Draw_VLc - Optimizations

In this section I would like to demonstrate possible steps to optimize vectorlist drawing.

The shown possible steps are not ment to be "complete", exclusively right or particulary engenious.

Each scenario of ones own coding base provides grounds for other or more specific optimization technics.

I will show different versions of printing a vectorlist, which in each version step will be slightly more optimized in specific ways.

The demonstration program(s) will draw a vectorlist which represents a "hut" like:

Note:
Below given cycle measurement may be different to you measurements. To have exactly the same measurements you must use exactly the same Vide configuration - which even I can't remember right now - but the relation of the values should be correct!

(The hut is pretty small, but as you will see, drawing a "small" list is easier to optimize - also in many games drawing many small objects is the predominant scenario)

Version 1 - 3164 cycles - basis

;***************************************************** ; DEFINE SECTION ;***************************************************** ; load vectrex bios routine definitions INCLUDE "VECTREX.I" ; vectrex function includes ;***************************************************** ; Variable / RAM SECTION ;***************************************************** ; insert your variables (RAM usage) in the BSS section ; user RAM starts at $c880 BSS ORG $c880 ; start of our ram space ;***************************************************** ; HEADER SECTION ;***************************************************** ; The cartridge ROM starts at address 0 CODE ORG 0 ; the first few bytes are mandatory, otherwise the BIOS will not load ; the ROM file, and will start MineStorm instead DB "g GCE 1998", $80 ; "g' is copyright sign DW music1 ; music from the rom DB $F8, $50, $20, -$80 ; hight, width, rel y, rel x (from 0,0) DB "DRAW LINES V1", $80 ; some game information, ending with $80 DB 0 ; end of game header ;***************************************************** ; CODE SECTION ;***************************************************** ; here the cartridge program starts off main: JSR Wait_Recal ; Vectrex BIOS recalibration JSR Intensity_5F ; Sets the intensity of the ; vector beam to $5f LDA #$b0 ; scalefactor STA VIA_t1_cnt_lo LDA #$00 ; position relative Y LDB #$00 ; position relative X JSR Moveto_d ; sets up VIA control register after a wait recal LDX #vData ; address of string JSR Draw_VLc ; Vectrex BIOS print routine BRA main ; and repeat forever ;***************************************************** ; DATA SECTION ;***************************************************** vData = VectorList VectorList: DB +$07 ; number of lines to draw DB +$00, +$06 ; draw to y, x DB +$07, +$00 ; draw to y, x DB +$05, -$03 ; draw to y, x DB -$05, -$03 ; draw to y, x DB -$07, +$00 ; draw to y, x DB +$07, +$06 ; draw to y, x DB +$00, -$06 ; draw to y, x DB -$07, +$06 ; draw to y, x

Version 2 - 1637 cycles - scale factor!

In the above version (for demonstration purposes) I used a large scale factor and small vector strengths. Using a small scale factor and large vector strength results in huge optimization effects.

The scale factor and strength changes are not 1:1 - you have to experiment to get the same "sizes" of the resulting lists.

In the example I changed the scale factor from $b0 to $06, and multiplied each (original) strength by 18.

;***************************************************** ; DEFINE SECTION ;***************************************************** ; load vectrex bios routine definitions INCLUDE "VECTREX.I" ; vectrex function includes ;***************************************************** ; Variable / RAM SECTION ;***************************************************** ; insert your variables (RAM usage) in the BSS section ; user RAM starts at $c880 BSS ORG $c880 ; start of our ram space ;***************************************************** ; HEADER SECTION ;***************************************************** ; The cartridge ROM starts at address 0 CODE ORG 0 ; the first few bytes are mandatory, otherwise the BIOS will not load ; the ROM file, and will start MineStorm instead DB "g GCE 1998", $80 ; "g' is copyright sign DW music1 ; music from the rom DB $F8, $50, $20, -$80 ; hight, width, rel y, rel x (from 0,0) DB "DRAW LINES V2", $80 ; some game information, ending with $80 DB 0 ; end of game header ;***************************************************** ; CODE SECTION ;***************************************************** ; here the cartridge program starts off main: JSR Wait_Recal ; Vectrex BIOS recalibration JSR Intensity_5F ; Sets the intensity of the ; vector beam to $5f LDA #$06 ; scalefactor STA VIA_t1_cnt_lo LDA #$00 ; position relative Y LDB #$00 ; position relative X JSR Moveto_d ; sets up VIA control register after a wait recal LDX #vData ; address of string JSR Draw_VLc ; Vectrex BIOS print routine BRA main ; and repeat forever ;***************************************************** ; DATA SECTION ;***************************************************** BLOW UP = 18 vData = VectorList VectorList: DB +$07 ; number of lines to draw DB +$00*BLOW_UP, +$06*BLOW_UP; draw to y, x DB +$07*BLOW_UP, +$00*BLOW_UP; draw to y, x DB +$05*BLOW_UP, -$03*BLOW_UP; draw to y, x DB -$05*BLOW_UP, -$03*BLOW_UP; draw to y, x DB -$07*BLOW_UP, +$00*BLOW_UP; draw to y, x DB +$07*BLOW_UP, +$06*BLOW_UP; draw to y, x DB +$00*BLOW_UP, -$06*BLOW_UP; draw to y, x DB -$07*BLOW_UP, +$06*BLOW_UP; draw to y, x

