nullsound: big refactor and new pipeline processing

This commit holds a massive refactoring of the FM and SSG processing that is hapenning at every step. Now every channel has a fixed sound pipeline where every NSS opcode records actions to perform for a row, and the processing of those actions and FX takes place at every step. SSG channel has its macro processing reworked. Now every register can be updated at every step of the macro definition, and loop support has been added. This is a breaking commit that requires recompiling the NSS instruments and streams. The user API stays the same.
dciabrin · Nov 22, 2024 · 5803d3a · 5803d3a
1 parent dad1c17
commit 5803d3a
Show file tree

Hide file tree

Showing 15 changed files with 1,537 additions and 1,807 deletions.
diff --git a/.github/scripts/test.sh b/.github/scripts/test.sh
@@ -5,4 +5,6 @@ if [ -n "$PY3PATH" ]; then
 fi
 cd examples
 export XDG_RUNTIME_DIR=$HOME
-./configure && make
+MAKE=$(which gmake make | head -1)
+$MAKE
+./configure && $MAKE
diff --git a/.github/workflows/build-and-publish.yaml b/.github/workflows/build-and-publish.yaml
@@ -87,7 +87,9 @@ jobs:
         run: ./.github/scripts/build.sh
 
       - name: Install dependencies for ngdevkit-examples
-        run: brew install pyyaml ngdevkit-gngeo imagemagick sox glew sdl2 sdl2_image
+        run: |
+          brew install ngdevkit-gngeo imagemagick sox glew sdl2 sdl2_image
+          $(brew --prefix python)/libexec/bin/pip install pyyaml --break-system-packages
 
       - name: Test by compiling ngdevkit-examples
         run: ./.github/scripts/test.sh

diff --git a/.github/workflows/build-tests.yaml b/.github/workflows/build-tests.yaml
@@ -87,7 +87,9 @@ jobs:
         run: ./.github/scripts/build.sh
 
       - name: Install dependencies for ngdevkit-examples
-        run: brew install pyyaml ngdevkit-gngeo imagemagick sox glew sdl2 sdl2_image
+        run: |
+          brew install ngdevkit-gngeo imagemagick sox glew sdl2 sdl2_image
+          $(brew --prefix python)/libexec/bin/pip install pyyaml --break-system-packages
 
       - name: Test by compiling ngdevkit-examples
         run: ./.github/scripts/test.sh
diff --git a/nullsound/Makefile.in b/nullsound/Makefile.in
@@ -19,7 +19,7 @@ all: nullsound.lib linkcheck.map
 -include ../Makefile.config
 
 INCLUDE_FILES=helpers ports ym2610
-OBJS=entrypoint bios-commands adpcm ym2610 stream timer nss-fm nss-adpcm nss-ssg fx-vibrato fx-slide volume
+OBJS=entrypoint bios-commands adpcm ym2610 stream timer nss-fm nss-adpcm nss-ssg fx-vibrato fx-slide fx-vol-slide volume
 LIB=nullsound.lib
 
 VERSION=@version@

diff --git a/nullsound/buffers.s b/nullsound/buffers.s
@@ -61,12 +61,33 @@ ssg_semitone_distance::
         .db     0x07, 0x07, 0x06, 0x06, 0x06, 0x05, 0x05, 0x05, 0x04, 0x04, 0x04, 0x04, 0x04, 0, 0, 0
         .db     0x04, 0x03, 0x03, 0x03, 0x03, 0x03, 0x02, 0x03, 0x02, 0x02, 0x02, 0x02, 0x02, 0, 0, 0
 
