Index
> DOS > How to convert program to work with 32bit numbers ? |
| Author |
|
|
Matrix 26 Aug 2005, 22:35
hi,
first, change that je to jae then change registers to appropriate 32 bit ones, Code: format MZ entry codeseg:main segment dataseg segment codeseg LF EQU 0AH CR EQU 0DH main: mov ax,49 ; uses di, bx, cx, dx ; ; square root of AX via Newton-Raphson iteration. mov di,ax mov ax,13 start_loop: mov bx,ax xor dx,dx mov ax,di div bx add ax,bx shr ax,1 mov cx,ax sub cx,bx cmp cx,2 jae start_loop CALL print_ax_2 READCHAR: MOV AH,1 INT 21H RET ; exit PRINTCHAR:MOV AH,2 INT 21H RET PRINT_NEWLINE: MOV DL,CR CALL PRINTCHAR MOV DL,LF CALL PRINTCHAR RET ;*************************************************************** print_ax_1: ;proc near MOV CL,4 MOV DL,AL SHR DL,CL CALL PRINT_AX MOV DL,AL AND DL,0FH CALL PRINT_AX MOV DL,AH SHR DL,CL CALL PRINT_AX MOV DL,AH AND DL,0FH CALL PRINT_AX ret ;print_ax_1 endp print_ax_2: ; proc near MOV CL,4 MOV DL,AH SHR DL,CL CALL PRINT_AX MOV DL,AH AND DL,0FH CALL PRINT_AX MOV DL,AL SHR DL,CL CALL PRINT_AX MOV DL,AL AND DL,0FH CALL PRINT_AX ret ;print_ax_2 endp PRINT_AX: ; PROC NEAR PUSH AX ADD DL,30H CMP DL,'9' JBE OK ADD DL,7 OK: MOV AH,2 INT 21H POP AX RET ;PRINT_AX ENDP it will be the same basically, use eax instead of ax, ebx instead of bx, edi instead of di... |
|||
26 Aug 2005, 22:35 |
|
|
umen 27 Aug 2005, 10:58
Hello and thanks for your replay
but my masm dont support using eax , edi and so on... basically i like the program to except big numbers and perform on them the Newton square root . so as u see i can't implant your suggestion |
|||
|
27 Aug 2005, 10:58 |
|
|
El Tangas 27 Aug 2005, 17:42
If you cant use 32 bit registers, you'll have to use dx:ax to contain the 32 bit number, and make some changes to the calculations:
Code: 1. xor dx,dx mov ax,di div bx you have to remove the xor dx,dx, of course 2. add ax,bx you have to extend this to 32 bits: add ax,bx adc dx,0 3. shr ax,1 extend to 32 bits: shr dx,1 rcr ax,1 4. In the final comparison, you have to check also that dx is zero. And you may have to change other things to, but this is the most important. |
|||
|
27 Aug 2005, 17:42 |
|
|
umen 27 Aug 2005, 20:52
Hi
first of all tnx for the replay ... well i did changed my code as you suggested and i got Divide overflow -- when i tried to run the program .. whets going on here ? can some one help me here ? here is your changed i did in the code ------------------------------------------------------------------------------- LF EQU 0AH CR EQU 0DH DSEG SEGMENT DSEG ENDS SSEG SEGMENT STACK DW 100H DUP(?) SSEG ENDS CSEG SEGMENT ASSUME CS:CSEG,DS:DSEG,SS:SSEG BEGIN: ; uses di, bx, cx, dx ; ; square root of AX via Newton-Raphson iteration. mov di,ax mov ax,133 start_loop: mov bx,ax ;xor dx,dx mov ax,di div bx add ax,bx adc dx,0 shr ax,1 rcr ax,1 mov cx,ax sub cx,bx cmp cx,2 jae start_loop CALL PRINT_AX_2 READCHAR: MOV AH,1 INT 21H RET PRINTCHAR:MOV AH,2 INT 21H RET PRINT_NEWLINE: MOV DL,CR CALL PRINTCHAR MOV DL,LF CALL PRINTCHAR RET ;*************************************************************** print_ax_1 proc near MOV CL,4 MOV DL,AL SHR DL,CL CALL PRINT_AX MOV DL,AL AND DL,0FH CALL PRINT_AX MOV DL,AH SHR DL,CL CALL PRINT_AX MOV DL,AH AND DL,0FH CALL PRINT_AX ret print_ax_1 endp print_ax_2 proc near MOV CL,4 MOV DL,AH SHR DL,CL CALL PRINT_AX MOV DL,AH AND DL,0FH CALL PRINT_AX MOV DL,AL SHR DL,CL CALL PRINT_AX MOV DL,AL AND DL,0FH CALL PRINT_AX ret print_ax_2 endp PRINT_AX PROC NEAR PUSH AX ADD DL,30H CMP DL,'9' JBE OK ADD DL,7 OK: MOV AH,2 INT 21H POP AX RET PRINT_AX ENDP CSEG ENDS END BEGIN |
