******************************************************************** * * CyberMand_Calc_P * * calculation routines for CyberMand (PowerPC version) * * assemble with: * powerasm CyberMand_Calc_P.s to CyberMand_Calc.o * * link with smalldata and smallcode enabled * * © 1997 HAAGE&PARTNER Computer GmbH * ******************************************************************** include graphics/gfx.i ;// XREF / XDEF xdef Calculation_PPC xdef FX_PPC xdef ChunkyToPlanar_PPC xref ScrDepth xref AreaWidth xref AreaHeight xref LeftPos xref TopPos xref RightPos xref BottomPos xref Radius xref Iterations xref Precision xref ChunkyBuffer xref HiddenBitmap xref ActualBitmap xref ZoomIn xref ZoomOut xref Zoom xref MouseDX xref MouseDY xref Translation xref WindowMode xref TransTable xref P_SINGLE xref P_DOUBLE xref _SCREENHEIGHT xref _SCREENWIDTH xref _SCREENDEPTH ;;/ ;// Macros **** citf: Convert Integer to Floating Signed (Double Precision) * * parameters: * 1: destination FP-Register * 2: intermediate FP-Register (MUST hold the value $4330000080000000) * 3: source GP-Register * * Important: this macro does NOT work in Large-Data Model!! citf macro xoris trash,\3,$8000 sw trash,_CITF_TEMP+4 lf \1,_CITF_TEMP fsub \1,\1,\2 endm ;;/ section "CyberMand_Calc",code cpu POWERPC smalldata smallcode ;// Calculation (calculates one mandelbrot frame) ****************************************************************************** * * Calculation * * calculates one mandelbrot frame with the formula z=z^2+c * the results are written into the chunkybuffer (at the left/top corner) * * double AND single precision ****************************************************************************** Calculation_PPC prolog 256 push r31 la r31,TransTable lb r12,WindowMode lh r11,ScrDepth lf f10,_CITF_0 lw r6,ChunkyBuffer lf f2,LeftPos ;f0 = c real part fmr f9,f2 ;save c real part lf f3,TopPos ;f1 = c imaginary part lf f4,RightPos lf f5,BottomPos fsub f4,f4,f2 ;calc width of mandel area fsub f5,f5,f3 ;calc height of mandel area lh r4,AreaWidth lh r5,AreaHeight citf f6,f10,r4 ;convert AreaWidth to FPDP citf f7,f10,r5 ;convert AreaHeight to FPDP fdiv f4,f4,f6 ;calc delta x (real) fdiv f5,f5,f7 ;calc delta y (imaginary) lf f6,Radius ;read termination radius lh r7,Iterations lb r8,Precision cmpwi r8,P_DOUBLE beq .calcdouble ******************** single precision version .calcsingle mr r8,r4 .Sloop mr r4,r8 .Sloop2 mtctr r7 lf f0,_0 ;f0 = z real part fmr f1,f0 ;f1 = z imaginary part fsubs f12,f6,f2 ;f12 = RADIUS - c real .Sloop3 fmsubs f8,f1,f1,f2 ;f8 = (z imag)^2 - c real fadds f7,f0,f0 ;f7 = 2 * z real fmadds f11,f0,f0,f8 ;f11 = (z real)^2 + f8 fmsubs f0,f0,f0,f8 ;z[new] real = (z real)^2 - f8 fmadds f1,f1,f7,f3 ;z[new] imag = f7 * z imag + c imag fcmpu f11,f12 ;f11 > RADIUS - c real? bdnzt LT,.Sloop3 .Sexit mfctr r9 ;get iteration value blt .Sjump addi r9,r9,1 .Sjump slw r9,r9,r11 fadds f2,f2,f4 ;c[new] real = c[old] real + dx addi r10,r7,1 divw r9,r9,r10 ;calc colortable offset tstw r12 beq .nowin slwi r9,r9,2 add r9,r31,r9 lbz r9,3(r9) .nowin subic. r4,r4,1 ;decrease width counter stb r9,0(r6) ;and store the pixel addi r6,r6,1 ;postincrement bne .Sloop2 subic. r5,r5,1 ;decrease