Version 3 - 1630 cycles - macro!

This version is a step "in between". Nothing really changed, only the call of "JSR Draw VLc" was replaced with a macro call of exactly the same routine.

;***************************************************** ; DEFINE SECTION ;***************************************************** ; load vectrex bios routine definitions INCLUDE "VECTREX.I" ; vectrex function includes DRAW_VLC macro direct $d0 ; following code uses d0 LDA ,X+ Draw_VL_a_macro? STA $C823 LDD ,X STA <VIA_port_a ;Send Y to A/D CLR <VIA_port_b ;Enable mux LEAX 2,X ;Point to next coordinate pair NOP ;Wait a moment INC <VIA_port_b ;Disable mux STB <VIA_port_a ;Send X to A/D LDD #$FF00 ;Shift reg=$FF (solid line), T1H=0 LF3ED\?: STA <VIA_shift_reg ;Put pattern in shift register STB <VIA_t1 cnt_hi ;Set T1H (scale factor?) LDD #$0040 ;B-reg = T1 interrupt bit LF3F4\?: BITB <VIA_int_flags ;Wait for T1 to time out BEQ LF3F4\? NOP ;Wait a moment more STA <VIA_shift_reg ;Clear shift register (blank output) LDA $C823 ;Decrement line count DECA BPL Draw_VL_a_macro\? ;Go back for more points JSR Check0Ref ;Reset zero reference if necessary endm ;***************************************************** ; Variable / RAM SECTION ;***************************************************** ; insert your variables (RAM usage) in the BSS section ; user RAM starts at $c880 BSS ORG $c880 ; start of our ram space ;***************************************************** ; HEADER SECTION ;***************************************************** ; The cartridge ROM starts at address 0 CODE ORG 0 ; the first few bytes are mandatory, otherwise the BIOS will not load ; the ROM file, and will start MineStorm instead DB "g GCE 1998", $80 ; "g' is copyright sign DW music1 ; music from the rom DB $F8, $50, $20, -$80 ; hight, width, rel y, rel x (from 0,0) DB "DRAW LINES V3", $80 ; some game information, ending with $80 DB 0 ; end of game header ;***************************************************** ; CODE SECTION ;***************************************************** ; here the cartridge program starts off main: JSR Wait_Recal ; Vectrex BIOS recalibration JSR Intensity_5F ; Sets the intensity of the ; vector beam to $5f LDA #$06 ; scalefactor STA VIA_t1_cnt_lo LDA #$00 ; position relative Y LDB #$00 ; position relative X JSR Moveto_d ; sets up VIA control register after a wait recal LDX #vData ; address of string DRAW_VLC ; Macro print routine BRA main ; and repeat forever ;***************************************************** ; DATA SECTION ;***************************************************** BLOW UP = 18 vData = VectorList VectorList: DB +$07 ; number of lines to draw DB +$00*BLOW_UP, +$06*BLOW_UP; draw to y, x DB +$07*BLOW_UP, +$00*BLOW_UP; draw to y, x DB +$05*BLOW_UP, -$03*BLOW_UP; draw to y, x DB -$05*BLOW_UP, -$03*BLOW_UP; draw to y, x DB -$07*BLOW_UP, +$00*BLOW_UP; draw to y, x DB +$07*BLOW_UP, +$06*BLOW_UP; draw to y, x DB +$00*BLOW_UP, -$06*BLOW_UP; draw to y, x DB -$07*BLOW_UP, +$06*BLOW_UP; draw to y, x

Version 4 - 1531 cycles - optimized macro!

