Cosmetics, doc updates, and optimizations.

README.md

@@ -99,8 +99,20 @@ the remaining code for playback.
 
 ### Timing
 
-Sample generation is a bit faster (I guess around 10-15%),
+#### Sample precalculation
-but most of the time is spent on `muls` operations, so this is the limiting factor.
+
+Sample generation is a faster than the original 1.0 player and also
+faster than the 1.5 player, which got a slightly better performance
+than the 1.0 one (compiler change?).
+
+According to my measurements on my set of Pretracker tunes,
+Raspberry Casket needs between 10% to 20% less instructions.
+Of these instructions, about 5% are `muls` operations and the new
+player is only able to shave off between 3% and 8% percent of those,
+so this is probably the limiting factor.
+
+#### Playback
+
 Raspberry Casket is about twice as fast as the old replayer for playback.
 
 Unfortunately, the replayer is still pretty slow and has high

src/raspberry_casket.asm

@@ -346,23 +346,23 @@ pre_PrepareCopperlist:
 ;********************************************************************
 pre_SongInit:
         IFNE    PRETRACKER_DONT_TRASH_REGS
-        movem.l d2/d7/a2-a5,-(sp)
+        movem.l d2-d7/a2-a5,-(sp)
         ENDC
         moveq.l #0,d0
-        move.l  $0000(a2),d1
+        movem.l (a2),d1/d3-d6                      ; song offsets $0000/$0004/$0008/$000c/$0010
         move.b  d1,d2
         move.b  d0,d1
         cmp.l   #$50525400,d1                   ; "PRE"-Text
         bne     .error
         moveq.l #MAX_INSTRUMENTS-1,d7           ; notice there's one extra name (available in 1.5, but not usable)!
         IFNE    PRETRACKER_SUPPORT_V1_5
-        cmpi.b  #$1e,d2
+        cmp.b   #$1e,d2
         bgt     .error
         bne.s   .nopatchv15
         move.l  $005c(a2),d0                    ; make song backward compatible
         ror.w   #8,d0
         move.l  d0,$003c(a2)
-        addq.l  #8,$0004(a2)                    ; skip over first pattern data
+        addq.l  #8,d3                           ; skip over first pattern data (offset $0004)
         moveq.l #2*MAX_INSTRUMENTS-1,d7         ; v1.5 has 32 slots (the other ones used for sfx)
 .nopatchv15
         ELSE
@@ -379,17 +379,13 @@ pre_SongInit:
         move.b  $003f(a2),sv_num_steps_b(a1)           ; number of steps!
         move.b  $0041(a2),sv_num_waves_b(a1)           ; number of instruments
 
-        move.l  $0004(a2),d0
+        add.l   a2,d3                           ; add to offset $0004
-        add.l   a2,d0
+        move.l  d3,sv_pos_data_adr(a1)          ; address to position data (POSD)
-        move.l  d0,sv_pos_data_adr(a1)          ; address to position data (POSD)
 
-        move.l  $0008(a2),d0
+        add.l   a2,d4                           ; add to offset $0008
-        add.l   a2,d0
+        move.l  d4,sv_patterns_ptr(a1)          ; address to pattern data (PATT)
-        move.l  d0,sv_patterns_ptr(a1)          ; address to pattern data (PATT)
-
-        move.l  $000c(a2),d0                    ; offset into instrument names
-        lea     (a2,d0.l),a0                    ; instrument names
 
+        lea     (a2,d5.l),a0                    ; offset (from $000c) into instrument names
 .instrnamesloop
         moveq.l #23-1,d0                        ; max 23 chars
 .inststrloop
@@ -466,8 +462,7 @@ pre_SongInit:
         dbra    d7,.instinfoloop
 
 .noinstsskip
-        move.l  $0010(a2),d0                    ; offset into wave names
+        lea     (a2,d6.l),a0                    ; offset (from $0010) into wave names
-        lea     (a2,d0.l),a0
         moveq.l #MAX_WAVES-1,d7
 .wavenamesloop
         moveq.l #23-1,d0                        ; max 23 chars
@@ -533,7 +528,7 @@ pre_SongInit:
 .error
 .exit
         IFNE    PRETRACKER_DONT_TRASH_REGS
-        movem.l (sp)+,d2/d7/a2-a5
+        movem.l (sp)+,d2-d7/a2-a5
         ENDC
         rts
 
@@ -845,7 +840,7 @@ pre_PlayerTick:
         bpl.s   .noclip_pat_vol_min
         moveq.l #0,d1
 .noclip_pat_vol_min
-        cmpi.b  #MAX_VOLUME,d1
+        cmp.b   #MAX_VOLUME,d1
         ble.s   .noclip_pat_vol_max
         moveq.l #MAX_VOLUME,d1
 .noclip_pat_vol_max
@@ -1165,7 +1160,7 @@ pre_PlayerTick:
 ; ----------------------------------------
 .pat_set_speed
         lea     pv_pat_speed_even_b(a4),a1
-        cmpi.b  #MAX_SPEED,d5
+        cmp.b   #MAX_SPEED,d5
         bhs.s   .pat_set_speed_shuffle
         move.b  d5,(a1)+            ; pv_pat_speed_even_b
         move.b  d5,(a1)+            ; pv_pat_speed_odd_b
@@ -1279,7 +1274,7 @@ pre_PlayerTick:
 
