From 1a6d39d63fb4ebfcec296258fcba43d0f93aa81a Mon Sep 17 00:00:00 2001
From: Vince Weaver <vincent.weaver@maine.edu>
Date: Tue, 2 Jan 2024 16:20:43 -0500
Subject: [PATCH] bubble: optimize more

---
 graphics/hgr/bubble/Makefile         |  17 +-
 graphics/hgr/bubble/bubble.s         | 149 +++++----
 graphics/hgr/bubble/bubble_rolled.s  | 455 +++++++++++++++++++++++++
 graphics/hgr/bubble/hgr_clear_part.s | 480 +++++++++++++++++++++++++++
 graphics/hgr/bubble/sin_unrolled.s   |  57 ++++
 5 files changed, 1087 insertions(+), 71 deletions(-)
 create mode 100644 graphics/hgr/bubble/bubble_rolled.s
 create mode 100644 graphics/hgr/bubble/hgr_clear_part.s
 create mode 100644 graphics/hgr/bubble/sin_unrolled.s

diff --git a/graphics/hgr/bubble/Makefile b/graphics/hgr/bubble/Makefile
index 7138c662..9457d0b9 100644
--- a/graphics/hgr/bubble/Makefile
+++ b/graphics/hgr/bubble/Makefile
@@ -7,12 +7,13 @@ EMPTY_DISK = ../../../empty_disk
 
 all:	bubble.dsk make_table
 
-bubble.dsk:	HELLO BUBBLE.BAS BUBBLE BUBBLE_ORIG DIAMOND
+bubble.dsk:	HELLO BUBBLE.BAS BUBBLE BUBBLE_ORIG BUBBLE_ROLLED DIAMOND
 	cp $(EMPTY_DISK)/empty.dsk bubble.dsk
 	$(DOS33) -y bubble.dsk SAVE A HELLO
 	$(DOS33) -y bubble.dsk SAVE A BUBBLE.BAS
 	$(DOS33) -y bubble.dsk BSAVE -a 0x0C00 BUBBLE
 	$(DOS33) -y bubble.dsk BSAVE -a 0x0C00 BUBBLE_ORIG
+	$(DOS33) -y bubble.dsk BSAVE -a 0x0C00 BUBBLE_ROLLED
 	$(DOS33) -y bubble.dsk BSAVE -a 0x0C00 DIAMOND
 
 ###
@@ -25,7 +26,7 @@ HELLO:	hello.bas
 BUBBLE:	bubble.o
 	ld65 -o BUBBLE bubble.o -C $(LINKER_SCRIPTS)/apple2_c00.inc
 
-bubble.o:	bubble.s hgr_clear_screen.s
+bubble.o:	bubble.s hgr_clear_part.s sin_unrolled.s
 	ca65 -o bubble.o bubble.s -l bubble.lst
 
 ###
@@ -36,6 +37,15 @@ BUBBLE_ORIG:	bubble_orig.o
 bubble_orig.o:	bubble_orig.s
 	ca65 -o bubble_orig.o bubble_orig.s -l bubble_orig.lst
 
+###
+
+BUBBLE_ROLLED:	bubble_rolled.o
+	ld65 -o BUBBLE_ROLLED bubble_rolled.o -C $(LINKER_SCRIPTS)/apple2_c00.inc
+
+bubble_rolled.o:	bubble_rolled.s
+	ca65 -o bubble_rolled.o bubble_rolled.s -l bubble_rolled.lst
+
+
 
 ###
 
@@ -63,4 +73,5 @@ make_table.o:	make_table.c
 ###
 
 clean:	
-	rm -f *~ *.o *.lst BUBBLE.BAS HELLO BUBBLE DIAMOND BUBBLE_ORIG
+	rm -f *~ *.o *.lst BUBBLE.BAS HELLO BUBBLE \
+		DIAMOND BUBBLE_ORIG BUBBLE_ROLLED
diff --git a/graphics/hgr/bubble/bubble.s b/graphics/hgr/bubble/bubble.s
index f496ddf1..a8de6365 100644
--- a/graphics/hgr/bubble/bubble.s
+++ b/graphics/hgr/bubble/bubble.s
@@ -10,8 +10,8 @@
 
 
 ; after fast graphics
-;	D7E77 = 884343 = 1.1fps
-;	DD06E = ?? (made J countdown, why longer?)
+;	D7E77(??) = 884343 = 1.1fps
+;	DD06E = (made J countdown, why longer?)
 ;	DB584 = destructive U when plotting
 ;	D57A2 = rotate right instead of left for HPLOT *32 (U)
 ;	D1D53 = same byt for V
@@ -19,8 +19,12 @@
 ;	AB2FC = optimize sine, keep H value in accumulator = 1.4fps
 ;	A9A38 = optimize cosine slightly
 ;	A50BF = use lookup table for sine sign (takes 256 more bytes)
+;	9F673 = clear screen, only clear X region we use
+;	9DD73 = clear screen, only clear Y region we use
+;	906FE = inline/unroll the sines
 ;		TODO: inline/unroll sine/cosine calls
 
+
 ; soft-switches
 
 KEYPRESS	= $C000
@@ -89,6 +93,7 @@ bubble:
 	;=======================
 	; init graphics
 
+	jsr	HGR
 	jsr	HGR2
 
 	;=======================
@@ -112,15 +117,18 @@ next_frame:
 
 	;===========================
 	; "fast" clear screen
-	; inline to save 12 cycles
+	; FIXME: inline to save 12 cycles
 
-	jsr	hgr_clear_screen
+	jsr	hgr_clear_part
+
+	; FIXME: see value of X after clear
 
 	ldx	#0							; 2
 	stx	I							; 3
 
 outer_loop:
 
+	;===========================================================
 	; setup R*I to inner loop
 	;	save NUM*4 (128) cycles at expense of 11 cycles
 