;***************************************************** ; DEFINE SECTION ;***************************************************** ; load vectrex bios routine definitions INCLUDE "VECTREX.I" ; vectrex function includes DRAW_VLC macro direct $d0 ; following code uses d0 LDD ,X++ ; STA $C823 ; CLRA _DRAW_VLA\?: STD VIA_port_b LDB ,X+ INC VIA_port_b ;Disable mux STB VIA_port_a ;Send X to A/D DEC VIA_shift_reg ;Put pattern in shift register STA VIA_t1_cnt_hi ;enable timer 1 LDB ,X+ ; 6 DEC $C823 ; 7 BEQ DRAW_END\? ; 3 Go back for more points LDA #$40 ;B-reg = T1 interrupt bit LF_1\?: BITA VIA_int_flags ;Wait for T1 to time out BEQ LF_1\? ; CLRA ;Wait a moment more STA VIA_shift_reg ;Clear shift register (blank output) BRA DRAW_VLA\? ; 3 _DRAW_END\?: LDA #$40 ;B-reg = T1 interrupt bit LF_2\?: BITA VIA_int_flags ;Wait for T1 to time out BEQ LF_2\? ; CLRA ;Wait a moment more STA VIA_shift_reg ; one more round STD VIA_port_b LDB ,X+ INC VIA_port_b ;Disable mux STB VIA_port_a ;Send X to A/D DEC VIA_shift_reg ;Put pattern in shift register STA VIA_t1_cnt_hi ;enable timer 1 LDA #$40 ;B-reg = T1 interrupt bit LF_3?: BITA VIA_int_flags ;Wait for T1 to time out BEQ LF_3\? CLR VIA_shift_reg ;Clear shift register (blank output) endm ;***************************************************** ; Variable / RAM SECTION ;***************************************************** ; insert your variables (RAM usage) in the BSS section ; user RAM starts at $c880 BSS ORG $c880 ; start of our ram space ;***************************************************** ; HEADER SECTION ;***************************************************** ; The cartridge ROM starts at address 0 CODE ORG 0 ; the first few bytes are mandatory, otherwise the BIOS will not load ; the ROM file, and will start MineStorm instead DB "g GCE 1998", $80 ; "g' is copyright sign DW music1 ; music from the rom DB $F8, $50, $20, -$80 ; hight, width, rel y, rel x (from 0,0) DB "DRAW LINES V4", $80 ; some game information, ending with $80 DB 0 ; end of game header ;***************************************************** ; CODE SECTION ;***************************************************** ; here the cartridge program starts off main: JSR Wait_Recal ; Vectrex BIOS recalibration JSR Intensity_5F ; Sets the intensity of the ; vector beam to $5f LDA #$06 ; scalefactor STA VIA_t1_cnt_lo LDA #$00 ; position relative Y LDB #$00 ; position relative X JSR Moveto_d ; sets up VIA control register after a wait recal LDX #vData ; address of string DRAW_VLC ; Macro print routine BRA main ; and repeat forever ;***************************************************** ; DATA SECTION ;***************************************************** BLOW_UP = 18 vData = VectorList VectorList: DB +$07 ; number of lines to draw DB +$00*BLOW_UP, +$06*BLOW_UP; draw to y, x DB +$07*BLOW_UP, +$00*BLOW_UP; draw to y, x DB +$05*BLOW_UP, -$03*BLOW_UP; draw to y, x DB -$05*BLOW_UP, -$03*BLOW_UP; draw to y, x DB -$07*BLOW_UP, +$00*BLOW_UP; draw to y, x DB +$07*BLOW_UP, +$06*BLOW_UP; draw to y, x DB +$00*BLOW_UP, -$06*BLOW_UP; draw to y, x DB -$07*BLOW_UP, +$06*BLOW_UP; draw to y, x

Version 5 - 1416 cycles - loop unrolling and cycle specific!

The loop in the vector drawing was unrolled, and the actual drawing routine does not have a wait loop anymore, since we now that the length of the loop (= scalefactor) is 6. We can use the information, that we only draw for 6 cycles, to NOT wait for these 6 cycles to finish - we know how many cycles our instructions take.

To do the loop unrolling and some of the math involved (jump into a unrolled vector drawing routine) we changed the first parameter of the vectorlist to have a unrolled offset counter instead of the vectorlist length.

The maximum unrolled loop in this version is 11 vectors (maximum vectorlist size).

The size of the header (10) and one line (16) is manually computed, but could also be automatically gotten from address "math".

