I think you should look at the average. My routine is dramatically faster if largest number is the first parameter. Other patterns will also deliver different times. I would be interested in what results your tests reveal.
Thanks again to Wilfried for his original ASM code. It was so easy to understand that I was inspired to take my first stab at an ASM routine. And, thanks to Rob for giving me the idea to count the comparisons and assignments as a metric. I don't think there is any way to improve on his Pascal code. What a great community this is. Kind regards, Jon P. Grewer > Message: 1 > Date: Wed, 29 Aug 2007 11:58:57 +0200 > From: CubicDesign <[EMAIL PROTECTED]> > Subject: Re: maximum of 3 numbers > To: Borland's Delphi Discussion List <[email protected]> > Message-ID: <[EMAIL PROTECTED]> > Content-Type: text/plain; charset=ISO-8859-1; format=flowed > > Hi Jon, > > You function returns the best time. > > Still I wonder if I should make and average of all times or to look only > for the smallest time. > I think I should look for the smallest time. > > > > > Jon P. Grewer wrote: > >> Using Wilfried's asm code and Rob's tight Delphi code as inspiration and >> jumping off point: This is my first ASM routine ever. >> >> function Max3(a, b, c: integer): integer; >> // eax : first param & Result >> // ecx : second param >> // edx : third param >> asm >> cmp eax, ecx >> jl @less // ecx > eax >> cmp eax, edx >> jl @less2 // eax > ecx but less than edx. edx is max. 2 >> compares, 1, assignment, 1 jump >> ret // eax is max. 2 compares, 0 assignments, 0 jumps >> >> @less: // ecx > eax.. Is ecx greater than edx? >> cmp ecx, edx >> jl @less2 // edx is max. 2 compares, 1 assignment, 2 jumps >> mov eax, ecx // ecx is max. 2 compares, 1 assignment, 1 jump >> ret >> >> @less2: >> mov eax, edx // edx is max >> end; >> >> Warmest regards and thanks for the inspiration, >> -- Jon Grewer >> >> >> _______________________________________________ Delphi mailing list -> [email protected] http://www.elists.org/mailman/listinfo/delphi
