About speed with different looping-techniques?

[Lars Netzel]

I have a Dataset with one table in it.

If I need to loop thru that table I know of two ways that I would choose..

1. Do a "For Each DataRow in Datatable"
2. You create an IEnumerator fron the Table.Rows object. And loop thru that
one (a bit more code needed).

Questions:

1. Which one would be fastest and Why?

2.In general... how do you actually measure a thing like that, are there a
very precise way of measuring or any guidlines that you can use in general
to figure out what techniques you should choose in different situations?
Best Regards/
Lars Netzel

[David Williams]

From a real quick test, it appears that managing the IEnumerator CAN be quicker, but I had to run through a few million loops to feel it. This makes since when you look at what a For Each loop is doing - creating the IEnumerator on the fly and cycling over it.

How do you measure something like that? By doing it so many times that you can see it affects. I normally do something like (typed here so beware typos):

Dim StartTime as DateTime

StartTime = DateTime.Now
For i as integer = 0 to Integer.MaxValu e
For Each item as Object In Collection
'do nothing - timing loop code
Next
Next
Console.WriteLi ne "Took " & DateTime.Now - StartTime & " seconds."

HTH
--
David Williams, VB.NET MVP
"Lars Netzel" wrote:

I have a Dataset with one table in it.

If I need to loop thru that table I know of two ways that I would choose..

1. Do a "For Each DataRow in Datatable"
2. You create an IEnumerator fron the Table.Rows object. And loop thru that
one (a bit more code needed).

Questions:

1. Which one would be fastest and Why?

2.In general... how do you actually measure a thing like that, are there a
very precise way of measuring or any guidlines that you can use in general
to figure out what techniques you should choose in different situations?
Best Regards/
Lars Netzel

[One Handed Man \( OHM - Terry Burns \)]

In addition, be aware, that you cannot remove rows during an enumerated loop
or you will end up with an exception.

If you do need to alter the contents that way, use a backwards counting
for/next loop

--

OHM ( Terry Burns )
. . . One-Handed-Man . . .

Time flies when you don't know what you're doing

"David Williams" <Da***********@ discussions.mic rosoft.com> wrote in message
news:58******** *************** ***********@mic rosoft.com...

From a real quick test, it appears that managing the IEnumerator CAN be quicker, but I had to run through a few million loops to feel it. This
makes since when you look at what a For Each loop is doing - creating the
IEnumerator on the fly and cycling over it.
How do you measure something like that? By doing it so many times that you can see it affects. I normally do something like (typed here so beware
typos):
Dim StartTime as DateTime

StartTime = DateTime.Now
For i as integer = 0 to Integer.MaxValu e
For Each item as Object In Collection
'do nothing - timing loop code
Next
Next
Console.WriteLi ne "Took " & DateTime.Now - StartTime & " seconds."

HTH
--
David Williams, VB.NET MVP
"Lars Netzel" wrote:

I have a Dataset with one table in it.

If I need to loop thru that table I know of two ways that I would choose..
1. Do a "For Each DataRow in Datatable"
2. You create an IEnumerator fron the Table.Rows object. And loop thru that one (a bit more code needed).

Questions:

1. Which one would be fastest and Why?

2.In general... how do you actually measure a thing like that, are there a very precise way of measuring or any guidlines that you can use in general to figure out what techniques you should choose in different situations?
Best Regards/
Lars Netzel

[Jay B. Harlow [MVP - Outlook]]

Lars,
In addition to the other comments.

Does it really matter?
I find its better to code for "correctnes s" (OO) and "readabilit y" first,
then code for performance only when a routine has proven to have performance
problems via profiling.

IMHO: For Each is much more readable then the IEnumerator loop, plus For
Each is more "correct" in that the For Each will call IEnumerator.Dis pose
for me if needed...
Also as David stated, the For Each is implemented in terms of IEnumerator,
there is not going to be any real difference in speed.
In addition to David's sample of using DateTime.Now to time code, I normally
use QueryPerformanc eCounter to time code.

http://support.microsoft.com/default...b;en-us;306978

Hope this helps
Jay

"Lars Netzel" <[stop_spam]@host.topdomain > wrote in message
news:ua******** *****@tk2msftng p13.phx.gbl...

I have a Dataset with one table in it.

If I need to loop thru that table I know of two ways that I would choose..

1. Do a "For Each DataRow in Datatable"
2. You create an IEnumerator fron the Table.Rows object. And loop thru that one (a bit more code needed).

Questions:

1. Which one would be fastest and Why?

2.In general... how do you actually measure a thing like that, are there a
very precise way of measuring or any guidlines that you can use in general
to figure out what techniques you should choose in different situations?
Best Regards/
Lars Netzel