@@ -128,6 +136,8 @@ outer_loop:
 	lda	rl,X							; 4
 	sta	rl_smc+1						; 4
 
+	; countdown NUM times
+
 	ldx	#NUM							; 2
 	stx	J							; 3
 
@@ -143,10 +153,17 @@ rl_smc:
 	lda	#0		; R*I					; 2
 	adc	XL							; 3
 	sta	RXL							; 3
+
+	; 8 cycles to always add 0 for high byte
+	; since we're also copying can't save time by branching if no carry
+	;	3 if no carry / 2+5=7 if carry for branch/check
+
 	lda	#0							; 2
 	adc	XH							; 3
 	sta	RXH							; 3
 
+no_rl_carry:
+
 	; fixed_add(i,0,vh,vl,&ivh,&ivl);
 
 	; precalc I+V for later use
@@ -161,17 +178,45 @@ rl_smc:
 
 	; U=SIN(I+V)+SIN(RR+X)
 
-	ldy	#0
-	jsr	sin
-	ldy	#2
-	jsr	sin
+;	ldy	#0
+	lda	IVL							; 3
+	sta	STEMP1L							; 3
+	lda	IVH							; 3
+
+;	jsr	sin
+.include "sin_unrolled.s"
+
+	lda	sin_table_low,X						; 4
+	sta	OUT1L							; 3
+	lda	sin_table_high,X					; 4
+	sta	OUT1H							; 3
+
+
+
+
+
+
+
+;	ldy	#2
+	lda	RXL							; 3
+	sta	STEMP1L							; 3
+	lda	RXH							; 3
+
+;	jsr	sin
+
+.include "sin_unrolled.s"
+
+;	lda	sin_table_low,X						; 4
+;	sta	OUT1L							; 3
+;	lda	sin_table_high,X					; 4
+;	sta	OUT1H							; 3
 
 	clc
 	lda	OUT1L
-	adc	OUT2L
+	adc	sin_table_low,X
 	sta	UL
 	lda	OUT1H
-	adc	OUT2H
+	adc	sin_table_high,X
 	sta	UH
 
 	; V=COS(I+V)+COS(RR+X)
@@ -269,23 +314,25 @@ rl_smc:
 ;	bpl	inner_loop
 done_j:
 
-	inc	I
-	lda	I
-	cmp	#NUM
-	beq	done_i
-	jmp	outer_loop
+	inc	I							; 5
+	lda	I							; 3
+	cmp	#NUM							; 2
+	beq	done_i							; 2/3
+	jmp	outer_loop						; 3
 done_i:
 
 	; t=t+(1.0/32.0);
 	;  1/2 1/4 1/8 1/16 | 1/32 1/64 1/128 1/256
 	;	$0x08
 
-	; TODO: is CLC necessary? (bcs=bge, bcc=blt)
-	;	carry always set here
+	; carry always set here as we got here from a BEQ
+	; 	(bcs=bge, bcc=blt)
 
-	clc								; 2
+
+;	clc								; 2
 	lda	TL							; 3
-	adc	#$8							; 2
+;	adc	#$8							; 2
+	adc	#$7	; really 8, carry always set			; 2
 	sta	TL							; 3
 	lda	#0							; 2
 	adc	TH							; 3
@@ -360,29 +407,14 @@ already_loaded:
 
 	; sl=fsinh[i];
 
-	; TODO: tradeoff size for speed by having lookup
+	; tradeoff size for speed by having lookup
 	;	table for sign bits
+	;	the sign lookup only saves like 2 cycles
 
-	lda	sin_table_low,X
-	sta	OUT1L,Y
-	lda	sin_table_high,X
-	sta	OUT1H,Y
-
-.if 0
-
-	lda	sin_lookup,X						; 4+
-	asl								; 2
+	lda	sin_table_low,X						; 4+
 	sta	OUT1L,Y							; 5
-
-	bcs	sin_negative						; 2/3
-sin_positive:
-	lda	#$0							; 2
-	beq	set_sin_sign			; bra			; 3
-sin_negative:
-	lda	#$FF							; 2
-set_sin_sign:
+	lda	sin_table_high,X					; 4+
 	sta	OUT1H,Y							; 5
-.endif
 
 	rts								; 6
 
@@ -402,36 +434,7 @@ cos:
 	jmp	already_loaded						; 3
 
 
-rl:
-.byte	$00,$06,$0C,$12,$19,$1F,$25,$2B
-.byte	$32,$38,$3E,$45,$4B,$51,$57,$5E
-.byte	$64,$6A,$71,$77,$7D,$83,$8A,$90
-.byte	$96,$9D,$A3,$A9,$AF,$B6,$BC,$C2
-
-.if 0
-sin_lookup:
-.byte	$00,$03,$06,$09,$0C,$0F,$12,$15,$18,$1C,$1F,$22,$25,$28,$2B,$2E
-.byte	$30,$33,$36,$39,$3C,$3F,$41,$44,$47,$49,$4C,$4E,$51,$53,$55,$58
-.byte	$5A,$5C,$5E,$60,$62,$64,$66,$68,$6A,$6C,$6D,$6F,$70,$72,$73,$74
-.byte	$76,$77,$78,$79,$7A,$7B,$7C,$7C,$7D,$7E,$7E,$7F,$7F,$7F,$7F,$7F
-.byte	$7F,$7F,$7F,$7F,$7F,$7F,$7E,$7E,$7D,$7C,$7C,$7B,$7A,$79,$78,$77
-.byte	$76,$75,$73,$72,$70,$6F,$6D,$6C,$6A,$68,$66,$64,$63,$61,$5E,$5C
-.byte	$5A,$58,$56,$53,$51,$4E,$4C,$49,$47,$44,$41,$3F,$3C,$39,$36,$34
-.byte	$31,$2E,$2B,$28,$25,$22,$1F,$1C,$19,$16,$12,$0F,$0C,$09,$06,$03
-.byte	$00,$FE,$FA,$F7,$F4,$F1,$EE,$EB,$E8,$E5,$E2,$DF,$DC,$D9,$D6,$D3
-.byte	$D0,$CD,$CA,$C7,$C4,$C2,$BF,$BC,$BA,$B7,$B4,$B2,$AF,$AD,$AB,$A8
-.byte	$A6,$A4,$A2,$A0,$9E,$9C,$9A,$98,$96,$95,$93,$91,$90,$8E,$8D,$8C
-.byte	$8A,$89,$88,$87,$86,$85,$84,$84,$83,$82,$82,$81,$81,$81,$81,$81
-.byte	$81,$81,$81,$81,$81,$81,$82,$82,$83,$84,$84,$85,$86,$87,$88,$89
-.byte	$8A,$8B,$8D,$8E,$8F,$91,$93,$94,$96,$98,$99,$9B,$9D,$9F,$A1,$A4
-.byte	$A6,$A8,$AA,$AD,$AF,$B1,$B4,$B6,$B9,$BC,$BE,$C1,$C4,$C7,$C9,$CC
-.byte	$CF,$D2,$D5,$D8,$DB,$DE,$E1,$E4,$E7,$EA,$ED,$F0,$F4,$F7,$FA,$FD
-.endif
-
-log_lookup:
-	.byte $81,$82,$84,$88,$90,$A0,$C0,$80
-
-.include "hgr_clear_screen.s"
+.include "hgr_clear_part.s"
 .include "hgr_table.s"
 
