[LAU] Intel Rate Matching Hub and USB soundcards

Clemens Ladisch clemens at ladisch.de
Sun Jun 13 13:24:40 UTC 2010


Jeremy Jongepier wrote:
> Does anybody know if there is any relationship between the Intel Rate
> Matching Hub and poor performance of USB soundcards?

The "Rate Matching" behaves just like a normal USB hub; this design
choice does not introduce any problems by itself.

However, the USB controllers in these chipsets (5 and 3400 series) are
just plain buggy; the bug list ("specification update") says:
| 6. Intel® 5 Series Chipset and Intel® 3400 Series Chipset Full-Speed
|    USB Isochronous Packet Truncation
| 
| Problem: Full-Speed isochronous-out transactions with multi-frame
| packets may be truncated, in the presence of Full or Low-Speed USB
| asynchronous transactions.
| • For this to occur, two devices, one performing Full-Speed
|   isochronous-out transactions and one performing asynchronous
|   transactions must be connected to the same USB controller
|   (Ports 0-7 and 8-13).
| 
| Implication: In the case of a USB audio device this issue may result
| in no audible impact or audible artifacts such as pops and clicks.
| 
| Note:
| • High-Speed and Low-Speed USB devices are not impacted by this issue.
| • Only devices supporting Full-Speed isochronous-out transactions that
|   Intel is aware of are audio devices, such as sound adapters,
|   speakers, and headphones.
| • Intel has only observed the issue when a Full-Speed audio devices
|   and Full-Speed USB web camera are connected to the same USB
|   controller.
| 
| Workaround: None.
| Status: Fixed

(that is, fixed in chip revision B3)


Regards,
Clemens


More information about the Linux-audio-user mailing list