CPU Cache, pipeline, memory, and hard disc access methods are often optimized for increasing, rather than decreasing order of access. So it's possible that iterating an array backwards could be considerably slower than doing it forwards.
Caveats - this is just theory, and in practice it may not matter. Also, I seem to remember that some Windows data structures (DIBs?) are in fact stored in reverse order, in which case perhaps the opposite advice applies. On May 17, 2007, at 6:39 AM, Mathieu Langlois wrote: > If order is not critical, you can use: > > For i = ubound(arr) downto 0 > ... > Next > > In that case, there is no variable, and ubound is calculated only once > (just as if you would have used a variable). > > UBound is a function, I doubt they recalculate the size every time, > but there is definitely function calling overhead. > > Math > > On 5/17/07, Daniel Stenning <[EMAIL PROTECTED]> wrote: >> When I have to use for/next for processing arrays, >> I like, I suspect many others, tend to put the array ubound into a >> local >> variable like so: >> >> Dim counter as integer = myarray.ubound // ( new rb syntax ) >> >> For I as integer = 0 to counter >> ... >> Next >> >> >> As opposed to , what would be more clean: >> >> >> For I as integer = 0 to myarray.ubound >> ... >> Next >> >> >> Now I just did some timings, and it seems that there is a slight >> ( not great >> ) performance advantage with using a local variable. I used the >> following >> code ( with an empty array ) >> >> //------- >> dim u as integer = UBound(arr) >> >> Result.Text = "Started" >> app.DoEvents >> >> for j as integer = 0 to 200000000 >> for i as integer = 0 to u >> Next >> Next >> >> >> Beep >> Result.Text= "FINISHED" >> //-------- >> >> >> and: >> >> // ------- >> Result.Text = "Started" >> app.DoEvents >> >> for j as integer = 0 to 200000000 >> for i as integer = 0 to UBound(arr) >> Next >> Next >> >> Beep >> Result.Text= "FINISHED" >> //-------- >> >> >> This makes me wonder: - Idoes Ubound use a method/function call >> to get the >> ubound value, or does the RB generate code to directly access a >> "property" >> or variable holding the internal array size ?. >> >> >> Naturally there would be some speed advantage to NOT using any >> function to >> get the size. >> >> >> Basically I would prefer not to have to use the local variable >> counter >> method, or as I often have done - creating a class property to >> hold the >> array count. It would be so much cleaner just to use UBOUND >> directly in the >> for loop, knowing that I incur no performance penalty for doing so. >> >> >> Anyone have info on this ? My test seems to suggest there is not >> a huge >> difference in speed, but a difference nevertheless. > _______________________________________________ > Unsubscribe or switch delivery mode: > <http://www.realsoftware.com/support/listmanager/> > > Search the archives: > <http://support.realsoftware.com/listarchives/lists.html> _______________________________________________ Unsubscribe or switch delivery mode: <http://www.realsoftware.com/support/listmanager/> Search the archives: <http://support.realsoftware.com/listarchives/lists.html>