 .align $100
@@ -469,3 +472,13 @@ sin_table_high:
 	.byte	$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF
 	.byte	$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF
 	.byte	$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF
+
+rl:
+.byte	$00,$06,$0C,$12,$19,$1F,$25,$2B
+.byte	$32,$38,$3E,$45,$4B,$51,$57,$5E
+.byte	$64,$6A,$71,$77,$7D,$83,$8A,$90
+.byte	$96,$9D,$A3,$A9,$AF,$B6,$BC,$C2
+
+log_lookup:
+	.byte $81,$82,$84,$88,$90,$A0,$C0,$80
+
diff --git a/graphics/hgr/bubble/bubble_rolled.s b/graphics/hgr/bubble/bubble_rolled.s
new file mode 100644
index 00000000..5ad5dc8b
--- /dev/null
+++ b/graphics/hgr/bubble/bubble_rolled.s
@@ -0,0 +1,455 @@
+; bubble universe -- Apple II Hires
+
+; original = 612 bytes
+; clear screen:
+;	bkgnd0 = $44198 = 278936 cycles = max ~4fps
+;	new: $A616 = 42518 = max ~22fps
+; hplot
+;	hplot0 = ($14E-$15C) $14E = 334 * 1024 = 342016 = max ~3fps
+;	lookup = 46 * 1024 = 47104 = max ~21fps
+
+
+; after fast graphics
+;	D7E77(??) = 884343 = 1.1fps
+;	DD06E = (made J countdown, why longer?)
+;	DB584 = destructive U when plotting
+;	D57A2 = rotate right instead of left for HPLOT *32 (U)
+;	D1D53 = same byt for V
+;	C2679 = optimize sine, don't care about bottom byte in addition
+;	AB2FC = optimize sine, keep H value in accumulator = 1.4fps
+;	A9A38 = optimize cosine slightly
+;	A50BF = use lookup table for sine sign (takes 256 more bytes)
+;	9F673 = clear screen, only clear X region we use
+;	9DD73 = clear screen, only clear Y region we use
+;		TODO: inline/unroll sine/cosine calls
+
+
+; soft-switches
+
+KEYPRESS	= $C000
+KEYRESET	= $C010
+PAGE1		= $C054
+PAGE2		= $C055
+
+; ROM routines
+
+BKGND0		= $F3F4         ; clear current page to A
+HGR2		= $F3D8		; set hires page2 and clear $4000-$5fff
+HGR		= $F3E2		; set hires page1 and clear $2000-$3fff
+HPLOT0		= $F457		; plot at (Y,X), (A)
+HLINRL		= $F530		; line to (X,A), (Y)
+HCOLOR1		= $F6F0		; set HGR_COLOR to value in X
+COLORTBL	= $F6F6
+WAIT		= $FCA8		; delay 1/2(26+27A+5A^2) us
+
+; zero page
+
+GBASL		= $26
+GBASH		= $27
+
+
+HPLOTXL		= $90
+HPLOTXH		= $91
+HPLOTYL		= $92
+HPLOTYH		= $93
+IVL		= $94
+IVH		= $95
+RXL		= $96
+RXH		= $97
+OUT1L		= $98
+OUT1H		= $99
+OUT2L		= $9A
+OUT2H		= $9B
+STEMP1L		= $9C
+STEMP1H		= $9D
+STEMP2L		= $9E
+STEMP2H		= $9F
+
+I		= $D0
+J		= $D1
+XL		= $D4
+XH		= $D5
+VL		= $D6
+VH		= $D7
+TL		= $DA
+TH		= $DB
+UL		= $DC
+UH		= $DD
+
+HGR_PAGE	= $E6
+
+; const
+
+NUM		= 32
+
+bubble:
+
+	;========================
+	; setup lookup tables
+
+	jsr	hgr_make_tables
+
+	;=======================
+	; init graphics
+
+	jsr	HGR
+	jsr	HGR2
+
+	;=======================
+	; init variables
+
+	lda	#0
+	sta	XL
+	sta	XH
+	sta	VL
+	sta	VH
+	sta	TL
+	sta	TH
+
+	;=========================
+	;=========================
+	; main loop
+	;=========================
+	;=========================
+
+next_frame:
+
+	;===========================
+	; "fast" clear screen
+	; FIXME: inline to save 12 cycles
+
+	jsr	hgr_clear_part
+
+	; FIXME: see value of X after clear
+
+	ldx	#0							; 2
+	stx	I							; 3
+
+outer_loop:
+
+	;===========================================================
+	; setup R*I to inner loop
+	;	save NUM*4 (128) cycles at expense of 11 cycles
+
+	ldx	I							; 3
+	lda	rl,X							; 4
+	sta	rl_smc+1						; 4
+
+	; countdown NUM times
+
+	ldx	#NUM							; 2
