TThread vs CPU usage?

[Pyrogine]

Yea, I was trying to say he was right to use a double buffer. Something happen (error message of some sort) and I then was not able to edit the post.... arrrg. I am doing polling too. I think you can enable a callback in DirectSound, but I've not yet had a chance to look into it. The polling has been working fine me. The Sleep seems to work good. The value that I came up with works well based on the buffer size. Smaller then this for the buffer size that I use then sound starts to overlap, larger then too much latency.

[alexione]

Actually, if you use OpenAL, there IS a way to check for buffer's state. Here is very good lesson how to do streaming in C++:

http://www.devmaster.net/articles/op...ls/lesson8.php

Pay special attention to function ogg_stream::update()

[arthurprs]

sleep is good, don't worry

[Andreaz]

Sleep(Time > 0) may cause the thread to sleep for to long due to the window times of the scheduler (16ms on win if i'm not misstaken), a way that quake used that works really well even for unthreaded applications is:

[pascal]
UpdateTime:= GetCurrentTime;
repeat
Sleep(0);

CurentTime:= GetCurrentTime;
until CurentTime- UpdateTime >= UpdateTime;
[/pascal][/pascal]

[arthurprs]

Sleep(Time > 0) may cause the thread to sleep for to long due to the window times of the scheduler (16ms on win if i'm not misstaken), a way that quake used that works really well even for unthreaded applications is:

[pascal]
UpdateTime:= GetCurrentTime;
repeat
Sleep(0);

CurentTime:= GetCurrentTime;
until CurentTime- UpdateTime >= UpdateTime;
[/pascal][/pascal]

it's precise but eats the cpu instead