;
output = (output * env) >> 11;
}
else
{
output = (int16_t) (m.noise * 2);
if ( !(REG(non) & vbit) )
{
output = (fwd [0] * in [0]) >> 11;
output += (fwd [1] * in [1]) >> 11;
output += (rev [1] * in [2]) >> 11;
output = (int16_t) output;
output += (rev [0] * in [3]) >> 11;
CLAMP16( output );
output &= ~1;
}
output = (output * env) >> 11 & ~1;
}
// Output
int l = output * v->volume [0];
int r = output * v->volume [1];
main_out_l += l;
main_out_r += r;
if ( REG(eon) & vbit )
{
echo_out_l += l;
echo_out_r += r;
}
}
pmon_input = output;
VREG(v_regs,outx) = (uint8_t) (output >> 8);
}
// Soft reset or end of sample
if ( REG(flg) & 0x80 || (brr_header & 3) == 1 )
{
v->env_mode = env_release;
env = 0;
}
if ( m.every_other_sample )
{
// KOFF
if ( m.t_koff & vbit )
v->env_mode = env_release;
// KON
if ( m.kon & vbit )
{
v->kon_delay = 5;
v->env_mode = env_attack;
REG(endx) &= ~vbit;
}
}
// Envelope
if ( !v->kon_delay )
{
if ( v->env_mode == env_release ) // 97%
{
env -= 0x8;
v->env = env;
if ( env <= 0 )
{
v->env = 0;
goto skip_brr; // no BRR decoding for you!
}
}
else // 3%
{
int rate;
int const adsr0 = VREG(v_regs,adsr0);
int env_data = VREG(v_regs,adsr1);
if ( adsr0 >= 0x80 ) // 97% ADSR
{
if ( v->env_mode > env_decay ) // 89%
{
env--;
env -= env >> 8;
rate = env_data & 0x1F;
// optimized handling
v->hidden_env = env;
if ( READ_COUNTER( rate ) )
goto exit_env;
v->env = env;
goto exit_env;
}
else if ( v->env_mode == env_decay )
{
env--;
env -= env >> 8;
rate = (adsr0 >> 3 & 0x0E) + 0x10;
}
else // env_attack
{
rate = (adsr0 & 0x0F) * 2 + 1;
env += rate < 31 ? 0x20 : 0x400;
}
}
else // GAIN
{
int mode;
env_data = VREG(v_regs,gain);
mode = env_data >> 5;
if ( mode < 4 ) // direct
{
env = env_data * 0x10;
rate = 31;
}
else
{
rate = env_data & 0x1F;
if ( mode == 4 ) // 4: linear decrease
{
env -= 0x20;
}
else if ( mode < 6 ) // 5: exponential decrease
{
env--;
env -= env >> 8;
}
else // 6,7: linear increase
{
env += 0x20;
if ( mode > 6 && (unsigned) v->hidden_env >= 0x600 )
env += 0x8 - 0x20; // 7: two-slope linear increase
}
}
}
// Sustain level
if ( (env >> 8) == (env_data >> 5) && v->env_mode == env_decay )
v->env_mode = env_sustain;
v->hidden_env = env;
// unsigned cast because linear decrease going negative also triggers this
if ( (unsigned) env > 0x7FF )
{
env = (env < 0 ? 0 : 0x7FF);
if ( v->env_mode == env_attack )
v->env_mode = env_decay;
}
if ( !READ_COUNTER( rate ) )
v->env = env; // nothing else is controlled by the counter
}
}
exit_env:
{
// Apply pitch
int old_pos = v->interp_pos;
int interp_pos = (old_pos & 0x3FFF) + pitch;
if ( interp_pos > 0x7FFF )
interp_pos = 0x7FFF;
v->interp_pos = interp_pos;
// BRR decode if necessary
if ( old_pos >= 0x4000 )
{
// Arrange the four input nybbles in 0xABCD order for easy decoding
int nybbles = ram [(v->brr_addr + v->brr_offset) & 0xFFFF] * 0x100 +
ram [(v->brr_addr + v->brr_offset + 1) & 0xFFFF];
// Advance read position
int const brr_block_size = 9;
int brr_offset = v->brr_offset;
if ( (brr_offset += 2) >= brr_block_size )
{
// Next BRR block
int brr_addr = (v->brr_addr + brr_block_size) & 0xFFFF;
assert( brr_offset == brr_block_size );
if ( brr_header & 1 )
{
brr_addr = SAMPLE_PTR( 1 );
if ( !v->kon_delay )
REG(endx) |= vbit;
}
v->brr_addr = brr_addr;
brr_offset = 1;
}
v->brr_offset = brr_offset;
// Decode
// 0: >>1 1: <<0 2: <<1 ... 12: <<11 13-15: >>4 <<11
static unsigned char const shifts [16 * 2] = {
13,12,12,12,12,12,12,12,12,12,12, 12, 12, 16, 16, 16,
0, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 11, 11, 11
};
int const scale = brr_header >> 4;
int const right_shift = shifts [scale];
int const left_shift = shifts [scale + 16];
// Write to next four samples in circular buffer
int* pos = v->buf_pos;
int* end;
// Decode four samples
for ( end = pos + 4; pos < end; pos++, nybbles <<= 4 )
{
// Extract upper nybble and scale appropriately
int s = ((int16_t) nybbles >> right_shift) << left_shift;
// Apply IIR filter (8 is the most commonly used)