+	stx	J							; 3
+
+inner_loop:
+
+	; fixed_add(rh[i],rl[i],xh,xl,&rxh,&rxl);
+	; note: rh is always 0
+
+	; pre-calc (R*I)+X for later use
+
+	clc								; 2
+rl_smc:
+	lda	#0		; R*I					; 2
+	adc	XL							; 3
+	sta	RXL							; 3
+
+	; 8 cycles to always add 0 for high byte
+	; since we're also copying can't save time by branching if no carry
+	;	3 if no carry / 2+5=7 if carry for branch/check
+
+	lda	#0							; 2
+	adc	XH							; 3
+	sta	RXH							; 3
+
+no_rl_carry:
+
+	; fixed_add(i,0,vh,vl,&ivh,&ivl);
+
+	; precalc I+V for later use
+	;	this is 8.8 fixed point so bottom byte of I is 0
+
+;	clc			; C should be 0 from prev		;
+	lda	VL							; 3
+	sta	IVL							; 3
+	lda	I							; 3
+	adc	VH							; 3
+	sta	IVH							; 3
+
+	; U=SIN(I+V)+SIN(RR+X)
+
+	ldy	#0
+	jsr	sin
+	ldy	#2
+	jsr	sin
+
+	clc
+	lda	OUT1L
+	adc	OUT2L
+	sta	UL
+	lda	OUT1H
+	adc	OUT2H
+	sta	UH
+
+	; V=COS(I+V)+COS(RR+X)
+
+	ldy	#0
+	jsr	cos
+	ldy	#2
+	jsr	cos
+
+	clc
+	lda	OUT1L
+	adc	OUT2L
+	sta	VL
+	lda	OUT1H
+	adc	OUT2H
+	sta	VH
+
+
+	; X=U+T
+	clc								; 2
+	lda	UL							; 3
+	adc	TL							; 3
+	sta	XL							; 3
+	lda	UH							; 3
+	adc	TH							; 3
+	sta	XH							; 3
+
+	; HPLOT 32*U+140,32*V+96
+
+	; U can be destroyed as we don't use it again?
+
+	; 01234567 89ABCDEF
+
+	; 56789ABC DEF00000
+
+	; we want 56789ABC, rotate right by 3 is two iterations faster?
+
+	lda	UL							; 3
+
+	lsr	UH							; 5
+	ror								; 2
+
+	lsr	UH							; 5
+	ror								; 2
+
+	lsr	UH							; 5
+	ror								; 2
+
+	clc								; 2
+	adc	#140							; 2
+	tax								; 2
+
+	; calculate Ypos
+
+	lda	VH
+	sta	HPLOTYL
+	lda	VL
+
+	lsr	HPLOTYL
+	ror
+
+	lsr	HPLOTYL
+	ror
+
+	lsr	HPLOTYL
+	ror
+
+	clc
+	adc	#96
+
+	; "fast" hplot, Xpos in X, Ypos in A
+
+	tay								; 2
+	lda	hposn_low,Y						; 4
+	sta	GBASL							; 3
+	clc								; 2
+	lda	hposn_high,Y						; 4
+	adc	HGR_PAGE						; 3
+	sta	GBASH							; 3
+; 21
+
+	ldy	div7_table,X						; 4
+
+	lda	mod7_table,X						; 4
+	tax								; 2
+; 31
+	lda	(GBASL),Y						; 5
+	ora	log_lookup,X						; 4
+	sta	(GBASL),Y						; 6
+; 46
+
+	dec	J							; 5
+	bmi	done_j							; 2/3
+	jmp	inner_loop						; 3
+;	bpl	inner_loop
+done_j:
+
+	inc	I							; 5
+	lda	I							; 3
+	cmp	#NUM							; 2
+	beq	done_i							; 2/3
+	jmp	outer_loop						; 3
+done_i:
+
+	; t=t+(1.0/32.0);
+	;  1/2 1/4 1/8 1/16 | 1/32 1/64 1/128 1/256
+	;	$0x08
+
+	; carry always set here as we got here from a BEQ
+	; 	(bcs=bge, bcc=blt)
+
+
+;	clc								; 2
+	lda	TL							; 3
+;	adc	#$8							; 2
+	adc	#$7	; really 8, carry always set			; 2
+	sta	TL							; 3
+	lda	#0							; 2
+	adc	TH							; 3
+	sta	TH							; 3
+
+end:
+	; flip pages
+
+	; if $20 (draw PAGE1) draw PAGE2, SHOW page1
+	; if $40 (draw PAGE2) draw PAGE1, SHOW page2
+
+	lda	HGR_PAGE
+	eor	#$60
+	sta	HGR_PAGE
+
+	cmp	#$40
+	bne	flip2
+flip1:
+	bit	PAGE1
+	jmp	next_frame
+flip2:
+	bit	PAGE2
+	jmp	next_frame
+
+
+
+
+	;=======================
+sin:
+
+	; / 6.28	is roughly the same as *0.16
+	;               = .5 .25 .125 .0625 .03125
+	; 1/6.28 = 0.16 =  0 0    1   0       1 0 0 0 = 0x28
+
+	; i=(i*0x28)>>8;
+
+	lda	IVL,Y		; note, uses absolute as no ZP equiv	; 4
+	sta	STEMP1L							; 3
+	lda	IVH,Y							; 4
+already_loaded:
+	; A has STEMP1H
+
+	; i2=i<<3;
+
+
+	asl	STEMP1L							; 5
+	rol								; 2
+	asl	STEMP1L							; 5
+	rol								; 2
