Why VB 6.0??

[javajawa]

Sorry about the "ABC". I haven't had enough sleep

I've just run the same test as you, both with the listbox visible and hidden. I was very careful about form and control sizes for the visible one, so that the paint region was the same. I've taken averages across a number of runs (as this thing is becoming slightly jerking due to encoding mp3s...)

It was all done form a Compiled exe. Not using the debugger.

Results:
Visible:
Vb.Net: 156ms (ranged from 110 - 172, including 5 172s out of ten runs)
vb6: 106ms (range 78 - 165, including 4 94s out of ten runs)

Hidden (NOTE: NOT THE SAME NUMBER OF RUNS!)
Vb.net 528ms little range
vb6 I got bored after about 30seconds...so one run was 38953.

[Joeman]

No problem. Could you download the listbox_speed.zip and run them both compiled without any modification, so we can get a more solid ground on results. EDIT: I want you because I did modify code a bit, so your code is no longer the same size on both the gui or coding wise

[GremlinSA]

okay so i did much the same for the Sorting Algo's...

I did a quick rewrite of the VB6 code into VB.NET, and the results are some what interesting....

VB6 left, VB.NET right...


NOW... Some notes...

1) The sorting is done directly within the Listbox...
2) None of the code has been optimised..
3) Sorting times are also dependant on how out out of sort the array is, and it's a little beyond thescope of this test to try and give both test apps the same list..

Anyone care to interpret these results, and perhaps tweek the code...

[GremlinSA]

This is very interesting Gremmy!

So, what do you suppose accounts for the differences between the differences? LOL I mean, the .net version is about twice the speed this time, whereas drawing boxes was relatively close.

What i've been thinking is that .NET has an Updated maths object, something a lot faster, taking advantage of some of the new Chip features..

For example, VB6's "K = K + 1" can be translated to "K += 1"...

That's about all I can figure...

[Joeman]

I downloaded you code for sorting, but I couldn't test the vb6 version because I couldn't find it in there. Did you Post the vb6 version?

[GremlinSA]

I took the ideal to use a lisbox to see how fast vb6 and vb.net can populate the listbox.

The results are:

vb6 - 907 ms
vb.net - 2218 ms

This one i think may be a little easier to explain..

The Listbox Item object in VB6 is smaller than the VB.NET object.. In .NET it uses more memory, and hence takes longer to create/Allocate the memory..

Look at the sorting algo's test to see what i'm saying...

[Joeman]

At least this explains why some people call the gui part slower. It makes alot more since now.

EDIT: I am referring to locked and delayed window components on the window

[GremlinSA]

I downloaded you code for sorting, but I couldn't test the vb6 version because I couldn't find it in there. Did you Post the vb6 version?

The original VB6 code is in this article

The VB6 code has been pretty much tweeked, short of sorting in the Listbox. This was done on purpose, to enable the 'display sort progress', and a few other things...

[GremlinSA]

Okay i spent a little time checking the code and found a few things..

1) in a few places I put "If Prog.Checked = 1 Then", this takes longer to execute than "If Prog.Checked Then", which is what it was suposed to be..

2) changing the declaration of "S_Temp" to "Integer" instead of "Object" in the Swap Sub looks to help speed it up..

New times for .NET with these changes.. from the top.. 11622ms, 3822ms, 2106ms, 1045ms, 249ms, 483ms, 359ms...

[JeffB]

Care if I add a speed test of mine? It's pure calculation and the results are pretty interesting. Now, see if you can tweak some compilation parameters or some code parameters and even the things out. Maybe testing with the unsafe keyword (for the c# app) could make things faster?

Anyway, the .zip file include the code for c++, vb6, c# and vb.net and the results I got on my computer. Do you get similar results?

JeffB

[Arjay]

Wow, look at all these trees. What is this place?

[WizBang]

What i've been thinking is that .NET has an Updated maths object, something a lot faster, taking advantage of some of the new Chip features..

For example, VB6's "K = K + 1" can be translated to "K += 1"...

That's about all I can figure...

I was wondering about that, but also about the speed of the graphics object that you used in the .net version for drawing the squares. Perhaps the more graphics work the code must do, the slower .net would become relative to the VB6 code using API.

So it might prove insightful to test a routine which checks the properties of a bunch of controls, such as buttons, checkboxes, dropdowns, etc. As long as it doesn't change any properties, the controls won't need to be redrawn, so it should give some idea of how efficiently the objects are accessed.

But, before I write that, I'll post the sorting test I slapped together earlier. This uses BubbleSort, and the CheckBox determines whether the array to be sorted is numeric or string. (the original button and code is still there too). A ListBox is included to show multiple test results.

[GremlinSA]

Okay i've started a whole bank of threads in the CODE WAR forum, and think it'll be better to take these tests there...

This is the control thread which will have links to each of the individule tests .. I think it is important to keep each test in a seperate thread to minimise the cross test comparison...

[George1111]

Perhaps one of the questions should have been

"Why VB.Net ?"

http://www.adit.co.uk/html/vb7rant.html

http://www.geocities.com/tablizer/oopbad.htm

[javajawa]

Indeed.
Having now worked with both, I really think a VB7 might STILL be needed. The only features of .NET i've liked are Docking and the proper classes. Oh, and the ability to (if a bit messily and difficultly) deal with pointers. None of these, however, rely on the .NET framework...