print42_a

Hello All! Is it possible even faster, higher, stronger... without opening cycles? ╒═══< Begin file: print42a.asm >════════ PR42_Y DB 0 ;Y-print position (0-23) PR42_X DB 0 ;X print position (0-255) ;************************************************ ;print character [A] PR42_A exx : ld bc,(PR42_Y) ;24 ld d,FONT/256 : ld e,a ;11 ;35 ;--------------------------------------- ld a,c : and %00011000 ;11 or #40 : ld h,a ;11 ld a,b : and %11111000 ;11 ld l,a ;4 ld a,c : and %00000111 ;11 or l : rrca : rrca : rrca ;16 ld l,a ;4 ;--------------------------------------- ;68 ld a,b : ld b,8 : rrca ;15 and %00000011 : jr z,PR42_A0 ;14 ;137 dec a : jr z,PR42_A2 ;11 ;148 dec a : jr z,PR42_A4 ;11 ;159 ;--------------------------------------- ;154 PR42_A6 ld c,%00000011 ;7 PR42_A7 ld a,(de) : rlca : rlca ;15 xor (hl) : and c : xor (hl) ;18 ld (hl),a : inc l ;11 ld a,(de) ;7 rlca: rlca: rrd;26 ld (hl),a : dec l : inc h,d ;19 djnz PR42_A7 ;13 exx : ret ;14 ; ---- ;7+109*8-5+14 = 888 + 154 = 1042;--------------------------------------- PR42_A4 ld c,%11000000 ;7 PR42_A5 rrd : ld a,(de) ;25 rrca : rrca : rrca : rrca ;16 rld : inc l ;22 xor (hl) : and c : xor (hl) ;18 ld (hl),a : dec l : inc h,d ;19 djnz PR42_A5 ;13 exx : ret ;14 ; ---- ;7+113*8-5+14 = 920 + 159 = 1079 ;--------------------------------------- PR42_A2 ld c,%00111111 ;7 PR42_A3 ld a,(de) : rrca : rrca ;15 xor (hl) : and c : xor (hl) ;18 ld (hl),a : inc h,d ;15 djnz PR42_A3 ;13 exx : ret ;14 ; ---- ;7+61*8-5+14 = 504 + 148 = 652 ;--------------------------------------- PR42_A0 ld c,%11111100 ;7 PR42_A1 ld a,(de) ;7 xor (hl) : and c : xor (hl) ;18 ld (hl),a : inc h,d ;15 djnz PR42_A1 ;13 exx : ret ;14 ; ---- ;7+53*8-5+14 = 440 + 137 = 577 ;--------------------------------------- ;min: 577 ;max: 1079 ;byte: 114 ╘═══< End file: print42a.asm >══════════ Счастливо. Михаил.

Hello Mikhail! 11 Mar 01 21:21, Mihail Zharov -> All: MZ> Is it possible to go even faster, higher, stronger... MZ> without opening cycles? MZ> ╒═══< Begin file: print42a.asm >════════ MZ> PR42_Y DB 0 ;Y-print position (0-23) MZ> PR42_X DB 0 ;X-print position (0-255) MZ> ;************************************************ MZ> ;print character [A] MZ> PR42_A exx : ld bc,(PR42_Y) ;24 [skipped] Without even delving into the depth of the code: PR42_A EXX LD BC,0 PR42_Y EQU $-2 ;Y-print position (0-23) PR42_X EQU $-1 ;X-print position (0-255) Won 3 bytes, and won 10 clock cycles. Obviously, this is suitable when the printer is in RAM. Alexand

Hello Mikhail! 11 Mar 01 21:21, Mihail Zharov -> All: MZ> Is it possible even faster, higher, stronger... MZ> without cycle expansion? MZ> ╒═══< Begin file: print42a.asm >════════ MZ> PR42_Y DB 0 ;Y-print position (0-23) MZ> PR42_X DB 0 ;X-print position (0-255) The least significant bit of the X coordinate is ignored, ^^^^^ i.e. your signet is honorable :) [--skip--] MZ> PR42_A6 ld c,%00000011 ;7 MZ> PR42_A7 ld a,(de) : rlca : rlca ;15 MZ> xor (hl) : and c : xor (hl) ;18 MZ> ld (hl),a : incl ;11 MZ> ld a,(de) ;7 MZ> rlca : rlca : rrd ;26 MZ> ld (hl),a : dec l : inc h,d ;19 MZ> djnz PR42_A7 ;13 MZ> exx : ret ;14 MZ> ;--- MZ> ;7+109*8-5+14 = 888 This flow can be speeded up a little. Here: HL = address on screen DE = font address B=8 PR42_A6 LD A,(HL) ;07 RRCA : RRCA : LD C,A ;12LD A,(DE) ;07 RLCA : RL C : RLCA : RL C ;24 LD (HL),C : INC L : RRD ;27 LD (HL),A : DEC L : INC H,D ;19 DJNZ PR42_A6 ;13 EXX : RET ;14 ;--- ;109*8-5+14 = 881 A total of 7 bars, and for the scorpion - another 15, by reducing odd-clock instructions in the cycle. [--skip--] ...