+	asl	STEMP1L							; 5
+	rol								; 2
+
+	; i1=i<<5;
+
+	ldx	STEMP1L							; 3
+	stx	STEMP2L							; 3
+
+	sta	STEMP1H							; 3
+
+	asl	STEMP2L							; 5
+	rol								; 2
+	asl	STEMP2L							; 5
+	rol								; 2
+
+	; i=(i1+i2)>>8;
+
+	; We ignore the low byte as we don't need it
+	; possibly inaccurate as we don't clear carry?
+
+	adc	STEMP1H							; 2
+	tax								; 2
+
+	; sl=fsinh[i];
+
+	; tradeoff size for speed by having lookup
+	;	table for sign bits
+	;	the sign lookup only saves like 2 cycles
+
+	lda	sin_table_low,X						; 4+
+	sta	OUT1L,Y							; 5
+	lda	sin_table_high,X					; 4+
+	sta	OUT1H,Y							; 5
+
+	rts								; 6
+
+	;=============================
+cos:
+	; 1.57 is roughly 0x0192 in 8.8
+
+	clc								; 2
+	lda	IVL,Y							; 4
+	adc	#$92							; 2
+	sta	STEMP1L							; 3
+
+	lda	IVH,Y							; 4
+	adc	#1							; 2
+;	sta	STEMP1H							; 3
+
+	jmp	already_loaded						; 3
+
+
+.include "hgr_clear_part.s"
+.include "hgr_table.s"
+
+.align $100
+sin_table_low:
+	.byte	$00,$06,$0C,$12,$19,$1F,$25,$2B,$31,$38,$3E,$44,$4A,$50,$56,$5C
+	.byte	$61,$67,$6D,$73,$78,$7E,$83,$88,$8E,$93,$98,$9D,$A2,$A7,$AB,$B0
+	.byte	$B4,$B9,$BD,$C1,$C5,$C9,$CD,$D1,$D4,$D8,$DB,$DE,$E1,$E4,$E7,$E9
+	.byte	$EC,$EE,$F0,$F3,$F4,$F6,$F8,$F9,$FB,$FC,$FD,$FE,$FE,$FF,$FF,$FF
+	.byte	$FF,$FF,$FF,$FF,$FE,$FE,$FD,$FC,$FB,$F9,$F8,$F6,$F5,$F3,$F1,$EE
+	.byte	$EC,$EA,$E7,$E4,$E1,$DE,$DB,$D8,$D5,$D1,$CD,$C9,$C6,$C2,$BD,$B9
+	.byte	$B5,$B0,$AC,$A7,$A2,$9D,$98,$93,$8E,$89,$83,$7E,$78,$73,$6D,$68
+	.byte	$62,$5C,$56,$50,$4A,$44,$3E,$38,$32,$2C,$25,$1F,$19,$13,$0C,$06
+	.byte	$00,$FB,$F4,$EE,$E8,$E2,$DB,$D5,$CF,$C9,$C3,$BD,$B7,$B1,$AB,$A5
+	.byte	$9F,$99,$93,$8E,$88,$83,$7D,$78,$73,$6D,$68,$63,$5E,$5A,$55,$50
+	.byte	$4C,$47,$43,$3F,$3B,$37,$33,$30,$2C,$29,$25,$22,$1F,$1C,$19,$17
+	.byte	$14,$12,$10,$0E,$0C,$0A,$08,$07,$06,$04,$03,$02,$02,$01,$01,$01
+	.byte	$01,$01,$01,$01,$02,$02,$03,$04,$05,$07,$08,$0A,$0B,$0D,$0F,$11
+	.byte	$14,$16,$19,$1C,$1E,$21,$25,$28,$2B,$2F,$32,$36,$3A,$3E,$42,$47
+	.byte	$4B,$4F,$54,$59,$5E,$62,$67,$6C,$72,$77,$7C,$82,$87,$8D,$92,$98
+	.byte	$9E,$A4,$AA,$AF,$B5,$BB,$C2,$C8,$CE,$D4,$DA,$E0,$E7,$ED,$F3,$F9
+sin_table_high:
+	.byte	$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00
+	.byte	$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00
+	.byte	$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00
+	.byte	$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00
+	.byte	$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00
+	.byte	$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00
+	.byte	$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00
+	.byte	$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00,$00
+	.byte	$00,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF
+	.byte	$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF
+	.byte	$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF
+	.byte	$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF
+	.byte	$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF
+	.byte	$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF
+	.byte	$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF
+	.byte	$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF
+
+rl:
+.byte	$00,$06,$0C,$12,$19,$1F,$25,$2B
+.byte	$32,$38,$3E,$45,$4B,$51,$57,$5E
+.byte	$64,$6A,$71,$77,$7D,$83,$8A,$90
