diff --git a/Makefile b/Makefile index 711adcd..bd14c7d 100644 --- a/Makefile +++ b/Makefile @@ -3,7 +3,7 @@ all : mandel.xex mandel.xex : mandel.o tables.o atari-asm-xex.cfg - ld65 -C ./atari-asm-xex.cfg --mapfile mandel.map -o $@ mandel.o tables.o + ld65 -C ./atari-asm-xex.cfg -o $@ mandel.o tables.o %.o : %.s ca65 -o $@ $< @@ -15,6 +15,4 @@ clean : rm -f tables.s rm -f *.o rm -f *.xex - rm -f mandel.map - diff --git a/mandel.s b/mandel.s index b52f24a..6837e00 100644 --- a/mandel.s +++ b/mandel.s @@ -126,10 +126,6 @@ KEY_7 = 51 KEY_8 = 53 KEY_9 = 48 KEY_0 = 50 -KEY_PERIOD = 34 -KEY_E = 42 -KEY_X = 22 -KEY_Y = 43 .struct float48 exponent .byte @@ -407,13 +403,6 @@ elapsed_work: elapsed_digit: .byte 0 -input_col: - .byte 0 -input_row: - .byte 0 -input_max: - .byte 0 - ; 2 + 9 * byte cycles .macro add bytes, dest, arg1, arg2 clc ; 2 cyc @@ -461,7 +450,7 @@ input_max: sub 4, dest, arg1, arg2 .endmacro -; 3 + 5 * (bytes - 1) cycles +; 3 + 5 * bytes cycles .macro shl bytes, arg asl arg ; 3 cyc .repeat bytes-1, i @@ -469,23 +458,22 @@ input_max: .endrepeat .endmacro -; 8 cycles +; 13 cycles .macro shl16 arg shl 2, arg .endmacro -; 13 cycles +; 18 cycles .macro shl24 arg shl 3, arg .endmacro -; 18 cycles +; 23 cycles .macro shl32 arg shl 4, arg .endmacro ; 6 * bytes cycles -; 4 * bytes bytes .macro copy bytes, dest, arg .repeat bytes, byte ; 6 * bytes cycles lda arg + byte ; 3 cyc @@ -494,7 +482,6 @@ input_max: .endmacro ; 12 cycles -; 8 bytes .macro copy16 dest, arg copy 2, dest, arg .endmacro @@ -529,19 +516,17 @@ input_max: neg 4, arg .endmacro -; 11-27 + 18 * shift cycles -; 65-81 cycles for shift=3 +; 11-27 + 23 * shift cycles +; 103-119 cycles for shift=4 .macro shift_round_16 arg, shift .repeat shift - shl32 arg ; 18 cycles + shl32 arg ; 23 cycles .endrepeat round16 arg ; 11-27 cycles .endmacro ; input: arg1, arg2 as fixed4.12 ; output: dest as fixed8.24 -; patch point jsr at 16 bytes in -imul16_patch_offset = 16 .macro imul16 dest, arg1, arg2 copy16 FR0, arg1 ; 12 cyc copy16 FR1, arg2 ; 12 cyc @@ -551,8 +536,6 @@ imul16_patch_offset = 16 ; input: arg as fixed4.12 ; output: dest as fixed8.24 -; patch point jsr at 8 bytes in -sqr16_patch_offset = 8 .macro sqr16 dest, arg copy16 FR0, arg ; 12 cyc jsr sqr16_func ; ? cyc @@ -698,6 +681,71 @@ bank_switch_table: .endif .endmacro +.proc imul8xe_init + + bank_switch 0 + lda #0 + sta EXTENDED_RAM + bank_switch 1 + lda #1 + sta EXTENDED_RAM + bank_switch 0 + lda EXTENDED_RAM + beq init + + ; no bank switching available, we just overwrite the value in base ram + rts + +init: + + ; patch imul16_func into a forwarding thunk to imul16xe_func + lda #$4c ; 'jmp' opcode + sta imul16_func + lda #.lobyte(imul16xe_func) + sta imul16_func + 1 + lda #.hibyte(imul16xe_func) + sta imul16_func + 2 + + ; ditto for sqr16_func -> sqr16xe_func + lda #$4c ; 'jmp' opcode + sta sqr16_func + lda #.lobyte(sqr16xe_func) + sta sqr16_func + 1 + lda #.hibyte(sqr16xe_func) + sta sqr16_func + 2 + + ; create