-;;; Sine precalc
+;;; Convert note flat representation to <octave,semitone> representation
 ;;; ------
-;;; a 64 bytes 2*Pi sign precalc encoded as  3-bit magnitude + 1-bit sign
-;;; This is used by the vibrato effect
+;;; Precalc for 8 octaves
+        .bndry  128
+note_to_octave_semitone::
+        .db     0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0a, 0x0b
+        .db     0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, 0x18, 0x19, 0x1a, 0x1b
+        .db     0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2a, 0x2b
+        .db     0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3a, 0x3b
+        .db     0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48, 0x49, 0x4a, 0x4b
+        .db     0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59, 0x5a, 0x5b
+        .db     0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6a, 0x6b
+        .db     0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7a, 0x7b
+
+;;; fixed-point sine precalc
+;;; ------
+;;; a 64 entries (128bytes) signed fixed point precalc of sin(x) for x in [0..2*pi],
+;;; encoded as [-1.0..1.0] as [s..iffffffff....]. This serves as a base increment
+;;; for vibrato displacement from 1..16, which yields a 9bits fixed point for
+;;; displacement of the current NSS note
+        .bndry  128
 sine::
-        .db      0,  0,  9, 10, 11, 11, 12, 13, 13, 14, 14, 15, 15, 15, 15, 15
-        .db     15, 15, 15, 15, 15, 15, 14, 14, 13, 13, 12, 11, 11, 10,  9,  0
-        .db      0,  0,  1,  2,  3,  3,  4,  5,  5,  6,  6,  7,  7,  7,  7,  7
-        .db      7,  7,  7,  7,  7,  7,  6,  6,  5,  5,  4,  3,  3,  2,  1,  0
+        .dw     0x0000, 0x0190, 0x0310, 0x04a0, 0x0610, 0x0780, 0x08e0, 0x0a20
+        .dw     0x0b50, 0x0c50, 0x0d40, 0x0e10, 0x0ec0, 0x0f40, 0x0fb0, 0x0fe0
+        .dw     0x1000, 0x0fe0, 0x0fb0, 0x0f40, 0x0ec0, 0x0e10, 0x0d40, 0x0c50
+        .dw     0x0b50, 0x0a20, 0x08e0, 0x0780, 0x0610, 0x04a0, 0x0310, 0x0190
+        .dw     0x8000, 0x8190, 0x8310, 0x84a0, 0x8610, 0x8780, 0x88e0, 0x8a20
+        .dw     0x8b50, 0x8c50, 0x8d40, 0x8e10, 0x8ec0, 0x8f40, 0x8fb0, 0x8fe0
+        .dw     0x9000, 0x8fe0, 0x8fb0, 0x8f40, 0x8ec0, 0x8e10, 0x8d40, 0x8c50
+        .dw     0x8b50, 0x8a20, 0x88e0, 0x8780, 0x8610, 0x84a0, 0x8310, 0x8190
diff --git a/nullsound/fx-slide.s b/nullsound/fx-slide.s
@@ -27,19 +27,20 @@
         .area  CODE
 
 
-;;; Enable slide effect for the current SSG channel
+;;; Enable slide effect for the current channel
 ;;; ------
-;;;   ix  : FM state for channel
+;;;   ix  : state for channel
 ;;;    a  : slide direction: 0 == up, 1 == down
 ;;; [ hl ]: speed (4bits) and depth (4bits)
 slide_init::
+        push    bc
+        push    de
+
         ;; b: slide direction (from a)
         ld      b, a
 
-        ;; slide fx on
-        ld      a, FX(ix)
-        set     1, a
-        ld      FX(ix), a
+        ;; enable slide FX
+        set     BIT_FX_SLIDE, FX(ix)
 
         ;; a: speed
         ld      a, (hl)
@@ -48,6 +49,7 @@ slide_init::
         rra
         rra
         and     #0xf
+
         ;; de: inc16 = speed / 8
         ld      d, a
         ld      e, #0
@@ -85,67 +87,86 @@ _post_depth_negate:
 
         inc     hl
 
+        ;; init semitone position, fixed point representation
+        ld      a, #0
+        ld      SLIDE_POS16(ix), a
+        ld      SLIDE_POS16+1(ix), a
+
+        ;; save target depth
+        ld      a, SLIDE_DEPTH(ix)
+        ld      SLIDE_END(ix), a
+
+        pop     de
+        pop     bc
+
         ret
 
 
-;;; Setup the end semitone for the currently configured slide
+;;; Enable slide effect for the current channel
 ;;; ------
-;;; ix : state for channel
-;;;  e : semitone to configure increments from
-slide_setup_increments::
+;;;   ix  : state for channel
+;;;    a  : slide direction: 0 == up, 1 == down
+;;; [ hl ]: speed (4bits) and depth (4bits)
+slide_pitch_init::
         push    bc
         push    de
 
-        ;; init semitone position, fixed point representation
-        ld      SLIDE_POS16+1(ix), e
-        ld      a, #0
-        ld      SLIDE_POS16(ix), a
-
-        ;; d: depth, negative if slide goes down
-        ld      d, SLIDE_DEPTH(ix)
+        ;; b: slide direction (from a)
+        ld      b, a
 
-        ;; c: note adjust. slide up: 4, slide down: -4
-        ld      c, #4
-        bit     7, d
-        jr      z, _post_inc_adjust
-        ld      c, #-4
-_post_inc_adjust:
+        ;; enable slide FX
+        set     BIT_FX_SLIDE, FX(ix)
 
-        ;; b: current octave
-        ld      a, SLIDE_POS16+1(ix)
-        and     #0xf0
-        ld      b, a
+        ;; a: speed
+        ld      a, (hl)
 
-        ;; e: target octave
-        ld      a, SLIDE_POS16+1(ix)
-        add     d
-        and     #0xf0
+        ;; de: inc16 = speed / 32
+        ld      d, a
+        ld      e, #0
+        srl     d
+        rr      e
+        srl     d
+        rr      e
+        srl     d
+        rr      e
+        srl     d
+        rr      e
+        srl     d
+        rr      e
+        ;; down: negate inc16
+        bit     0, b
+        jr      z, _post_inc16_negate2
+        ld      a, #0
+        sub     e
         ld      e, a
-
-        ;; if current and target octave differ, skip missing steps in the depth
-        ld      a, b
-        cp      e
-        jr      z, _post_depth_adjust
-        ld      a, d
-        add     c               ; slide up: 4, slide down: -4
+        ld      a, #0
+        sbc     d
         ld      d, a
-_post_depth_adjust:
+_post_inc16_negate2:
+        ld      SLIDE_INC16(ix), e
+        ld      SLIDE_INC16+1(ix), d
 