+.byte	$96,$9D,$A3,$A9,$AF,$B6,$BC,$C2
+
+log_lookup:
+	.byte $81,$82,$84,$88,$90,$A0,$C0,$80
+
diff --git a/graphics/hgr/bubble/hgr_clear_part.s b/graphics/hgr/bubble/hgr_clear_part.s
new file mode 100644
index 00000000..2ba63b10
--- /dev/null
+++ b/graphics/hgr/bubble/hgr_clear_part.s
@@ -0,0 +1,480 @@
+	; we only draw to
+	;	Xpositions 140+-64 ( 76-204, /7 = 10 - 30)
+	;	Ypositions 96+-64 = 32 - 160
+
+hgr_clear_part:
+
+	lda	#0			; color
+
+	ldy	HGR_PAGE
+	cpy	#$40
+	bne	hgr_page1_clearscreen
+	jmp	hgr_page2_clearscreen
+
+hgr_page1_clearscreen:
+
+	ldy	#10
+hgr_page1_cls_loop:
+.if 0
+	sta	$2000,Y		; 0
+	sta	$2400,Y		; 1
+	sta	$2800,Y		; 2
+	sta	$2C00,Y		; 3
+	sta	$3000,Y		; 4
+	sta	$3400,Y		; 5
+	sta	$3800,Y		; 6
+	sta	$3C00,Y		; 7
+
+	sta	$2080,Y		; 8
+	sta	$2480,Y		; 9
+	sta	$2880,Y		; 10
+	sta	$2C80,Y		; 11
+	sta	$3080,Y		; 12
+	sta	$3480,Y		; 13
+	sta	$3880,Y		; 14
+	sta	$3C80,Y		; 15
+
+	sta	$2100,Y		; 16
+	sta	$2500,Y		; 17
+	sta	$2900,Y		; 18
+	sta	$2D00,Y		; 19
+	sta	$3100,Y		; 20
+	sta	$3500,Y		; 21
+	sta	$3900,Y		; 22
+	sta	$3D00,Y		; 23
+
+	sta	$2180,Y		; 24
+	sta	$2580,Y		; 25
+	sta	$2980,Y		; 26
+	sta	$2D80,Y		; 27
+	sta	$3180,Y		; 28
+	sta	$3580,Y		; 29
+	sta	$3980,Y		; 30
+	sta	$3D80,Y		; 31
+.endif
+	sta	$2200,Y		; 32
+	sta	$2600,Y		; 33
+	sta	$2A00,Y		; 34
+	sta	$2E00,Y		; 35
+	sta	$3200,Y		; 36
+	sta	$3600,Y		; 37
+	sta	$3A00,Y		; 38
+	sta	$3E00,Y		; 39
+
+	sta	$2280,Y		; 40
+	sta	$2680,Y		; 41
+	sta	$2A80,Y		; 42
+	sta	$2E80,Y		; 43
+	sta	$3280,Y		; 44
+	sta	$3680,Y		; 45
+	sta	$3A80,Y		; 46
+	sta	$3E80,Y		; 47
+
+	sta	$2300,Y		; 48
+	sta	$2700,Y		; 49
+	sta	$2B00,Y		; 50
+	sta	$2F00,Y		; 51
+	sta	$3300,Y		; 52
+	sta	$3700,Y		; 53
+	sta	$3B00,Y		; 54
+	sta	$3F00,Y		; 55
+
+	sta	$2380,Y		; 56
+	sta	$2780,Y		; 57
+	sta	$2B80,Y		; 58
+	sta	$2F80,Y		; 59
+	sta	$3380,Y		; 60
+	sta	$3780,Y		; 61
+	sta	$3B80,Y		; 62
+	sta	$3F80,Y		; 63
+
+	;=======
+
+	sta	$2028,Y		; 64
+	sta	$2428,Y		; 1
+	sta	$2828,Y		; 2
+	sta	$2C28,Y		; 3
+	sta	$3028,Y		; 4
+	sta	$3428,Y		; 5
+	sta	$3828,Y		; 6
+	sta	$3C28,Y		; 7
+
+	sta	$20A8,Y		; 72
+	sta	$24A8,Y		; 9
+	sta	$28A8,Y		; 10
+	sta	$2CA8,Y		; 11
+	sta	$30A8,Y		; 12
+	sta	$34A8,Y		; 13
+	sta	$38A8,Y		; 14
+	sta	$3CA8,Y		; 15
+
+	sta	$2128,Y		; 80
+	sta	$2528,Y		; 17
+	sta	$2928,Y		; 18
+	sta	$2D28,Y		; 19
+	sta	$3128,Y		; 20
+	sta	$3528,Y		; 21
+	sta	$3928,Y		; 22
+	sta	$3D28,Y		; 23
+
+	sta	$21A8,Y		; 88
+	sta	$25A8,Y		; 25
+	sta	$29A8,Y		; 26
+	sta	$2DA8,Y		; 27
+	sta	$31A8,Y		; 28
+	sta	$35A8,Y		; 29
+	sta	$39A8,Y		; 30
+	sta	$3DA8,Y		; 31
+
+	sta	$2228,Y		; 96
+	sta	$2628,Y		; 33
+	sta	$2A28,Y		; 34
+	sta	$2E28,Y		; 35
+	sta	$3228,Y		; 36
+	sta	$3628,Y		; 37
+	sta	$3A28,Y		; 38
+	sta	$3E28,Y		; 39
+
+	sta	$22A8,Y		; 104
+	sta	$26A8,Y		; 41
+	sta	$2AA8,Y		; 42
+	sta	$2EA8,Y		; 43
+	sta	$32A8,Y		; 44
+	sta	$36A8,Y		; 45
+	sta	$3AA8,Y		; 46
+	sta	$3EA8,Y		; 47
+
+	sta	$2328,Y		; 112
+	sta	$2728,Y		; 49
+	sta	$2B28,Y		; 50
+	sta	$2F28,Y		; 51
+	sta	$3328,Y		; 52
+	sta	$3728,Y		; 53
+	sta	$3B28,Y		; 54
+	sta	$3F28,Y		; 55
+
+	sta	$23A8,Y		; 120
+	sta	$27A8,Y		; 57
+	sta	$2BA8,Y		; 58
+	sta	$2FA8,Y		; 59
+	sta	$33A8,Y		; 60
+	sta	$37A8,Y		; 61
+	sta	$3BA8,Y		; 62
+	sta	$3FA8,Y		; 63
+
+	;=========
+
+	sta	$2050,Y		; 128
+	sta	$2450,Y		; 1
+	sta	$2850,Y		; 2
+	sta	$2C50,Y		; 3
+	sta	$3050,Y		; 4
+	sta	$3450,Y		; 5
+	sta	$3850,Y		; 6
+	sta	$3C50,Y		; 7
+
+	sta	$20D0,Y		; 136
+	sta	$24D0,Y		; 9
+	sta	$28D0,Y		; 10
+	sta	$2CD0,Y		; 11
+	sta	$30D0,Y		; 12
+	sta	$34D0,Y		; 13
+	sta	$38D0,Y		; 14
+	sta	$3CD0,Y		; 15
+
+	sta	$2150,Y		; 144
+	sta	$2550,Y		; 17
+	sta	$2950,Y		; 18
+	sta	$2D50,Y		; 19
+	sta	$3150,Y		; 20
+	sta	$3550,Y		; 21
+	sta	$3950,Y		; 22
+	sta	$3D50,Y		; 23
+
+	sta	$21D0,Y		; 152
+	sta	$25D0,Y		; 25
+	sta	$29D0,Y		; 26
+	sta	$2DD0,Y		; 27
+	sta	$31D0,Y		; 28
+	sta	$35D0,Y		; 29
+	sta	$39D0,Y		; 30
+	sta	$3DD0,Y		; 31
+.if 0
+	sta	$2250,Y		; 160
+	sta	$2650,Y		; 33
+	sta	$2A50,Y		; 34
+	sta	$2E50,Y		; 35
+	sta	$3250,Y		; 36
+	sta	$3650,Y		; 37
+	sta	$3A50,Y		; 38
+	sta	$3E50,Y		; 39
+
+	sta	$22D0,Y		; 168
+	sta	$26D0,Y		; 41
+	sta	$2AD0,Y		; 42
+	sta	$2ED0,Y		; 43
+	sta	$32D0,Y		; 44
+	sta	$36D0,Y		; 45
+	sta	$3AD0,Y		; 46
+	sta	$3ED0,Y		; 47
+
+	sta	$2350,Y		; 176
+	sta	$2750,Y		; 49
+	sta	$2B50,Y		; 50
+	sta	$2F50,Y		; 51
+	sta	$3350,Y		; 52
+	sta	$3750,Y		; 53
+	sta	$3B50,Y		; 54
+	sta	$3F50,Y		; 55
+
+	sta	$23D0,Y		; 184
+	sta	$27D0,Y		; 57
+	sta	$2BD0,Y		; 58
+	sta	$2FD0,Y		; 59
+	sta	$33D0,Y		; 60
+	sta	$37D0,Y		; 61
+	sta	$3BD0,Y		; 62
+	sta	$3FD0,Y		; 63
+.endif
+	iny
+	cpy	#30
+	beq	hgr_page1_cls_done
+	jmp	hgr_page1_cls_loop
+
+hgr_page1_cls_done:
+
+	rts
+
+
+hgr_page2_clearscreen:
+
+	ldy	#10
+hgr_page2_cls_loop:
+.if 0
+	sta	$4000,Y		; 0
+	sta	$4400,Y		; 1
+	sta	$4800,Y		; 2
+	sta	$4C00,Y		; 3
+	sta	$5000,Y		; 4
+	sta	$5400,Y		; 5
+	sta	$5800,Y		; 6
+	sta	$5C00,Y		; 7
+
+	sta	$4080,Y		; 8
+	sta	$4480,Y		; 9
+	sta	$4880,Y		; 10
+	sta	$4C80,Y		; 11
+	sta	$5080,Y		; 12
+	sta	$5480,Y		; 13
+	sta	$5880,Y		; 14
+	sta	$5C80,Y		; 15
+
+	sta	$4100,Y		; 16
+	sta	$4500,Y		; 17
+	sta	$4900,Y		; 18
+	sta	$4D00,Y		; 19
+	sta	$5100,Y		; 20
+	sta	$5500,Y		; 21
+	sta	$5900,Y		; 22
+	sta	$5D00,Y		; 23
+
+	sta	$4180,Y		; 24
+	sta	$4580,Y		; 25
+	sta	$4980,Y		; 26
+	sta	$4D80,Y		; 27
+	sta	$5180,Y		; 28
+	sta	$5580,Y		; 29
+	sta	$5980,Y		; 30
+	sta	$5D80,Y		; 31
+.endif
+	sta	$4200,Y		; 32
+	sta	$4600,Y		; 33
+	sta	$4A00,Y		; 34
+	sta	$4E00,Y		; 35
+	sta	$5200,Y		; 36
+	sta	$5600,Y		; 37
+	sta	$5A00,Y		; 38
+	sta	$5E00,Y		; 39
+
+	sta	$4280,Y		; 40
+	sta	$4680,Y		; 41
+	sta	$4A80,Y		; 42
+	sta	$4E80,Y		; 43
+	sta	$5280,Y		; 44
+	sta	$5680,Y		; 45
+	sta	$5A80,Y		; 46
+	sta	$5E80,Y		; 47
+
+	sta	$4300,Y		; 48
+	sta	$4700,Y		; 49
+	sta	$4B00,Y		; 50
+	sta	$4F00,Y		; 51
+	sta	$5300,Y		; 52
+	sta	$5700,Y		; 53
+	sta	$5B00,Y		; 54
+	sta	$5F00,Y		; 55
+
+	sta	$4380,Y		; 56
+	sta	$4780,Y		; 57
+	sta	$4B80,Y		; 58
+	sta	$4F80,Y		; 59
+	sta	$5380,Y		; 60
+	sta	$5780,Y		; 61
+	sta	$5B80,Y		; 62
+	sta	$5F80,Y		; 63
+
+	;=======
+
+	sta	$4028,Y		; 64
+	sta	$4428,Y		; 1
+	sta	$4828,Y		; 2
+	sta	$4C28,Y		; 3
+	sta	$5028,Y		; 4
+	sta	$5428,Y		; 5
+	sta	$5828,Y		; 6
+	sta	$5C28,Y		; 7
+
+	sta	$40A8,Y		; 72
+	sta	$44A8,Y		; 9
+	sta	$48A8,Y		; 10
+	sta	$4CA8,Y		; 11
+	sta	$50A8,Y		; 12
+	sta	$54A8,Y		; 13
+	sta	$58A8,Y		; 14
+	sta	$5CA8,Y		; 15
+
+	sta	$4128,Y		; 80
+	sta	$4528,Y		; 17
+	sta	$4928,Y		; 18
+	sta	$4D28,Y		; 19
+	sta	$5128,Y		; 20
+	sta	$5528,Y		; 21
+	sta	$5928,Y		; 22
+	sta	$5D28,Y		; 23
+
+	sta	$41A8,Y		; 88
+	sta	$45A8,Y		; 25
+	sta	$49A8,Y		; 26
+	sta	$4DA8,Y		; 27
+	sta	$51A8,Y		; 28
+	sta	$55A8,Y		; 29
+	sta	$59A8,Y		; 30
+	sta	$5DA8,Y		; 31
+
