|
Hello!
I've got a problem with multithreading.
One should think that every Thread-Model causes overhead and is responsible
for a little bit lost of performance on a singel-Core CPU!
In my Case, the sequentiell version is about 3 % slower than the Threading-versions.
Has anybody an explanation for this?
//the worker-function
public void Function()
{
for (int i = 0; i < 10000; i++)
{
for (int j = 0; j < 10000; j++)
{
j = j + 1;
j = j - 1;
}
}
}
//Needed to set the AutoResetEvent -> Overhead?
public void FunctionNT(Object obj)
{
Function();
Object tmp = obj;
if (tmp != null)
{
((AutoResetEvent)obj).Set();
}
}
// sequentiell Version
class SequentiellTest : ThreadingParent
{
public SequentiellTest(int piterations)
{
this.iterations = piterations;
}
public void Run()
{
Worker wrk = new Worker();
for (int i = 0; i < iterations; i++)
{
wrk.Function();
}
}
}
// Version newThread
class NewThreadTest : ThreadingParent
{
public NewThreadTest(int piterations)
{
this.iterations = piterations;
}
public void Run()
{
AutoResetEvent[] ready = new AutoResetEvent[iterations];// = new AutoResetEvent(false);
Worker wrk = new Worker();
Thread[] tda = new Thread[iterations];
for (int i = 0; i < iterations; i++)
{
ready[i] = new AutoResetEvent(false);
tda[i] = new Thread(wrk.FunctionNT);
tda[i].Start(ready[i]);
}
//wait until all Threads are Ready
for (int j = 0; j < iterations; j++)
{
ready[j].WaitOne();
}
}
}
// Example for a Calling-Function
private void NewThreadStart()
{
//start the Stopwatch
stp.Start();
//Init NewThreadClass
NewThreadTest ntt = new NewThreadTest(Convert.ToInt16(tbIterations.Text));
ntt.Run();
//Stop the Stopwatch
stp.Stop();
}
|
|
|
|
|
Hi all..
I need to know what is meant by satellite assembly..and what is the purpose of it..?
Regards,
Balaguru
|
|
|
|
|
Try Googling it!
Trinity: Neo... nobody has ever done this before.
Neo: That's why it's going to work.
|
|
|
|
|
Hi..
Tell me the exact definition if u know..I know that i can get if from google but i couldn't do it due to certain issues..Help in this regard..
Regards,
Balaguru
|
|
|
|
|
|
For me satellite assembly mean is when you create an application using feature like globalistion and localisation, create resource to extract region specific information. In this senerio the assemblies behave according to region ....
Regards,
Jaiprakash M Bankolli
jaiprakash.bankolli@gmail.com
http://jaiprakash.blog.com/
|
|
|
|
|
Hi guys,
I find (especially in .NET 1.1) that many UI type classes are kept "sealed". Any specific reason for that?
Examples: Classes like Tooltip, Progressbar etc.
Typcially, UI classes are supposed to be extensible. I can understand if 3rd party vendors keep their UI classes sealed (for more business oppurtunity  ) and therefore only provide developers to extended the functionality by providing properties or other extension mechanisms like interfaces on the controls/components (for custom rendering of the component) etc.
But why would Framework providers like Microsoft, keep their (of course, not all!) UI classes sealed?
|
|
|
|
|
One particular reason is so that you can't end up doing something that breaks the underlying object. Bear in mind that some of the controls are effectively wrappers for API calls you can appreciate that it would be very easy for you to do something that just broke them.
Deja View - the feeling that you've seen this post before.
|
|
|
|
|
My understanding is that classes like Tooltip etc are final form of any feature/ UI type. If you try to customse it you will not achieve much ... so they are SEALED. This make the developer use it as it is in either custome or user control
Regards,
Jaiprakash M Bankolli
jaiprakash.bankolli@gmail.com
Blog: http://jaiprakash.blog.com/
Suggestions: http://jaitoimprove.blog.com/
|
|
|
|
|
Hi there,
My unmanaged C++ application (MFC) is consuming events provided by a .NET object exposed by a VB.NET application (.NET 1.1 is used). Things were working fine till yesterday. But today, when I tested the executables on the same test PC, the client hangs.
Further analysis revealed that the client was able to create the object in .NET application using IDispatch call, then find connection point successfully. But it hangs when it tries to call IConnectionPoint::Advice.
