|
Maybe they should build a factory. 
> The problem with computers is that they do what you tell them to do and not what you want them to do. <
> Life: great graphics, but the gameplay sux. <
|
|
|
|
|
Code-o-mat wrote: Maybe they should build a factory.
Or in this case a vectory... 
There are three kinds of people in the world - those who can count and those who can't...
|
|
|
|
|
yeah i agree with both ov u.. 
|
|
|
|
|
Well, since we are fashionable developers, we have a Class Factory...
If the Lord God Almighty had consulted me before embarking upon the Creation, I would have recommended something simpler.
-- Alfonso the Wise, 13th Century King of Castile.
This is going on my arrogant assumptions. You may have a superb reason why I'm completely wrong.
-- Iain Clarke
[My articles]
modified on Wednesday, June 10, 2009 7:49 AM
|
|
|
|
|
I have a complex application with many background threads running and I use "CSingleLock::Lock" extensively to lock access to critical data.
I recently noticed a bug where one of my threads was running very slowly and making one of the application's operations almost unusable. After much debugging, I narrowed it down to the thread's repeated use of "CSingleLock::Lock" which is used (along with a CMutex member variable) to lock out some critical data to stop it being used by a different thread, in a different function, at the same time.
My first (obvious) thought was that this second thread was locking the mutex and so causing my original thread to slow down (even though, due to how the threads are used, it is highly unlikely that they will need to access the data at the same time). However I did extensive tests to prove that this was not the case, including stopping the second thread and making sure that my original thread is the only one using the mutex.
It seems that just calling "CSingleLock::Lock" is enough to slow my thread down, regardless of whether the mutex is locked by another thread.
The thread's priority is 0. If I change this to 1 then the problem does not happen. It also only happens on single core PCs.
Has anybody out there encountered a similar issue when using low priority threads?
Is this a common feature of multithreading which I just haven’t seen before?
|
|
|
|
|
Ever since I read this[^] and was able to reproduce some of the issues listed, I stopped using MFC synchronization classes (years ago) and haven't looked back.
Not that I'm providing this as a solution to your query, but using the Win32 APIs directly for synchronization works sweetly enough for me. You might want to consider.
It is a crappy thing, but it's life -^ Carlo Pallini
|
|
|
|
|
Thank you, very interesting.
I'm moving on to C++/CLI programming in MsDev 2008 soon. Do you know of any similar problems with the equivalent .NET synchronisation classes (e.g. Mutex) ?
|
|
|
|
|
I don't pretend to be a .NET expert, but I can almost certainly say that you won't face such issues (as listed in the article by Joe) with .NET. I say it because everything is taken care of by the framework for you (how it does internally is another thing) and it's pretty impossible to shoot yourself in the foot with .NET.
Unfortunately, that doesn't stop a programmer from writing bad code and therefore, a badly formed, bloated, etc., program is possible to write. So, if you use a synchronization object improperly, expect the program to be slow. Things like memory damage, memory/resource leak, memory fragmentation, etc., are impossible though (I chose C# over C++ for a couple of server programs that I wrote for this reason).
It is a crappy thing, but it's life -^ Carlo Pallini
|
|
|
|
|
LetsMond wrote: It seems that just calling "CSingleLock::Lock" is enough to slow my thread down, regardless of whether the mutex is locked by another thread.
Locking a mutex is quite an expensive operation, even if the lock operation is immediately successful. In addition, the lock attempt (using, for example, WaitForSingleObject) is a kernel synchronisation point, which allows higher priority threads to take over the thread's time slice. Sounds like this is what's happening?
Java, Basic, who cares - it's all a bunch of tree-hugging hippy cr*p
|
|
|
|
|
LetsMond wrote:
Has anybody out there encountered a similar issue when using low priority threads?
Absolutely.
LetsMond wrote:
Is this a common feature of multithreading which I just haven’t seen before?
You will probably get better performance using the Interlocked Functions[^] rather than the CSingleLock class[^]. In high-performance threads you need to avoid synchronization with kernel objects.
Note that all synchronization techniques are relatively expensive because all of them use a memory barrier[^] which essentially means that are using SFENCE [^]or MFENCE[^] instructions to serialize memory access.
Best Wishes,
-David Delaune
|
|
|
|
|
How to add a CEdit-derived class output the text/icon in the beginning and-or in the end of CEdit window.
I can allocate an area for the text/icon by SetMargins method, but in what method and how is better output a text or icon not clearly.
It is desirable, that other CEdit functional was not broken
Similar to CEdit from Data group and MsgID groups (http://www.canusb.com/canusb_labview.jpg[^])
|
|
|
|
|
|
hi guys
i try copy some chinese word "你好" from a .txt file as edit control's caption
but i fail.it show some garbled text
how to solve this problem
could help me
thank you;
|
|
|
|
|
Hi,
What do you mean by it fails? The garbled text is in Chinese or not? If not Chinese, do you see question marks instead? Is your application fully Unicode?
It is a crappy thing, but it's life -^ Carlo Pallini
|
|
|
|
|
Rajesh R Subramanian wrote: Hi,
What do you mean by it fails? The garbled text is in Chinese or not?
We have nothing, so far. We need facts, Watson, facts!
--Sherlock Charles
If the Lord God Almighty had consulted me before embarking upon the Creation, I would have recommended something simpler.
-- Alfonso the Wise, 13th Century King of Castile.
This is going on my arrogant assumptions. You may have a superb reason why I'm completely wrong.
-- Iain Clarke
[My articles]
|
|
|
|
|
open ur .rc file in notepad and save it as unicode...
|
|
|
|
|
ali kanju wrote: open ur .rc file in notepad and save it as unicode...
Uh...What? I have no .rc file, but I don't care at all. On the other hand, I need facts... 
--Sherlock Charles Again
If the Lord God Almighty had consulted me before embarking upon the Creation, I would have recommended something simpler.
-- Alfonso the Wise, 13th Century King of Castile.
This is going on my arrogant assumptions. You may have a superb reason why I'm completely wrong.
-- Iain Clarke
[My articles]
|
|
|
|
|
|
Facts, facts. For instance is a fact that you missed the OP, replying instead to me.
--Sherlock Charles Again & Again
If the Lord God Almighty had consulted me before embarking upon the Creation, I would have recommended something simpler.
-- Alfonso the Wise, 13th Century King of Castile.
This is going on my arrogant assumptions. You may have a superb reason why I'm completely wrong.
-- Iain Clarke
[My articles]
|
|
|
|
|
|
OP - Original Post.
If the Lord God Almighty had consulted me before embarking upon the Creation, I would have recommended something simpler.
-- Alfonso the Wise, 13th Century King of Castile.
This is going on my arrogant assumptions. You may have a superb reason why I'm completely wrong.
-- Iain Clarke
[My articles]
|
|
|
|
|
|
|
What do you mean by it fails?
A:fail is means after copy the chinese word is not i want just so;
The garbled text is in Chinese or not?
A:the garbled text is also chinese
you can see below:
when before copy chinese word "你好"
after copy the chinese word in edit caption is "斕疑"
Is your application fully Unicode?
A:my application's character set is MBCS
|
|
|
|
|
jeansea wrote: A:my application's character set is MBCS
To begin with, you can do an Unicode build.
It is a crappy thing, but it's life -^ Carlo Pallini
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
