C# Code optimization

Question

who could be faster ? and why ?

1:

Point point = new Point(25,25);   //any numbers..
Point point2 = new Point(20,95);  //any numbers..

Graphics g = CreateGraphics();
g.DrawLine(point,point2);

OR

2:

Graphics g = CreateGraphics();
g.DrawLine(new Point(25,25),new Point(20,95));

sugest you benchmark using System.Diagnostics.StopWatch. Question has little or no value to a wider audience... — Mitch Wheat, Jun 15 '09 at 00:41
I think the question is valid, don't see why the downvotes !! — Jhonny D. Cano -Leftware-, Jun 15 '09 at 00:41
The question is valid, but pointless. It's micro-optimization at best, preference at worst. — Eric, Jun 15 '09 at 00:43
I think the question is valid in its generalised form. I find arul's answer interesting. — ChristianLinnell, Jun 15 '09 at 00:45

score 15 · Answer 1 · answered Jun 15 '09 at 00:41

15

None of them, since both snippets will compile to the same MSIL code representation.

Besides, this is a micro-optimization, which you should avoid before actually knowing that it is the bottleneck.

answered Jun 15 '09 at 00:41

arul

13,998
1
57
77

score 4 · Accepted Answer · answered Jun 15 '09 at 00:41

4

Neither, there is no real difference between them except for the loss in readability (especially after a JITing).

answered Jun 15 '09 at 00:41

Aaron Maenpaa

119,832
11
95
108

what about memory? i thought that the first one create in memory two points (objects) but the second one doesn't create in memory? is that right ???? – Sherif Jun 15 '09 at 01:00
1

@mavric - are created the same way -- on the stack (at least I think Point is a structure). The only difference even with reference types is that in #1, in debug mode, the Point objects won't be collected until you exit the method. In release mode, the objects will have the same lifetime. – Jonathan Rupp Jun 15 '09 at 01:29

score 2 · Answer 3 · answered Jun 15 '09 at 01:10

2

Micro Optimization, huh? One notable playwright says code readability is more important than micro optimizations, and I agree.

answered Jun 15 '09 at 01:10

Factor Mystic

26,279
16
79
95

score 2 · Answer 4 · answered Jun 15 '09 at 01:16

2

Neither is faster, the fastest one would be the one that avoids allocating those points in the Render path altogether and precreates them earlier

answered Jun 15 '09 at 01:16

Ana Betts

73,868
16
141
209

score 0 · Answer 5 · answered Jun 15 '09 at 00:41

0

2 might be faster because you're not creating intermediate pointers to the object before passing it to g.DrawLine; however, if that's the only place you use point and point2 then the compiler will likely optimize so the binary is the same anyway.

answered Jun 15 '09 at 00:41

Charles Ma

47,141
22
87
101

score 0 · Answer 6 · edited May 23 '17 at 12:12

It all depends on what else you're doing.

If you're doing almost nothing else (which I doubt) then this is the "bottleneck".

If this is the "bottleneck" then you still don't know what's dominant:

two "new"s and associated constructors, destructors, and garbage collection.
actually rendering lines.

The second is probably unavoidable, but the first is only there for style reasons.

You can find out by profiling or this simple method.

Or you can avoid the whole question with:

g.DrawLine(25,25,20,95);

C# Code optimization

6 Answers6