the lookup table + ; go through the input set, in four 16KB chunks + + arg1 = FR1 + arg2 = FR2 + result = FR0 + + lda #$00 + sta arg1 + sta arg2 + sta ptr + lda #$40 + sta ptr + 1 + + ; $00 * $00 -> $3f * $ff + bank_switch 0 + jsr imul8xe_init_section + + ; $40 * $00 -> $7f * $ff + bank_switch 1 + jsr imul8xe_init_section + + ; $80 * $00 -> $bf * $ff + bank_switch 2 + jsr imul8xe_init_section + + ; $c0 * $00 -> $ff * $ff + bank_switch 3 + jsr imul8xe_init_section + + rts +.endproc ; Initialize a 16 KB chunk of the table ; input: multipliers in temp @@ -935,17 +983,6 @@ common: .endproc -; rounds to 16-bit first! -; input in FR0, 32 bits signed 6.26 fixed -; output in FR0, Atari float -; clobbers a, x, y, FR0, FR1 -.proc fixed6_26_to_float - shift_round_16 FR0, 3 - copy16 FR0, FR0 + 2 - jsr fixed3_13_to_float - rts -.endproc - ; input in FR0, Atari float ; output in FR0, 16 bits signed 3.13 fixed ; clobbers a, x, y, FR0, FR1 @@ -1097,15 +1134,12 @@ keep_going: shift_round_16 zy, 3 ; zx_2 = zx * zx -fixup_sqr16_1: sqr16 zx_2, zx + 2 ; zy_2 = zy * zy -fixup_sqr16_2: sqr16 zy_2, zy + 2 ; zx_zy = zx * zy -fixup_imul16_1: imul16 zx_zy, zx + 2, zy + 2 ; dist = zx_2 + zy_2 @@ -1569,7 +1603,7 @@ number_keys: beq five cpy #KEY_6 beq six - jmp letter_keys + jmp skip_char one: ldx #0 @@ -1588,21 +1622,7 @@ five: jmp load_key_viewport six: ldx #5 - jmp load_key_viewport - -letter_keys: - cpy #KEY_X - bne not_x - jsr input_x - jmp done -not_x: - cpy #KEY_Y - bne not_y - jsr input_y - jmp done -not_y: - jmp skip_char - + ; fall through load_key_viewport: jsr load_viewport ; fall through @@ -1612,23 +1632,6 @@ done: .endproc -.proc input_x - ldx #col_x - ldy #1 - jsr input_number - - - rts -.endproc - -.proc input_y - rts -.endproc - -.proc input_number - rts -.endproc - .proc clear_screen ; zero the range from framebuffer_top to display_list lda #.lobyte(framebuffer_top) @@ -1676,7 +1679,9 @@ zero_byte_loop: draw_string_const str_x copy32 FR0, ox - jsr fixed6_26_to_float + shift_round_16 FR0, 3 + copy16 FR0, FR0 + 2 + jsr fixed3_13_to_float jsr FASC jsr draw_string @@ -1685,7 +1690,9 @@ zero_byte_loop: draw_string_const str_y copy32 FR0, oy - jsr fixed6_26_to_float + shift_round_16 FR0, 3 + copy16 FR0, FR0 + 2 + jsr fixed3_13_to_float jsr FASC jsr draw_string @@ -1984,25 +1991,51 @@ update_status: lda FR0 + 1 sta elapsed_work + 1 - draw_string_const str_space - - .macro do_countdown divisor, digits - ldx #.lobyte(divisor) - ldy #.hibyte(divisor) - lda #.lobyte(digits) - sta INBUFF - lda #.hibyte(digits) - sta INBUFF + 1 - jsr countdown + ;jsr IFP + ;jsr FASC + ;jsr draw_string + + .macro countdown divisor, digits + .scope + ; count the hours + ldx #0 + countdown_loop: + lda elapsed_work + 1 + cmp #.hibyte(divisor) + bcc countdown_done + lda elapsed_work + cmp #.lobyte(divisor) + bcc countdown_done + sec + lda elapsed_work + sbc #.lobyte(divisor) + sta elapsed_work + lda elapsed_work + 1 + sbc #.hibyte(divisor) + sta elapsed_work + 