DN5@PSUVM.BITNET (02/03/89)
Hi netlanders:
I am working on a DA which needs some random numbers. My problem is that
I always get the same sequence. I have tried setting randSeed, but this
doesn't seem to do anything (actually I set it before I call Random, then
save this value, then reset the original value, so that it won't mess up
an application). However, anything I do (including attempting a FOR loop
repeating for a somewhat random time (GetTimeDate + CountTicks), seems to
do nothing: I still get the same sequence.
I am working on a Mac II under Multifinder. This is a nasty problem,
because everything else works.
Another problem: How do I submit something to comp.sources.mac? I would
like to submit this DA whenever it is finished.
Jay, etc... (D. Jay Newman)
ps.
I am using MPW 3.0 B something.kish@.ucalgary.ca (Bradley John Kish) (02/04/89)
In article <69823DN5@PSUVM> DN5@PSUVM.BITNET writes: >Hi netlanders: > > I am working on a DA which needs some random numbers. My problem is that >I always get the same sequence. I have tried setting randSeed, but this >doesn't seem to do anything (actually I set it before I call Random, then >save this value, then reset the original value, so that it won't mess up >an application). However, anything I do (including attempting a FOR loop >repeating for a somewhat random time (GetTimeDate + CountTicks), seems to >do nothing: I still get the same sequence. The problem has to do with the way the Mac handles DA's and the Random number seed. The random number seed is stored as a GLOBAL and DA's aren't allowed to touch with them. I can't remember what what nasty things happen if you do mess with them, though. To get around this problem you have to write your own random number generator. Any good book on computer algorithms will have one. To start off randomly, set your seed to the system clock. Hope this was of some help. Good Luck! Brad Kish. *** What's all this disclaimer jazz anyway? kish@enel.UCalgary.CA
kaufman@polya.Stanford.EDU (Marc T. Kaufman) (02/04/89)
In article <69823DN5@PSUVM> DN5@PSUVM.BITNET writes: >Hi netlanders: > I am working on a DA which needs some random numbers... > I am working on a Mac II under Multifinder. This is a nasty problem, >because everything else works. >ps. > I am using MPW 3.0 B something. Have I got a deal for you. The function below will do what you want. I wrote it as an exercise to see just how fast a '020 will go: Marc Kaufman (kaufman@polya.stanford.edu) --------------------- CUT HERE, submit to MPW Assembler ---------------------- ******************************************************************* * ROUTINE Random.a * FUNCTION Provide Minimal-Standard random number generator * CALLING C-compatible calling sequences * * This generator is taken from CACM, October 1988, p.1192 * * It was designed by Stephen K. Park and Keith W. Miller * * For future reference: Possible better multipliers are 48271 and 69621 ******************************************************************* CASE ON MACHINE MC68020 PROC ENTRY MSRandomSeed Start DC.W ' ) Written by Marc Kaufman, Kaufman Research, 1988 ' MSRandomSeed DC.L 1 ; Place to store seed value ENDPROC ******************************************************************* * ROUTINE MSGetSeed * FUNCTION Return the current seed value * INPUT none * OUTPUT D0 = the seed ******************************************************************* MSGetSeed FUNC EXPORT move.l MSRandomSeed,D0 rts ENDFUNC ******************************************************************* * ROUTINE MSRandom * FUNCTION Return the next random number * INPUT none * OUTPUT D0 = the number ******************************************************************* MSRandom FUNC EXPORT lea MSRandomSeed,A0 move.l (A0),D1 mulu.l #16807,D0:D1 ; long multiply divu.l #$7fffffff,D0:D1 ; modulo 2^31 -1 move.l D0,(A0) rts ENDFUNC ******************************************************************* * ROUTINE MSRanSet * FUNCTION Set the random number seed * INPUT The seed value * OUTPUT The seed value ******************************************************************* MSRanSet FUNC EXPORT move.l 4(A7),D0 ; the value the user wants to use and.l #$7fffffff,D0 beq.s MSGetSeed ; zero is not a valid value cmp.l #$7fffffff,D0 beq.s MSGetSeed ; neither is 2^31-1 lea MSRandomSeed,A0 move.l D0,(A0) rts ENDFUNC END
jkjl@munnari.oz (John Lim) (02/06/89)
In article <69823DN5@PSUVM> DN5@PSUVM.BITNET writes: >Hi netlanders: > > I am working on a DA which needs some random numbers. My problem is that >I always get the same sequence. I have tried setting randSeed, but this You just cannot simply reset randSeed in C by : randSeed = TickCount(); because LSC thinks that randSeed is a DA global offset by A4 and not A5 (which is what u want). What you need to do is use some assembly to access randSeed. Sorry but i dont have my notes and docs here so i cant show it to you from memory. >doesn't seem to do anything (actually I set it before I call Random, then >save this value, then reset the original value, so that it won't mess up >an application). However, anything I do (including attempting a FOR loop >repeating for a somewhat random time (GetTimeDate + CountTicks), seems to >do nothing: I still get the same sequence. Why dont you use your own random number generator ? john
lsr@Apple.COM (Larry Rosenstein) (02/07/89)
In article <2671@munnari.oz> jkjl@munnari.UUCP (John Lim) writes: >In article <69823DN5@PSUVM> DN5@PSUVM.BITNET writes: >> I am working on a DA which needs some random numbers. My problem is that >>I always get the same sequence. I have tried setting randSeed, but this > > You just cannot simply reset randSeed in C by : > > randSeed = TickCount(); > > because LSC thinks that randSeed is a DA global offset by A4 and > not A5 (which is what u want). randSeed happens to be a Quickdraw global variable. A5 always points to a pointer to the Quickdraw variables. I think assembly code such as: MOVE.L (A5),A0 MOVE.L randSeed(A0),D0 will get randSeed. From the MPW assembler interfaces, it looks like randSeed is -$7E. Looking at the code for _Random in Macsbug seems to confirm this. -- Larry Rosenstein, Object Specialist Apple Computer, Inc. 20525 Mariani Ave, MS 46-B Cupertino, CA 95014 AppleLink:Rosenstein1 domain:lsr@Apple.COM UUCP:{sun,voder,nsc,decwrl}!apple!lsr