height counter fmr f2,f9 ;restore LeftPos fadds f3,f3,f5 ;c[new] imag = c[old] imag + dy bne .Sloop b .end ******************** double precision version .calcdouble mr r8,r4 .loop mr r4,r8 .loop2 mtctr r7 lf f0,_0 ;f0 = z real part fmr f1,f0 ;f1 = z imaginary part fsub f12,f6,f2 ;f12 = RADIUS - c real .loop3 fmsub f8,f1,f1,f2 ;f8 = (z imag)^2 - c real fadd f7,f0,f0 ;f7 = 2 * z real fmadd f11,f0,f0,f8 ;f11 = (z real)^2 + f8 fmsub f0,f0,f0,f8 ;z[new] real = (z real)^2 - f8 fmadd f1,f1,f7,f3 ;z[new] imag = f7 * z imag + c imag fcmpu f11,f12 ;f11 > RADIUS - c real? bdnzt LT,.loop3 .exit mfctr r9 ;get iteration value blt .jump addi r9,r9,1 .jump slw r9,r9,r11 fadd f2,f2,f4 ;c[new] real = c[old] real + dx addi r10,r7,1 divw r9,r9,r10 ;calc colortable offset tstw r12 beq .nowin2 slwi r9,r9,2 add r9,r31,r9 lbz r9,3(r9) .nowin2 subic. r4,r4,1 ;decrease width counter stb r9,0(r6) ;and store the pixel addi r6,r6,1 ;postincrement bne .loop2 subic. r5,r5,1 ;decrease height counter fmr f2,f9 ;restore LeftPos fadd f3,f3,f5 ;c[new] imag = c[old] imag + dy bne .loop .end pop r31 epilog ;;/ ;// FX (calculates zoom and translation effects) ****************************************************************************** * * FX * * calculates zoom and translation effects (only Double Precision) * ****************************************************************************** FX_PPC prolog 256 lf f6,_CITF_0 lb r4,ZoomIn mr. r4,r4 ;check if we should zoom in beq .nozoomin lb r4,Zoom ;read zoom value srwi r4,r4,1 ;half zoom value citf f4,f6,r4 ;convert to FPDP lf f0,LeftPos lf f1,RightPos fsub f2,f1,f0 ;calc area width fmul f2,f2,f4 lf f3,_100 fdiv f2,f2,f3 ;calc offset value fadd f0,f0,f2 ;increase LeftPos fsub f1,f1,f2 ;decrease RightPos sf f0,LeftPos ;and write back sf f1,RightPos lf f0,TopPos ;now completely the same lf f1,BottomPos ;for vertical direction fsub f2,f1,f0 fmul f2,f2,f4 fdiv f2,f2,f3 fadd f0,f0,f2 fsub f1,f1,f2 sf f0,TopPos sf f1,BottomPos b .mousemove .nozoomin lb r4,ZoomOut mr. r4,r4 ;check if we should zoom out beq .mousemove lb r4,Zoom li r5,100 mullw r6,r4,r5 subf r5,r4,r5 divw r4,r6,r5 ;calc zoom value srwi r4,r4,1 ;half zoom value citf f4,f6,r4 ;convert to FPDP lf f0,LeftPos lf f1,RightPos fsub f2,f1,f0 ;calc area width fmul f2,f2,f4 lf f3,_100 fdiv f2,f2,f3 ;calc offset value fsub f0,f0,f2 ;decrease LeftPos fadd f1,f1,f2 ;increase RightPos sf f0,LeftPos ;and write back sf f1,RightPos lf f0,TopPos ;now completely the same lf f1,BottomPos ;for vertical direction fsub f2,f1,f0 fmul f2,f2,f4 fdiv f2,f2,f3 fsub f0,f0,f2 fadd f1,f1,f2 sf f0,TopPos sf f1,BottomPos .mousemove lh r4,MouseDX extsh. r4,r4 beq .mousemoveY ;check if we should translate lf f0,LeftPos lf f1,RightPos fsub f2,f1,f0 ;calc area width citf f3,f6,r4 ;and convert to FPDP fmul f2,f2,f3 lh r7,AreaWidth citf f4,f6,r7 lb r8,Translation ;get speed citf f5,f6,r8 fmul f2,f2,f5 fdiv f2,f2,f4 ;calc translation value fadd f0,f0,f2 ;do translation fadd f1,f1,f2 sf f0,LeftPos ;and write back sf f1,RightPos .mousemoveY lh r4,MouseDY extsh. r4,r4 beq .end ;check if we should translate lf f0,TopPos lf f1,BottomPos fsub f2,f1,f0 ;calc area width citf f3,f6,r4 ;and convert to FPDP fmul