Re: interpreting HALLab IO Cycle telemetry
Re: interpreting HALLab IO Cycle telemetry
- Subject: Re: interpreting HALLab IO Cycle telemetry
- From: Jeff Moore <email@hidden>
- Date: Wed, 3 May 2006 14:43:28 -0700
I should mention that the purpose of this check is to enforces that
devices that have a clock whose rate stays basically the same all the
time don't start varying wildly. To trip this check in the HAL the
following conditions have to be met:
1) The device claims it's timeline is constant rate by setting
kIOAudioEngineClockIsStableKey to a non-zero value. If this key is
not present on the engine, then the HAL assumes that the clock is
stable if the device is connected via USB.
2) The new time stamp in question is more than a millisecond before
the time the HAL expects it to be.
Given that you are working with a USB device, I'd say that having a
time stamp that is more than a millisecond early most likely means
that you have some issues with your time stamp code in your driver.
The reason why is that USB packets are clocked by the millisecond and
an error this big is likely a miscalculation somewhere as opposed to
reflecting what the device is actually doing.
On the other hand, if the device's rate scalar is really varying by
that much, you will need to be sure to set
kIOAudioEngineClockIsStableKey to a non-zero value in the IORegistry
entry for your engine.
On May 2, 2006, at 5:55 PM, Jeff Moore wrote:
That would be a bad time stamp coming from the driver. The bad time
stamp is causing the HAL to resynch.
I would get a screenshot of the telemetry and pull together as much
supporting info as you can and file a bug with the folks that wrote
the driver.
On May 2, 2006, at 2:54 PM, david tay wrote:
While tracking the playback from an application such as Amadeus II
to my usb audio device, on occasion I get an entry such as
0 Early: 113 2414.361 (+0.02) Zero: ( 114688, 2291.211, 0.000000)
and an audible pop.
How would one go about interpreting such entries?
It doesn't appear that there is an overload situation.
--
Jeff Moore
Core Audio
Apple
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Coreaudio-api mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden