I am seeing maximum throughput of a little over 400kbits/second using an Airprime PC5220 card and the generic usb serial driver. A windows driver can get over twice the throughput.
Throughput to me being TCP throughput after I have started up pppd on the ttyUSB0 port. I noticed a comment from you on one of the mailing lists suggesting that the generic USB driver is not optimized for speed and suggests that I should take a look at the empeg driver for an example of a driver with higher throughput. The empeg driver seems to do basically the same thing inside the bulk_read_callback that the generic driver does. It just pops the bytes into the flip buffer and resubmits the urb. I looked in the ir-usb driver which can use multiple urbs, but it looks like this happens when the particular port offers multiple endpoints. I highly suspect that the problem is on the incoming side, (as the only data flowing the other way are the TCP ACKs). Can you give me some insight into where I should look for solutions to my problem? Any hints or pointers to resources will be appreciated. Especially any ideas as to where you think the bottleneck might be. I saw (either on a list or in one of your L.J. articles) that it is possible to use a pool of urbs for one endpoint. Is this true for a bulk read endpoint? Is there an example floating around? Thank you for any pointers, -jon PS I sent this to the linux-usb-users list, should I have sent it to linux-usb-devel? ------------------------------------------------------- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt _______________________________________________ [email protected] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-users