;***************************************************** ; DEFINE SECTION ;***************************************************** ; load vectrex bios routine definitions INCLUDE "VECTREX.I" ; vectrex function includes ONE_LINE_LENGTH EQU 16 LENGTH_OF_HEADER EQU (4+2+2+2) MAX_LINE_NUM EQU 23 ONE_LINE_DRAW macro ; cycles INC VIA_port_b ; [6] Disable mux STB VIA_port_a ; [6] Send X to A/D DEC VIA_shift_reg ; [6] Put pattern in shift register CLR VIA_t1_cnt_hi ; [4] enable timer 1 LDD ,X++ ; [8] load Y coordinate to A, X coordinate to B STA VIA_port_a ; [4] Send Y to A/D CLR VIA_port_b ; [4] enable mux, thus y integrators are set to Y CLR VIA_shift_reg ; [4] turn the lights off endm MY_SPRITE_DRAW_VLC_UNLOOP_SMALL macro LDD ,X++ ;(2) STA VIA_port_a ;(2) [4] Send Y to A/D CLR VIA_port_b ;(2) [4] enable mux, thus y integrators are set to Y JMP [i_jump] ;(4) JUMP_INTO_ALL\?: JUMP_INTO_11\?: ONE_LINE_DRAW JUMP_INTO_10\?: ONE_LINE_DRAW JUMP_INTO_09\?: ONE_LINE_DRAW JUMP_INTO_08\?: ONE_LINE_DRAW JUMP_INTO_07\?: ONE_LINE_DRAW JUMP_INTO_06\?: ONE_LINE_DRAW JUMP_INTO_05\?: ONE_LINE_DRAW JUMP_INTO_04\?: ONE_LINE_DRAW JUMP_INTO_03\?: ONE_LINE_DRAW JUMP_INTO_02\?: ONE_LINE_DRAW JUMP_INTO_01\?: INC VIA_port_b ; [6] Disable mux STB VIA_port_a ; [6] Send X to A/D DEC VIA_shift_reg ; [6] Put pattern in shift register CLR VIA_t1_cnt_hi ; [4] enable timer 1 NOP 3 endm ;***************************************************** ; Variable / RAM SECTION ;***************************************************** ; insert your variables (RAM usage) in the BSS section ; user RAM starts at $c880 BSS ORG $c880 ; start of our ram space ;***************************************************** ; HEADER SECTION ;***************************************************** ; The cartridge ROM starts at address 0 CODE ORG 0 ; the first few bytes are mandatory, otherwise the BIOS will not load ; the ROM file, and will start MineStorm instead DB "g GCE 1998", $80 ; "g' is copyright sign DW music1 ; music from the rom DB $F8, $50, $20, -$80 ; hight, width, rel y, rel x (from 0,0) DB "DRAW LINES V5", $80 ; some game information, ending with $80 DB 0 ; end of game header ;***************************************************** ; CODE SECTION ;***************************************************** ; here the cartridge program starts off main: JSR Wait_Recal ; Vectrex BIOS recalibration JSR Intensity_5F ; Sets the intensity of the ; vector beam to $5f LDA #$06 ; scalefactor STA VIA_t1_cnt_lo LDA #$00 ; position relative Y LDB #$00 ; position relative X JSR Moveto_d ; sets up VIA control register after a wait recal LDX #vData ; address of string LDD ,X++ ; load offset of vector list draw ADDD #(unloop_start_address0+LENGTH_OF_HEADER-12*ONE_LINE_LENGTH); 4=JMP [], 2=LDD ,X++ STD i_jump unloop_start_address0: MY_SPRITE_DRAW_VLC_UNLOOP_SMALL NOP 2 CLR VIA_shift_reg ; [4] turn the lights off BRA main ; and repeat forever ;***************************************************** ; DATA SECTION ;***************************************************** BLOW UP = 18 vData = VectorList VectorList: DW (MAX LINE NUM-(7+1))*ONE LINE LENGTH DB +$00*BLOW_UP, +$06*BLOW_UP; draw to y, x DB +$07*BLOW_UP, +$00*BLOW_UP; draw to y, x DB +$05*BLOW_UP, -$03*BLOW_UP; draw to y, x DB -$05*BLOW_UP, -$03*BLOW_UP; draw to y, x DB -$07*BLOW_UP, +$00*BLOW_UP; draw to y, x DB +$07*BLOW_UP, +$06*BLOW_UP; draw to y, x DB +$00*BLOW_UP, -$06*BLOW_UP; draw to y, x DB -$07*BLOW_UP, +$06*BLOW_UP; draw to y, x