int const filter = brr_header & 0x0C;
int const p1 = pos [brr_buf_size - 1];
int const p2 = pos [brr_buf_size - 2] >> 1;
if ( filter >= 8 )
{
s += p1;
s -= p2;
if ( filter == 8 ) // s += p1 * 0.953125 - p2 * 0.46875
{
s += p2 >> 4;
s += (p1 * -3) >> 6;
}
else // s += p1 * 0.8984375 - p2 * 0.40625
{
s += (p1 * -13) >> 7;
s += (p2 * 3) >> 4;
}
}
else if ( filter ) // s += p1 * 0.46875
{
s += p1 >> 1;
s += (-p1) >> 5;
}
// Adjust and write sample
CLAMP16( s );
s = (int16_t) (s * 2);
pos [brr_buf_size] = pos [0] = s; // second copy simplifies wrap-around
}
if ( pos >= &v->buf [brr_buf_size] )
pos = v->buf;
v->buf_pos = pos;
}
}
skip_brr:
// Next voice
vbit <<= 1;
v_regs += 0x10;
v++;
}
while ( vbit < 0x100 );
// Echo position
int echo_offset = m.echo_offset;
uint8_t* const echo_ptr = &ram [(REG(esa) * 0x100 + echo_offset) & 0xFFFF];
if ( !echo_offset )
m.echo_length = (REG(edl) & 0x0F) * 0x800;
echo_offset += 4;
if ( echo_offset >= m.echo_length )
echo_offset = 0;
m.echo_offset = echo_offset;
// FIR
int echo_in_l = GET_LE16SA( echo_ptr + 0 );
int echo_in_r = GET_LE16SA( echo_ptr + 2 );
int (*echo_hist_pos) [2] = m.echo_hist_pos;
if ( ++echo_hist_pos >= &m.echo_hist [echo_hist_size] )
echo_hist_pos = m.echo_hist;
m.echo_hist_pos = echo_hist_pos;
echo_hist_pos [0] [0] = echo_hist_pos [8] [0] = echo_in_l;
echo_hist_pos [0] [1] = echo_hist_pos [8] [1] = echo_in_r;
#define CALC_FIR_( i, in ) ((in) * (int8_t) REG(fir + i * 0x10))
echo_in_l = CALC_FIR_( 7, echo_in_l );
echo_in_r = CALC_FIR_( 7, echo_in_r );
#define CALC_FIR( i, ch ) CALC_FIR_( i, echo_hist_pos [i + 1] [ch] )
#define DO_FIR( i )\
echo_in_l += CALC_FIR( i, 0 );\
echo_in_r += CALC_FIR( i, 1 );
DO_FIR( 0 );
DO_FIR( 1 );
DO_FIR( 2 );
#if defined (__MWERKS__) && __MWERKS__ < 0x3200
__eieio(); // keeps compiler from stupidly "caching" things in memory
#endif
DO_FIR( 3 );
DO_FIR( 4 );
DO_FIR( 5 );
DO_FIR( 6 );
// Echo out
if ( !(REG(flg) & 0x20) )
{
int l = (echo_out_l >> 7) + ((echo_in_l * (int8_t) REG(efb)) >> 14);
int r = (echo_out_r >> 7) + ((echo_in_r * (int8_t) REG(efb)) >> 14);
// just to help pass more validation tests
#if SPC_MORE_ACCURACY
l &= ~1;
r &= ~1;
#endif
CLAMP16( l );
CLAMP16( r );
SET_LE16A( echo_ptr + 0, l );
SET_LE16A( echo_ptr + 2, r );
}
// Sound out
int l = (main_out_l * mvoll + echo_in_l * (int8_t) REG(evoll)) >> 14;
int r = (main_out_r * mvolr + echo_in_r * (int8_t) REG(evolr)) >> 14;
CLAMP16( l );
CLAMP16( r );
if ( (REG(flg) & 0x40) )
{
l = 0;
r = 0;
}
sample_t* out = m.out;
WRITE_SAMPLES( l, r, out );
m.out = out;
}
while ( --count );
}
//// Setup
void SPC_DSP::mute_voices( int mask )
{
m.mute_mask = mask;
for ( int i = 0; i < voice_count; i++ )
{
m.voices [i].enabled = (mask >> i & 1) - 1;
update_voice_vol( i * 0x10 );
}
}
void SPC_DSP::init( void* ram_64k )
{
m.ram = (uint8_t*) ram_64k;
mute_voices( 0 );
disable_surround( false );
set_output( 0, 0 );
reset();
#ifndef NDEBUG
// be sure this sign-extends
assert( (int16_t) 0x8000 == -0x8000 );
// be sure right shift preserves sign
assert( (-1 >> 1) == -1 );
// check clamp macro
int i;
i = +0x8000; CLAMP16( i ); assert( i == +0x7FFF );
i = -0x8001; CLAMP16( i ); assert( i == -0x8000 );
blargg_verify_byte_order();
#endif
}
void SPC_DSP::soft_reset_common()
{
require( m.ram ); // init() must have been called already
m.noise = 0x4000;
m.echo_hist_pos = m.echo_hist;
m.every_other_sample = 1;
m.echo_offset = 0;
m.phase = 0;
init_counter();
}
void SPC_DSP::soft_reset()
{
REG(flg) = 0xE0;
soft_reset_common();
}
void SPC_DSP::load( uint8_t const regs [register_count] )
{
memcpy( m.regs, regs, sizeof m.regs );
memset( &m.regs [register_count], 0, offsetof (state_t,ram) - register_count );
// Internal state
int i;
for ( i = voice_count; --i >= 0; )
{
voice_t& v = m.voices [i];
v.brr_offset = 1;
v.buf_pos = v.buf;
}
m.new_kon = REG(kon);
mute_voices( m.mute_mask );
soft_reset_common();
}
void SPC_DSP::reset() { load( initial_regs ); }
Reference in a new issue View git blame Copy permalink