+	sta	$4228,Y		; 96
+	sta	$4628,Y		; 33
+	sta	$4A28,Y		; 34
+	sta	$4E28,Y		; 35
+	sta	$5228,Y		; 36
+	sta	$5628,Y		; 37
+	sta	$5A28,Y		; 38
+	sta	$5E28,Y		; 39
+
+	sta	$42A8,Y		; 104
+	sta	$46A8,Y		; 41
+	sta	$4AA8,Y		; 42
+	sta	$4EA8,Y		; 43
+	sta	$52A8,Y		; 44
+	sta	$56A8,Y		; 45
+	sta	$5AA8,Y		; 46
+	sta	$5EA8,Y		; 47
+
+	sta	$4328,Y		; 112
+	sta	$4728,Y		; 49
+	sta	$4B28,Y		; 50
+	sta	$4F28,Y		; 51
+	sta	$5328,Y		; 52
+	sta	$5728,Y		; 53
+	sta	$5B28,Y		; 54
+	sta	$5F28,Y		; 55
+
+	sta	$43A8,Y		; 120
+	sta	$47A8,Y		; 57
+	sta	$4BA8,Y		; 58
+	sta	$4FA8,Y		; 59
+	sta	$53A8,Y		; 60
+	sta	$57A8,Y		; 61
+	sta	$5BA8,Y		; 62
+	sta	$5FA8,Y		; 63
+
+	;=========
+
+	sta	$4050,Y		; 128
+	sta	$4450,Y		; 1
+	sta	$4850,Y		; 2
+	sta	$4C50,Y		; 3
+	sta	$5050,Y		; 4
+	sta	$5450,Y		; 5
+	sta	$5850,Y		; 6
+	sta	$5C50,Y		; 7
+
+	sta	$40D0,Y		; 136
+	sta	$44D0,Y		; 9
+	sta	$48D0,Y		; 10
+	sta	$4CD0,Y		; 11
+	sta	$50D0,Y		; 12
+	sta	$54D0,Y		; 13
+	sta	$58D0,Y		; 14
+	sta	$5CD0,Y		; 15
+
+	sta	$4150,Y		; 144
+	sta	$4550,Y		; 17
+	sta	$4950,Y		; 18
+	sta	$4D50,Y		; 19
+	sta	$5150,Y		; 20
+	sta	$5550,Y		; 21
+	sta	$5950,Y		; 22
+	sta	$5D50,Y		; 23
+
+	sta	$41D0,Y		; 152
+	sta	$45D0,Y		; 25
+	sta	$49D0,Y		; 26
+	sta	$4DD0,Y		; 27
+	sta	$51D0,Y		; 28
+	sta	$55D0,Y		; 29
+	sta	$59D0,Y		; 30
+	sta	$5DD0,Y		; 31
+.if 0
+	sta	$4250,Y		; 160
+	sta	$4650,Y		; 33
+	sta	$4A50,Y		; 34
+	sta	$4E50,Y		; 35
+	sta	$5250,Y		; 36
+	sta	$5650,Y		; 37
+	sta	$5A50,Y		; 38
+	sta	$5E50,Y		; 39
+
+	sta	$42D0,Y		; 168
+	sta	$46D0,Y		; 41
+	sta	$4AD0,Y		; 42
+	sta	$4ED0,Y		; 43
+	sta	$52D0,Y		; 44
+	sta	$56D0,Y		; 45
+	sta	$5AD0,Y		; 46
+	sta	$5ED0,Y		; 47
+
+	sta	$4350,Y		; 176
+	sta	$4750,Y		; 49
+	sta	$4B50,Y		; 50
+	sta	$4F50,Y		; 51
+	sta	$5350,Y		; 52
+	sta	$5750,Y		; 53
+	sta	$5B50,Y		; 54
+	sta	$5F50,Y		; 55
+
+	sta	$43D0,Y		; 184
+	sta	$47D0,Y		; 57
+	sta	$4BD0,Y		; 58
+	sta	$4FD0,Y		; 59
+	sta	$53D0,Y		; 60
+	sta	$57D0,Y		; 61
+	sta	$5BD0,Y		; 62
+	sta	$5FD0,Y		; 63
+.endif
+	iny
+	cpy	#30
+	beq	done_hgr_page2_cls
+	jmp	hgr_page2_cls_loop
+done_hgr_page2_cls:
+	rts
+
diff --git a/graphics/hgr/bubble/sin_unrolled.s b/graphics/hgr/bubble/sin_unrolled.s
new file mode 100644
index 00000000..267a3baf
--- /dev/null
+++ b/graphics/hgr/bubble/sin_unrolled.s
@@ -0,0 +1,57 @@
+
+	;=======================
+;sin:
+
+	; / 6.28	is roughly the same as *0.16
+	;               = .5 .25 .125 .0625 .03125
+	; 1/6.28 = 0.16 =  0 0    1   0       1 0 0 0 = 0x28
+
+	; i=(i*0x28)>>8;
+
+;	lda	IVL,Y		; note, uses absolute as no ZP equiv	; 4
+;	sta	STEMP1L							; 3
+;	lda	IVH,Y							; 4
+
+	; A has STEMP1H
+
+	; i2=i<<3;
+
+
+	asl	STEMP1L							; 5
+	rol								; 2
+	asl	STEMP1L							; 5
+	rol								; 2
+	asl	STEMP1L							; 5
+	rol								; 2
+
+	; i1=i<<5;
+
+	ldx	STEMP1L							; 3
+	stx	STEMP2L							; 3
+
+	sta	STEMP1H							; 3
+
+	asl	STEMP2L							; 5
+	rol								; 2
+	asl	STEMP2L							; 5
+	rol								; 2
+
+	; i=(i1+i2)>>8;
+
+	; We ignore the low byte as we don't need it
+	; possibly inaccurate as we don't clear carry?
+
+	adc	STEMP1H							; 2
+	tax								; 2
+
+	; sl=fsinh[i];
+
+	; tradeoff size for speed by having lookup
+	;	table for sign bits
+	;	the sign lookup only saves like 2 cycles
+
+;	lda	sin_table_low,X						; 4+
+;	sta	OUT1L,Y							; 5
+;	lda	sin_table_high,X					; 4+
+;	sta	OUT1H,Y							; 5
+