-        ;; a: current octave/note
-        ld      a, SLIDE_POS16+1(ix)
+        ;; depth
+        ld      a, #127
+        ;; down: negate depth
+        ;; we also need to go one seminote below, to account for the
+        ;; fractional parts of the slide.
+        bit     0, b
+        jr      z, _post_depth_negate2
+        neg
+        dec     a
+_post_depth_negate2:
+        ld      SLIDE_DEPTH(ix), a
 
-        ;; d: target octave/note
-        add     d
-        ld      d, a
+        inc     hl
 
-        ;; when target note is a missing step, adjust to the next note
-        and     #0xf
-        cp      #12
-        ld      a, d
-        jr      c, _post_target_note
-        add     c             ; slide up: 4, slide down: -4
-_post_target_note:
-        ;; save target octave/note
+        ;; init semitone position, fixed point representation
+        ld      a, #0
+        ld      SLIDE_POS16(ix), a
+        ld      SLIDE_POS16+1(ix), a
+
+        ;; save target depth
+        ld      a, SLIDE_DEPTH(ix)
         ld      SLIDE_END(ix), a
 
         pop     de
@@ -211,20 +232,17 @@ _post_sign_chk:
         ret
 
 
-;;; Increment current fixed point position in the semitone table and
-;;; stop effects when the target position is reached
+;;; Increment current fixed point displacement and
+;;; stop effects when the target displacement is reached
 ;;; ------
 ;;; IN:
 ;;;   ix : state for channel
 ;;;    c : slide direction: 0 == up, 1 == down
 ;;; OUT:
 ;;;    a : whether effect is finished (0: finished, 1: still running)
-;;;    d : when effect is finished, target semitone
+;;;    d : when effect is finished, target displacement
+;;; de modified
 eval_slide_step:
-        ;; ix: state for the current channel
-        push    hl
-        pop     ix
-
         ;; c: 0 slide up, 1 slide down
         ld      a, SLIDE_INC16+1(ix)
         rlc     a
@@ -234,7 +252,7 @@ eval_slide_step:
         ;; INC16 increment is 1/8 semitone (0x0020) * depth
         ;; negative for slide down
 
-        ;; add/sub increment to the current semitone POS16
+        ;; add/sub increment to the current semitone displacement POS16
         ;; e: fractional part
         ld      a, SLIDE_INC16(ix)
         add     SLIDE_POS16(ix)
@@ -243,23 +261,8 @@ eval_slide_step:
         ;; d: integer part
         ld      a, SLIDE_INC16+1(ix)
         adc     SLIDE_POS16+1(ix)
+        ld      SLIDE_POS16+1(ix), a
         ld      d, a
-        ;; do we need to skip missing steps in the note table
-        and     #0xf
-        cp      #0xc
-        jr      c, _post_skip
-        ld      a, d
-        ;; slide direction
-        bit     0, c
-        jr      z, _slide_dist_up
-        add     #-4
-        ld      d, a
-        jr      _post_skip
-_slide_dist_up:
-        add     #4
-        ld      d, a
-_post_skip:
-        ld      SLIDE_POS16+1(ix), d
 
         ;; have we reached the end of the slide?
         ;; slide up:   continue if cur < end
@@ -276,31 +279,18 @@ _slide_cp:
         jr      c, _slide_intermediate
 
         ;; slide is finished, stop effect
-        ld      (ix), #0
+        res     BIT_FX_SLIDE, FX(ix)
 
-        ;; d: clamp the last slide pos to the target semitone
+        ;; d: clamp the last slide pos to the target displacement
         ld      d, SLIDE_END(ix)
 
         ;; for slide down, we finish one note below the real target to play
-        ;; all ticks with fractional parts. Adjust the end note back if needed
+        ;; all ticks with fractional parts. Adjust the end displacement back if needed
         bit     0, c
         jr      z, _post_adjust
-        ld      a, d
-        and     #0xf
-        cp      #11
-        jr      c, _neg_inc_adjust
-        ;; adjust to next note (after the missing steps in the note table)
-        ld      a, d
-        add     #5
-        ld      d, a
-        jr      _post_adjust
-_neg_inc_adjust:
-        ;; adjust to next note
-        ld      a, d
-        inc     a
-        ld      d, a
+        inc     d
 _post_adjust:
-        ;; effect is finished, new semitone in d
+        ;; effect is finished, new displacement in d
         ld      a, #0
         ret