Re: [Semi-OT] 192kHz/96kHz soundcard recommendations ?
Re: [Semi-OT] 192kHz/96kHz soundcard recommendations ?
- Subject: Re: [Semi-OT] 192kHz/96kHz soundcard recommendations ?
- From: Mike Kent <email@hidden>
- Date: Fri, 12 Dec 2003 09:24:02 -0800
>
You have to remember that USB devices indicate their bandwidth
>
requirements at enumeration. When playback starts, the device requests
>
that bandwidth; if not enough bandwidth is available at that time, then
>
the playback doesn't start.
That is correct.
>
Likewise, if you're playing back from a USB
>
hard disk, that too will request bandwidth and get it all or nothing.
That is not correct.
A USB Audio Device uses isochronous transport. It requests the necessary
bandwidth at enumeration time. If the necessary USB bandwidth is available
(not already reserved) then the host reserves that much bandwidth for the
device.
On the other hand, a hard drive uses asynchronous transfer. It tells the
host the maximum packet size it can handle. The bandwidth it may use is any
left over, up to that max packet size, after all isochronous transfers have
been completed. No bandwidth is actually reserved for asynchronous devices.
So audio devices get reserved bandwidth, hard drives do not.
If something interrupts audio data flow over USB it is not the fault of USB,
the problem is occurring higher up in the host. USB has guaranteed the
necessary bandwidth. On older operating systems (OS9 or Win98 before SE) USB
Audio isochronous transfers were sometimes not given high enough priority in
the host. So audio playback data was occasionally blocked by some other
process (such as networking or graphics) before it got sent from the OS to
USB. Audio data packets were occasionally lost.
Today's Operating Systems (OSX or WinXP) generally do not suffer the same
type of audio data loss. USB audio threads have been given a higher priority
than in past operating systems. Today's faster CPU and platforms also help.
USB 2.0 may also help somewhat to preserve integrity of the data stream in
that the OS should at least be aware that USB 2.0 is running at 48 times the
speed of USB 1.1. The host should allocated sufficient resources to
servicing a much larger data bus. So the same stream of audio data that was
playing on 12Mb probably is more securely insured on a 480Mb bus.
Going back to what I said about USB hard drives not having guaranteed
bandwidth, I probably wouldn't rely on a USB storage device for time
sensitive applications such as pulling data to feed multi-channel audio if
that drive is sharing the USB with other high bandwidth devices.
But USB audio works pretty well. In general USB Audio data dropouts that
used to be problematic in some early USB host systems are rare today. The
Edirol UA-1000 reliably runs 20 channels of 24bit 96KHz audio on USB 2.0.
Regards,
Mike.
-----------------------------------------------
| Mike Kent, Manager of Technical Relations |
| Roland Corporation |
| Email: email@hidden |
| Tel: (604)543-2973 Fax: (604)543-2983 |
-----------------------------------------------
_______________________________________________
coreaudio-api mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/coreaudio-api
Do not post admin requests to the list. They will be ignored.