1 + inx + jmp countdown_loop + countdown_done: + lda digits,x + eor #$80 + sta elapsed_digit + lda #.lobyte(elapsed_digit) + sta INBUFF + lda #.hibyte(elapsed_digit) + sta INBUFF + 1 + jsr draw_string + .endscope .endmacro - do_countdown 36000, digits_space - do_countdown 3600, digits_zero + + draw_string_const str_space + countdown 36000, digits_space + countdown 3600, digits_zero draw_string_const str_h - do_countdown 600, digits_zero - do_countdown 60, digits_zero + countdown 600, digits_zero + countdown 60, digits_zero draw_string_const str_m - do_countdown 10, digits_zero - do_countdown 1, digits_zero + countdown 10, digits_zero + countdown 1, digits_zero draw_string_const str_s skipped: @@ -2064,118 +2097,3 @@ loop: jmp main_loop .endproc - -; digit string in INBUFF -; divisor X/Y -; clobbers temp, calls draw_string -.proc countdown - divisor = temp - stx divisor - sty divisor + 1 - - ; count the hours - ldy #0 -countdown_loop: - lda elapsed_work + 1 - cmp divisor + 1 - beq countdown_lobyte - bcc countdown_done - bcs countdown_inc -countdown_lobyte: - lda elapsed_work - cmp divisor - bcc countdown_done -countdown_inc: - sec - lda elapsed_work - sbc divisor - sta elapsed_work - lda elapsed_work + 1 - sbc divisor + 1 - sta elapsed_work + 1 - iny - jmp countdown_loop -countdown_done: - lda (INBUFF),y - eor #$80 - sta elapsed_digit - lda #.lobyte(elapsed_digit) - sta INBUFF - lda #.hibyte(elapsed_digit) - sta INBUFF + 1 - jsr draw_string - rts -.endproc - -.proc imul8xe_init - - bank_switch 0 - lda #0 - sta EXTENDED_RAM - bank_switch 1 - lda #1 - sta EXTENDED_RAM - bank_switch 0 - lda EXTENDED_RAM - beq init - - ; no bank switching available, we just overwrite the value in base ram - rts - -init: - - ; patch imul16_func into a forwarding thunk to imul16xe_func - lda #$4c ; 'jmp' opcode - sta imul16_func - lda #.lobyte(imul16xe_func) - sta imul16_func + 1 - sta mandelbrot::fixup_imul16_1 + imul16_patch_offset + 1 - lda #.hibyte(imul16xe_func) - sta imul16_func + 2 - sta mandelbrot::fixup_imul16_1 + imul16_patch_offset + 2 - - ; ditto for sqr16_func -> sqr16xe_func - lda #$4c ; 'jmp' opcode - sta sqr16_func - lda #.lobyte(sqr16xe_func) - sta sqr16_func + 1 - sta mandelbrot::fixup_sqr16_1 + sqr16_patch_offset + 1 - sta mandelbrot::fixup_sqr16_2 + sqr16_patch_offset + 1 - lda #.hibyte(sqr16xe_func) - sta sqr16_func + 2 - sta mandelbrot::fixup_sqr16_1 + sqr16_patch_offset + 2 - sta mandelbrot::fixup_sqr16_2 + sqr16_patch_offset + 2 - - - ; create the lookup table - ; go through the input set, in four 16KB chunks - - arg1 = FR1 - arg2 = FR2 - result = FR0 - - lda #$00 - sta arg1 - sta arg2 - sta ptr - lda #$40 - sta ptr + 1 - - ; $00 * $00 -> $3f * $ff - bank_switch 0 - jsr imul8xe_init_section - - ; $40 * $00 -> $7f * $ff - bank_switch 1 - jsr imul8xe_init_section - - ; $80 * $00 -> $bf * $ff - bank_switch 2 - jsr imul8xe_init_section - - ; $c0 * $00 -> $ff * $ff - bank_switch 3 - jsr imul8xe_init_section - - rts -.endproc