BeagleBone: I know it's not apples to apples

For background, I’m using one of Roberts Squeeze generated images for my BeagleBone, That part works great.

We also have an application that we are using on the Bone that interfaces with a USB device. With no hub involved. When we connect to the device and start the application, the system load goes to approx .50 at 15 minute averages. The process that’s running is only taking approx 8-10% of the cpu and there is no file system activity with this application.

The exact same application built on the Raspberry Pi, and connected to the same usb device, via the same setup and we see .08-.1 load averages and the same or similar CPU utilization.

Is there THAT much difference in the USB implementation between the Bone and the Pi? I have to believe that the issue is one of either the kernel driver or the usb driver… It’s just odd that the system load is so low on the Pi and yet so hi on the bone, both running at 700mhz, etc.

This is running Roberts latest omap-psp kernel as well for reference 3.2.25-psp20 I believe

Alan

It's the musb based usb host on the BeagleBone... If you were to
compare your application on a BeagleBoard while using the ehci port,
you'd see a similar comparison at this point in time... With the 3.2
fork, the musb port mostly works. If you want to get your hands
dirty, I'd retry with Koen's 3.6-rc based branch on github (but note,
it's a little limited for new/most users, (no working mmc drivers
yet))..

https://github.com/beagleboard/kernel/tree/beaglebone-3.6

Regards,

Robert as always - THANKS for the quick reply… I’m not of the bleeding edge group. So I’m perfectly happy to wait… Hopefully we’ll see something there soon… I’ll remember to check from time to time, that would be awesome to have that resolved.

Alan