Hello Mikhail! 11 Mar 01 21:21, Mihail Zharov -> All: MZ> Is it possible to go even faster, higher, stronger... MZ> without opening cycles? I watched it a little more :) [skipped] MZ> ld a,b : ld b,8 : rrca ;15 MZ> and %00000011 : jr z,PR42_A0 ;14 MZ> ;137 MZ> dec a : jr z,PR42_A2 ;11 MZ> ;148 MZ> dec a : jr z,PR42_A4 ;11 MZ> ;159 MZ> ;--------------------------------------- MZ> ;154 MZ> PR42_A6 ld c,%00000011 ;7 [skipped] This place for choosing the type of print can be optimized a little. From the first RRCA to the PR42_A marks? LD A,B: LD B,8 : RRCA RRCA: JR NC,PR42_A04 RRCA: JR C,PR42_A6 PR42_A2 ........ PR42_A04 RRCA: JR C, PR42_A4 PR42_A0 .......... PR42_A4 .......... PR42_A6 .......... In total, I ended up with one byte less. Now there are temporary schedules. We count the cycles spent from the moment after RRCA to the moment hits on print points depending on content [A] A old version new version 0 19 27 1 30 222 41 32 3 36 27 Since all 4 types of symbols occur equally, the average time to transition to the old one = 31.5 cycles, to the new one = 27 cycles. Total is one byte shorter and about 4.5 clock cycles faster ;-) Alexand

Hello, Alexander! Saturday 17 March 2001, at 04:17:01, Aleksandr Majorov -> Mikhail Zharov: AM>>> Since all 4 types of characters occur equally, then AM>>> average time for transition according to the old method = 31.5 cycles, AM>>> new = 27 cycles. AM>>> Total one byte shorter and about 4.5 clock cycles faster ;-) MZ>> I did the math and it turned out - you will gain 18 cycles when printing MZ>> 4 characters in a row... ;) AM> So everything is correct - 18 / 4 = 4.5 AM> But you counted for “in a row”, and I counted for “on average” - this AM> IMHO more convenient AM> :) Perhaps. But I wanted a more accurate result ;) Here is the final version(?) Here two more streams have been reduced by Pavel Vasiliev. Is there anything else that can be done? ;) ╒═══< Begin file: print42a.asm >════════ ;************************************************ ;PRINTING SYMBOL (A) IN 6x8 FONT - 42 ;************************************************ PR42_A exx ;4 ld bc,#0000 ;10 PR42_Y EQU $-2 ;Y-coordinate (0-023) PR42_X EQU $-1 ;X-coordinate (0-255) ld d,FONT/256 : ld e,a ;11 ;--------------------------------------- ld a,c : and %00011000 ;11 or #40 : ld h,a ;11 ld a,b : and %11111000 ;11 ld l,a ;4 ld a,c : and %00000111 ;11 or l : rrca : rrca : rrca ;16 ld l,a ;4 ;--------------------------------------- ld a,b : ld b,8 : rrca ;15 rrca : jr nc,PR42_A0 ;124 rrca : jr c,PR42_A6 ;135 ; ---- ; 130 ;--------------------------------------- PR42_A2 ld c,%00111111 ;7 PR42_A3 ld a,(de) : rrca : rrca ;15 xor (hl) : and c : xor (hl) ;18 ld (hl),a : inc h,d ;15 djnz PR42_A3 ;13 exx : ret ;14 ; ---- ;7+61*8-5+14 = 504 + 130 = 634 ;--------------------------------------- PR42_A0 rrca : jr c,PR42_A4 ;11 ld c,%00000011 ;7 ex de,hl ;4 PR42_A1 ld a,(de) ;7 and c : or (hl) ;11 ld (de),a : inc h,d ;15 djnz PR42_A1 ;13 exx : ret ;14 ; ---- ;22+46*8-5+14 = 399 + 124 = 523 ;--------------------------------------- PR42_A4 ld c,%11000000 ;7 PR42_A5 rrd : ld a,(de) ;25 rrca : rrca : rrca : rrca ;16 rld : inc l ;22 xor (hl) : and c : xor (hl) ;18 ld (hl),a : dec l : inc h,d ;19 djnz PR42_A5 ;13 exx : network ;14 ; ---- ;7+113*8-5+14 = 920 + 140 = 1060 ;------------------------------------ PR42_A6 ld a,(of) ;7 rlca : rlca : ld c,a ;12 xor (hl) : and %00000011 ;14 xor (hl) ;7 ld (hl),a : inc l ;11 ld a,c : rrd ;22 ld (hl),a : dec l : inc h,d ;19 djnz PR42_A6 ;13 exx : network ;14 ; ---- ;105*8-5+14 = 849 + 135 = 984 ;------------------------------------ ;min: 523 ;max: 1060 ;bytes: 110 ╘═══< End file: print42a.asm >══════════ Happily. Michael.