Extract X1-010 core from submodule

This commit is contained in:
cam900 2022-03-09 01:00:09 +09:00
parent 6c432bc42e
commit 66eb40e55e
6 changed files with 355 additions and 9 deletions

4
.gitmodules vendored
View file

@ -19,10 +19,6 @@
[submodule "extern/adpcm"] [submodule "extern/adpcm"]
path = extern/adpcm path = extern/adpcm
url = https://github.com/superctr/adpcm url = https://github.com/superctr/adpcm
[submodule "extern/cam900_vgsound_emu"]
path = extern/cam900_vgsound_emu
url = https://github.com/cam900/vgsound_emu
branch = main
[submodule "extern/Nuked-OPL3"] [submodule "extern/Nuked-OPL3"]
path = extern/Nuked-OPL3 path = extern/Nuked-OPL3
url = https://github.com/nukeykt/Nuked-OPL3.git url = https://github.com/nukeykt/Nuked-OPL3.git

View file

@ -228,8 +228,6 @@ extern/opm/opm.c
extern/Nuked-OPLL/opll.c extern/Nuked-OPLL/opll.c
extern/Nuked-OPL3/opl3.c extern/Nuked-OPL3/opl3.c
extern/cam900_vgsound_emu/x1_010/x1_010.cpp
src/engine/platform/sound/sn76496.cpp src/engine/platform/sound/sn76496.cpp
src/engine/platform/sound/ay8910.cpp src/engine/platform/sound/ay8910.cpp
src/engine/platform/sound/saa1099.cpp src/engine/platform/sound/saa1099.cpp
@ -268,6 +266,8 @@ src/engine/platform/sound/lynx/Mikey.cpp
src/engine/platform/sound/qsound.c src/engine/platform/sound/qsound.c
src/engine/platform/sound/x1_010/x1_010.cpp
src/engine/platform/sound/swan.cpp src/engine/platform/sound/swan.cpp
src/engine/platform/ym2610Interface.cpp src/engine/platform/ym2610Interface.cpp

