dachay2tnr
One Hit Wonder
Found this post on the Sonar newsgroup. It was posted by Ron Kuper who works for Cakewalk. He was responding to some user posts who claimed they were getting better CPU usage measurements with ASIO than they did with WDM.
I thought you all might find it of some interest.
I thought you all might find it of some interest.
I won't rule out the possibility that some ASIO drivers are more efficient that WDM drivers. However, it's important to note that the CPU meter when in WDM mode actually measures a bit more work than when in ASIO mode, due to differences in these driver architectures.
In both driver modes, the CPU meter measures how much time is spent performing audio DSP. But in WDM mode, the meter also measures how much time it takes to send new audio data buffers down into the driver. In ASIO mode, the data buffer transfer is handled by the driver, so we don't get to measure that bit.
A good analogy to help understand this goes as follows. It's the difference between making a telephone call vs. answering a telephone call. If you measure how much time it takes to call someone and say the word "boo", it would come up as taking more time than if you *answered* the phone and said "boo".
This is because when you make a call it takes time to dial, so you
would measure that, but when you answer a call the time to dial isn't measured. Somebody has to take the time to dial, it's just not you.
Dialing is analogous to transfering audio data to the driver. In WDM mode, SONAR does the "dialing"; in ASIO mode, the driver does the "dialing". I would bet the net total CPU usage in either case is about the same.
Clear as mud?