f2,f2,f3 lh r7,AreaHeight citf f4,f6,r7 lb r8,Translation ;get speed citf f5,f6,r8 fmul f2,f2,f5 fdiv f2,f2,f4 ;calc translation value fadd f0,f0,f2 ;do translation fadd f1,f1,f2 sf f0,TopPos ;and write back sf f1,BottomPos .end epilog ;;/ ;// ChunkyToPlanar (converts chunky data to bitplane data) merge macro and r8,\1,\3 xor \1,r8,\1 and r9,\2,\3 xor \2,r9,\2 slwi r8,r8,\4 srwi \2,\2,\4 or \1,\1,\2 or \2,r9,r8 endm merge4 macro and r8,\1,\3 and \2,\2,\3 slwi r8,r8,4 or \2,\2,r8 endm ****************************************************************************** * * ChunkyToPlanar * * converts chunky data into bitplane data (using Bitmap Structure) * works ONLY with a depth of 4! * * original algorithm by James McCoull (I found the source somewhere * on the AMINET CD's) * ****************************************************************************** ChunkyToPlanar_PPC prolog 256 pushgpr r14-r30 li r29,0 lh r30,AreaHeight add r30,r29,r30 li r4,_SCREENDEPTH ;security check cmpwi r4,4 ;if depth <> 4 then exit bne .end liw r4,$00ff00ff ;load mask values liw r5,$0f0f0f0f liw r6,$33333333 liw r7,$55555555 lw r10,ChunkyBuffer lw r11,HiddenBitmap la r11,bm_Planes+16(r11) ;a1 -> after 4 plane ptrs lwzu r12,-4(r11) ;read 4 plane ptrs lwzu r13,-4(r11) lwzu r14,-4(r11) lwzu r15,-4(r11) lh r16,AreaHeight ;r16 = number of rows li r18,_SCREENHEIGHT subf r18,r16,r18 srwi r18,r18,1 ;r18 = starting row li r19,_SCREENWIDTH srwi r20,r19,3 ;r20 = screen width [bytes] mullw r18,r18,r20 ;r18 = vertical offset lh r17,AreaWidth ;r17 = number of columns subf r19,r17,r19 srwi r19,r19,4 ;r19 = horizontal offset add r18,r18,r19 ;r18 = offset to the 1st pixel add r19,r19,r19 ;r19 = modulo value srwi r17,r17,5 ;r17 = number of columns / 32 add r12,r12,r18 ;increase planeptrs add r13,r13,r18 add r14,r14,r18 add r15,r15,r18 subf r29,r29,r16 subf r30,r30,r16 .loop cmplw r16,r29 bgt .loopend cmplw r16,r30 ble .loopend mtctr r17 ;inner loop counter .loop2 lwz r21,0(r10) ;now the magic stuff... lwz r23,4(r10) ;first read chunky data lwz r20,8(r10) lwz r22,12(r10) lwz r24,2(r10) lwz r25,10(r10) lwz r26,6(r10) lwz r27,14(r10) lhz r18,16(r10) mh r21,r18 lhz r18,24(r10) mh r20,r18 lhz r18,20(r10) mh r23,r18 lhz r18,28(r10) mh r22,r18 lhz r18,18(r10) mh r24,r18 lhz r18,26(r10) mh r25,r18 lhz r18,22(r10) mh r26,r18 lhz r18,30(r10) mh r27,r18 addi r10,r10,32 merge r21,r20,r4,8 ;now convert it merge r23,r22,r4,8 merge4 r21,r23,r5 merge4 r20,r22,r5 merge r24,r25,r4,8 merge r26,r27,r4,8 merge4 r24,r26,r5 merge4 r25,r27,r5 merge r23,r26,r6,2 merge r22,r27,r6,2 merge r23,r22,r7,1 merge r26,r27,r7,1 stw r23,0(r12) ;and store it into the planes addi r12,r12,4 stw r22,0(r13) addi r13,r13,4 stw r26,0(r14) addi r14,r14,4 stw r27,0(r15) addi r15,r15,4 bdnz .loop2 b .cont .loopend slwi r21,r17,2 add r12,r12,r21 add r13,r13,r21 add r14,r14,r21 add r15,r15,r21 slwi r21,r21,3 add r10,r10,r21 .cont subic. r16,r16,1 add r12,r12,r19 ;get modulo value and add r13,r13,r19 ;increase planeptrs add r14,r14,r19 add r15,r15,r19 bne .loop .end popgpr epilog ;;/ section "CyberMand_Data",data dcb.b 32 _100 dc.d 100 _0 dc.d 0 _CITF_0 dc.l $43300000,$80000000 _CITF_TEMP dc.l $43300000,0 dy dc.d 0 dcb.b 32