@ -1 +0,0 @@
Subproject commit 3f8b5c5c6b996588afec7e4ce7469abccf16ecbe

View file

@ -0,0 +1,224 @@
/*
License: BSD-3-Clause
see https://github.com/cam900/vgsound_emu/LICENSE for more details
Copyright holders: cam900
Seta/Allumer X1-010 Emulation core
the chip has 16 voices, all voices can be switchable to Wavetable or PCM sample playback mode.
It has also 2 output channels, but no known hardware using this feature for stereo sound.
Wavetable needs to paired with envelope, it's always enabled and similar as AY PSG's one
but its shape is stored at RAM.
PCM volume is stored by each register.
Both volume is 4bit per output.
Everything except PCM sample is stored at paired 8 bit RAM.
RAM layout (common case: Address bit 12 is swapped when RAM is shared with CPU)
-----------------------------
0000...007f Voice Registers
0000...0007 Voice 0 Register
Address Bits Description
7654 3210
0 x--- ---- Frequency divider*
---- -x-- Envelope one-shot mode
---- --x- Sound format
---- --0- PCM
---- --1- Wavetable
---- ---x Keyon/off
PCM case:
1 xxxx xxxx Volume (Each nibble is for each output)
2 xxxx xxxx Frequency*
4 xxxx xxxx Start address / 4096
5 xxxx xxxx 0x100 - (End address / 4096)
Wavetable case:
1 ---x xxxx Wavetable data select
2 xxxx xxxx Frequency LSB*
3 xxxx xxxx "" MSB
4 xxxx xxxx Envelope period (.10 fixed point, Low 8 bit)
5 ---x xxxx Envelope shape select (!= 0 : Reserved for Voice registers)
0008...000f Voice 1 Register
...
0078...007f Voice 15 Register
-----------------------------
0080...0fff Envelope shape data (Same as volume; Each nibble is for each output)
0080...00ff Envelope shape data 1
0100...017f Envelope shape data 2
...
0f80...0fff Envelope shape data 31
-----------------------------
1000...1fff Wavetable data
1000...107f Wavetable data 0
1080...10ff Wavetable data 1
...
1f80...1fff Wavetable data 31
-----------------------------
* Frequency is 4.4 fixed point for PCM,
6.10 for Wavetable.
Frequency divider is higher precision or just right shift?
needs verification.
*/
#include "x1_010.hpp"
void x1_010_core::tick()
{
// reset output
m_out[0] = m_out[1] = 0;
for (int i = 0; i < 16; i++)
{
voice_t &v = m_voice[i];
v.tick();
m_out[0] += v.data * v.vol_out[0];
m_out[1] += v.data * v.vol_out[1];
}
}
void x1_010_core::voice_t::tick()
{
data = vol_out[0] = vol_out[1] = 0;
if (flag.keyon)
{
if (flag.wavetable) // Wavetable
{
// envelope, each nibble is for each output
u8 vol = m_host.m_envelope[(bitfield(end_envshape, 0, 5) << 7) | bitfield(env_acc, 10, 7)];
vol_out[0] = bitfield(vol, 4, 4);
vol_out[1] = bitfield(vol, 0, 4);
env_acc += start_envfreq;
if (flag.env_oneshot && bitfield(env_acc, 17))
flag.keyon = false;
else
env_acc = bitfield(env_acc, 0, 17);
// get wavetable data
data = m_host.m_wave[(bitfield(vol_wave, 0, 5) << 7) | bitfield(acc, 10, 7)];
acc = bitfield(acc + (freq >> flag.div), 0, 17);
}
else // PCM sample
{
// volume register, each nibble is for each output
vol_out[0] = bitfield(vol_wave, 4, 4);
vol_out[1] = bitfield(vol_wave, 0, 4);
// get PCM sample
data = m_host.m_intf.read_byte(bitfield(acc, 4, 20));
acc += bitfield(freq, 0, 8) >> flag.div;
if ((acc >> 16) > (0xff ^ end_envshape))
flag.keyon = false;
}
}
}
u8 x1_010_core::ram_r(u16 offset)
{
if (offset & 0x1000) // wavetable data
return m_wave[offset & 0xfff];
else if (offset & 0xf80) // envelope shape data
return m_envelope[offset & 0xfff];
else // channel register
return m_voice[bitfield(offset, 3, 4)].reg_r(offset & 0x7);
}
void x1_010_core::ram_w(u16 offset, u8 data)
{
if (offset & 0x1000) // wavetable data
m_wave[offset & 0xfff] = data;
else if (offset & 0xf80) // envelope shape data
m_envelope[offset & 0xfff] = data;
else // channel register
m_voice[bitfield(offset, 3, 4)].reg_w(offset & 0x7, data);
}
u8 x1_010_core::voice_t::reg_r(u8 offset)
{
switch (offset & 0x7)
{
case 0x00: return (flag.div << 7)
| (flag.env_oneshot << 2)
| (flag.wavetable << 1)
| (flag.keyon << 0);
case 0x01: return vol_wave;
case 0x02: return bitfield(freq, 0, 8);
case 0x03: return bitfield(freq, 8, 8);
case 0x04: return start_envfreq;
case 0x05: return end_envshape;
default: break;
}
return 0;
}
void x1_010_core::voice_t::reg_w(u8 offset, u8 data)
{
switch (offset & 0x7)
{
case 0x00:
{
const bool prev_keyon = flag.keyon;
flag.div = bitfield(data, 7);
flag.env_oneshot = bitfield(data, 2);
flag.wavetable = bitfield(data, 1);
flag.keyon = bitfield(data, 0);
if (!prev_keyon && flag.keyon) // Key on
{
acc = flag.wavetable ? 0 : (u32(start_envfreq) << 16);
env_acc = 0;
}
break;
}
case 0x01:
vol_wave = data;
break;
case 0x02:
freq = (freq & 0xff00) | data;
break;
case 0x03:
freq = (freq & 0x00ff) | (u16(data) << 8);
break;
case 0x04:
start_envfreq = data;
break;
case 0x05:
end_envshape = data;
break;
default:
break;
}
}
void x1_010_core::voice_t::reset()
{
flag.reset();
vol_wave = 0;
freq = 0;
start_envfreq = 0;
end_envshape = 0;
acc = 0;
env_acc = 0;
data = 0;
vol_out[0] = vol_out[1] = 0;
}
void x1_010_core::reset()
{
for (int i = 0; i < 16; i++)
m_voice[i].reset();
std::fill_n(&m_envelope[0], 0x1000, 0);
std::fill_n(&m_wave[0], 0x1000, 0);
m_out[0] = m_out[1] = 0;
}