|||
|
27 Aug 2005, 20:52 |
|
|
El Tangas 28 Aug 2005, 19:02
Well, you have to know what you're doing, that is, you have to understand the Newton algorithm when applied to the square root and the workings of the div instruction. For the program to work, you have to preserve the original number. In the original code:
mov di,ax and restore it before the div: mov ax,di div bx The div operation works like this: the 32 bit number in dx:ax is divided by the div operand (bx in this case). The result is placed in ax, and the remainder in dx, so both registers are changed. You have to preserve and restore dx also: preserve: mov di,ax mov si,dx restore before the division: mov dx,si mov ax,di div bx maybe this will work. |
|||
|
28 Aug 2005, 19:02 |
|
|
umen 28 Aug 2005, 20:09
Hello ant tnx for the quick replay but i stiil get "Divide overflow" after the changes you gave me .
im dispread here ... can some one help please? thanks |
|||
|
28 Aug 2005, 20:09 |
|
|
Matrix 28 Aug 2005, 23:13
unfortunatly when the end product doesnt fit into 16 bits destiantion, ax, it gives this "Divide Overflow ..."
if you must use 16 bit assembler for 32 bit, then prefer using db s to extend to 32 bit like in turbo pascal, because it generates 286 code only. Code: db $66 xor ax,ax ; this means xor eax,eax db $66 stosw ; stosd db $66 mov ax, $faaa db $feee ; mov eax, $feeefaaa ... or simply complie with fasm and db it into your 286 assembler code. if you just want to get square root of a number, you can use other codes, or simply use fpu. |
|||
|
28 Aug 2005, 23:13 |
|
|
umen 29 Aug 2005, 19:22
Hello and tnx again
confused again .. cant i use 32 bit numbers ( big numbers ) with 2 registers ? in 16 masm ? yes i like to get square root of number to understand and learn from it. if i just could find working code example of such thing , im sure i could learn allot . |
|||
|
29 Aug 2005, 19:22 |
|
|
shoorick 30 Aug 2005, 08:18
of course, you can use some registers to work with long numbers, but you have to update higher parts depending on carry flag after each operation with lower part. but i cann't help you with division with this way (have no mathematical education
 )- i think, it is possible to find examples for this around i-net. a suggestion: if your pc is not 8086, place .386 (or .486 etc. - why hesitate?) at the start of source and use 32-bit registers and operations.
regards! |
|||
|
30 Aug 2005, 08:18 |
|
|
umen 30 Aug 2005, 19:21
hi , tnx for the replay what is "i-net" ?
and where i can find sources there? |
|||
|
30 Aug 2005, 19:21 |
|
|
UCM 30 Aug 2005, 19:21
internet
|
|||
|
30 Aug 2005, 19:21 |
|
|
shoorick 31 Aug 2005, 07:10
here is a program to play with double registers numbering. use 16-bit debugger to check how it works. (at least you have to have debug.exe in your system). you can change numbers to check and use calculator in hex mode to ensure if it works right.
_________________ UNICODE forever! |
|||||||||||
|
31 Aug 2005, 07:10 |
|
< Last Thread | Next Thread > |
Forum Rules:
|