 ; ----------------------------------------
 .pat_set_volume
-        cmpi.b  #MAX_VOLUME,d5
+        cmp.b   #MAX_VOLUME,d5
         bls.s   .pat_set_volume_nomax
         moveq.l #MAX_VOLUME,d5
 .pat_set_volume_nomax
@@ -1445,7 +1440,7 @@ pre_PlayerTick:
 ;        IFNE    PRETRACKER_PARANOIA_MODE ; new step is never written
 ;        move.w  pcd_inst_new_step_w(a5),d1
 ;        blt.s   .inst_no_new_step_pos
-;        cmpi.w  #$20,d1
+;        cmp.w   #$20,d1
 ;        ble.s   .inst_good_new_step_pos
 ;        moveq.l #$20,d1
 ;.inst_good_new_step_pos
@@ -1792,7 +1787,7 @@ pre_PlayerTick:
         bsr     .inst_select_wave_subroutine
 
 .inst_wave_selected
-        cmpi.b  #$FF,d2
+        cmp.b   #$FF,d2
         beq.s   .inst_pat_loop_exit3
         subq.b  #1,d2
         move.b  d2,pcd_inst_line_ticks_b(a5)
@@ -1832,7 +1827,7 @@ pre_PlayerTick:
 .adsr_decay_and_release
         moveq.l #0,d4
         move.b  pcd_adsr_phase_speed_b(a5),d4
-        cmpi.b  #$8f,d4
+        cmp.b   #$8f,d4
         bhs.s   .adsr_absurd_slow_release
         move.b  d4,d5
         addq.b  #1,d5
@@ -1884,7 +1879,7 @@ pre_PlayerTick:
 
 .adsr_attack
         add.w   uii_adsr_attack(a2),d2
-        cmpi.w  #MAX_VOLUME<<4,d2
+        cmp.w   #MAX_VOLUME<<4,d2
         blt.s   .adsr_done
 
 .adsr_do_decay
@@ -2111,7 +2106,7 @@ pre_PlayerTick:
 ; select right sample corresponding to current pitch
 
         move.w  pcd_out_len_w(a5),d3
-        cmpi.w  #$219,d0
+        cmp.w   #$219,d0
         ble     .noclippitchhigh
         move.w  #$231,d6                    ; That's probably B-3+1, mapping to period $71 (although $7c is the last safe value)
         btst    #2,wi_flags_b(a3)
@@ -2184,7 +2179,7 @@ pre_PlayerTick:
         add.w   d1,d1
         add.w   d1,d1
         sub.w   d1,d0
-        cmpi.w  #$231,d0
+        cmp.w   #$231,d0
         ble.s   .noclippitchhigh
         move.w  #$231,d0
 .noclippitchhigh
@@ -2237,7 +2232,7 @@ pre_PlayerTick:
         moveq.l #MAX_TRACK_DELAY-1,d0   ; load from last buffer
 
         ; handle track delay
-        cmpi.b  #$FF,d3
+        cmp.b   #$FF,d3
         beq.s   .clear_track_delay
 
         ; advance and wrap offset

src/raspberry_casket_wavegen.asm

@@ -1194,8 +1194,7 @@ pre_WaveGen:
 ;--------------------------------------------------------------------
 ; a3: waveinfo
 ;
-; d6: wetness (word)
+; uses all data registers and a0-a1 (a2/a3 unchanged)
-; uses all data registers and a0-a2 (a3 unchanged)
 pre_Modulator:
         tst.b   wi_mod_wetness_b(a3)
         beq.s   .earlyexit
@@ -1240,21 +1239,21 @@ pre_Modulator:
         moveq.l #0,d3
 
 .innerloop
+        moveq.l #0,d1
         add.w   d7,d3
         addq.w  #8,d3
         smi     d1
         ext.w   d1
-        eor.w   d3,d1
+        eor.w   d3,d1               ; flip order if it was negative
         lsr.w   #6,d1               ; 4 bit key is bits 14 to 11, needs to be 8 to 5
         and.w   #15<<5,d1
-        ext.l   d1
 
         add.l   d5,d1
         lsr.l   #6,d1
 
         move.w  d2,d0
         sub.w   d1,d0
-        bmi.s   .isneg
+        bmi.s   .is_outside_sample
 
         move.b  (a0,d0.w),d1
         ext.w   d1
@@ -1269,7 +1268,7 @@ pre_Modulator:
 
         CLIPTO8BITAFTERADD d1
         move.b  d1,(a0,d2.w)
-.isneg
+.is_outside_sample
         addq.w  #1,d2
         cmp.w   d4,d2
         bcs.s   .innerloop