Following is the call stack of the thread (I created dump using adplus.vbs in hang mode)
0007f5ac 7c90e9ab 7c8094f2 00000001 0007f5d8 ntdll!KiFastSystemCallRet
0007f5b0 7c8094f2 00000001 0007f5d8 00000001 ntdll!ZwWaitForMultipleObjects+0xc
0007f64c 7c809c86 00000001 0007f75c 00000000 kernel32!WaitForMultipleObjectsEx+0x12c
0007f668 79297940 00000001 0007f75c 00000000 kernel32!WaitForMultipleObjects+0x18
0007f6ec 7923455f 00000000 00000001 00000001 mscorwks!NonNT5WaitRoutine+0x17a
0007f714 792335a3 00000001 0007f75c 00000000 mscorwks!MsgWaitHelper+0x62
0007f754 792272a0 00000208 00148ce0 0007f7c8 mscorwks!ComPlusWrapperCleanupList::CleanUpCurrentWrappers+0xd2
0007f768 79226ec9 0161537c 000b8e58 0007f7c8 mscorwks!ComPlusWrapper::Init+0x6f
0007f7a4 79226dff 00000000 00000000 00000000 mscorwks!COMInterfaceMarshaler::CreateObjectRef+0x62
0007f7b4 79225cb7 00000000 0012a430 000b8e58 mscorwks!COMInterfaceMarshaler::FindOrCreateObjectRef+0x43
0007fc5c 79343821 00000000 00000000 00000000 mscorwks!GetObjectRefFromComIP+0x1c5
0007fd44 7c1acfd2 0012a430 0101ea50 0101e7f8 mscorwks!ConnectionPoint::Advise+0x151
*** WARNING: Unable to verify checksum for MySample.exe
*** ERROR: Module load completed but symbols could not be loaded for MySample.exe
0007fd64 0040f8b5 0012a430 00427884 0101ea50 mfc71!AfxConnectionAdvise+0x41
Any idea what is going wrong?
S o h a i l K a d i w a l a
To Err Is Human; to Debug, Divine
modified 21-Apr-21 21:01pm.
|
|
|
|
|
This is more of a Muse than an actualy question, but answers would be welcome.
I find occasionaly that i may use an object for a short period of time and after that, never want to use it again. It seems a waste to let something go through the full GC process, when i could kill it there and then.
An example would be a StreamReader that i use to read an input stream. It only exists in the scope of the function, and at the end of that it will no longer be needed. If it existed I could just call GC.Kill(reader) and be done with it.
What are peoples thoughts on this?
Tris
-------------------------------
Carrier Bags - 21st Century Tumbleweed.
|
|
|
|
|
Why don't you Dispose it?
|
|
|
|
|
I didn't mean disposable objects specificaly (Perhaps streamreader was a bad example).
I'm just wondering why there is no option to destroy objects that you know you have no further use for, rather than cluttering up the GC.
T
-------------------------------
Carrier Bags - 21st Century Tumbleweed.
|
|
|
|
|
It's already cluttered up to death anyway. All strings are immutable objects. String manipulations create and destroy lots of small objects, that just hang around until the GC gets around to collecting them.
It's a nice idea on paper, but it also introduces problems that can't be fixed until the next GC, like memory fragmentation. The GC maintains the managed heap, compacting memory when it can and when it needs to. It's performance is also self-tuning. Having you just kill and forcibly collect any object you want just destroys the ability of the tuner to do it's job.
Your idea is actually introducing the ability to create more clutter for the GC, not less.
Dave Kreskowiak
Microsoft MVP
Visual Developer - Visual Basic
2006, 2007
|
|
|
|
|
Ah, i see.
It was only a technical daydream, i just wondered why it wasn't done 
Cheers
Tris
-------------------------------
Carrier Bags - 21st Century Tumbleweed.
|
|
|
|
|
Because the GC never needs to visit dead objects (it only visits alive objects). So, you could easily implement a GC.Kill method all by yourself like this:
class GC
{
void Kill(object o)
{
}
}
See? Easy 
|
|
|
|
|
Hullo,
IS it possible to build applications on .net and run them on a a freeBSD,linux or any other open source OS.
If so? To what degree would this work?
cheers
Waskira
The level of your experience depends on your view
|
|
|
|
|
Try the Mono Project.
-------------------------------
Carrier Bags - 21st Century Tumbleweed.
|
|
|
|
|
I tired that one time. Took me weeks to get it working right on my Fedora Core 6 box and then it was still... I dunno'.. odd feeling...
|
|
|
|
|
Check out the Mono Mono[^] project.
It is not 100% feature complete when compared to the Microsoft .NET Framework, but it is pretty close.
-----------------------------
In just two days, tomorrow will be yesterday.
|
|
|
|
|
And they are playing serious catch up, so although it will unlikely ever be feature complete with the latest .NET version, it will probably be very close to 2.0 soon, an that's quite usable.
|
|
|
|
|
Is there a way to run a native Windows application if .NET is not installed on the target machine?
There was a way to do this with MS-DOS and Windows applications, by attaching a stub[^] application to the main executable. Then, if the application was launched in MS-DOS, the stub would run instead, usually providing a suitable error message.
Does something similar exist for .NET 2.0?
|
|
|
|
|
No the .net framework must be installed.
Ben
|
|
|
|
|
The .NET Framework must be install for the app to work.
You could create a stub, written in unmanaged C++ that MIGHT work, but I haven't heard of anything available or anyone writing one.
Dave Kreskowiak
Microsoft MVP
Visual Developer - Visual Basic
2006, 2007
|
|
|
|
|
Dave Kreskowiak wrote: You could create a stub, written in unmanaged C++ that MIGHT work
Yes, but how?
The MS-DOS stub was trivial - it was just a one line .EXE that told you to go get Windows. It was added at link time by the /STUB linker argument. I can't find any documentation (apart from vague references to statements like ".NET will contain a win32 stub...") on how to do this - or even if such a thing is possible!
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