View file

@ -0,0 +1,127 @@
/*
License: BSD-3-Clause
see https://github.com/cam900/vgsound_emu/LICENSE for more details
Copyright holders: cam900
Seta/Allumer X1-010 Emulation core
See x1_010.cpp for more info.
*/
#include <algorithm>
#include <memory>
#ifndef _VGSOUND_EMU_X1_010_HPP
#define _VGSOUND_EMU_X1_010_HPP
#pragma once
typedef unsigned char u8;
typedef unsigned short u16;
typedef unsigned int u32;
typedef signed char s8;
typedef signed int s32;
template<typename T> T bitfield(T in, u8 pos, u8 len = 1)
{
return (in >> pos) & (len ? (T(1 << len) - 1) : 1);
}
class x1_010_mem_intf
{
public:
virtual u8 read_byte(u32 address) { return 0; }
};
class x1_010_core
{
friend class x1_010_mem_intf;
public:
// constructor
x1_010_core(x1_010_mem_intf &intf)
: m_voice{*this,*this,*this,*this,
*this,*this,*this,*this,
*this,*this,*this,*this,
*this,*this,*this,*this}
, m_intf(intf)
{
m_envelope = std::make_unique<u8[]>(0x1000);
m_wave = std::make_unique<u8[]>(0x1000);
std::fill_n(&m_envelope[0], 0x1000, 0);
std::fill_n(&m_wave[0], 0x1000, 0);
}
// register accessor
u8 ram_r(u16 offset);
void ram_w(u16 offset, u8 data);
// getters
s32 output(u8 channel) { return m_out[channel & 1]; }
// internal state
void reset();
void tick();
private:
// 16 voices in chip
struct voice_t
{
// constructor
voice_t(x1_010_core &host) : m_host(host) {}
// internal state
void reset();
void tick();
// register accessor
u8 reg_r(u8 offset);
void reg_w(u8 offset, u8 data);
// registers
x1_010_core &m_host;
struct flag_t
{
u8 div : 1;
u8 env_oneshot : 1;
u8 wavetable : 1;
u8 keyon : 1;
void reset()
{
div = 0;
env_oneshot = 0;
wavetable = 0;
keyon = 0;
}
flag_t()
: div(0)
, env_oneshot(0)
, wavetable(0)
, keyon(0)
{ }
};
flag_t flag;
u8 vol_wave = 0;
u16 freq = 0;
u8 start_envfreq = 0;
u8 end_envshape = 0;
// internal registers
u32 acc = 0;
u32 env_acc = 0;
s8 data = 0;
u8 vol_out[2] = {0};
};
voice_t m_voice[16];
// RAM
std::unique_ptr<u8[]> m_envelope = nullptr;
std::unique_ptr<u8[]> m_wave = nullptr;
// output data
s32 m_out[2] = {0};
x1_010_mem_intf &m_intf;
};
#endif

View file

@ -24,9 +24,9 @@
#include "../dispatch.h" #include "../dispatch.h"
#include "../engine.h" #include "../engine.h"
#include "../macroInt.h" #include "../macroInt.h"
#include "../../../extern/cam900_vgsound_emu/x1_010/x1_010.hpp" #include "sound/x1_010/x1_010.hpp"
class DivX1_010Interface: public vgsound_emu_mem_intf { class DivX1_010Interface: public x1_010_mem_intf {
public: public:
DivEngine* parent; DivEngine* parent;
int sampleBank; int sampleBank;