On Fri, 27 Jul 2012, Hans Petter Selasky wrote: > On Friday 27 July 2012 10:52:11 Alexey Filin wrote: > > USB is a good choice for some bus adapters, it is used widely, cheap, > > simple. For 2-byte read/write we could use either two bulk IN/OUT > > endpoints (131 * 8kHz ~ 1 MHz) or one control endpoint (table 5-3, 42 > > * 8 kHz = 336 kHz). Ethernet is overkill for the task even if it could > > provide 1 us delay. > > Hi, > > If you submit one million URB's per second, you will have a significant system > overhead? Right? Have you ever tried this application under FreeBSD 8+ ? There > we allow so-called multi-frame transfers, that means you sort of bind > toghether multiple jobs into a single URB submit. > > For example you can submit 100 2-byte short-packet-jobs in a single submit, > and have only one completion interrupt from the EHCI controller! > > For the XHCI you can even double buffer such transfers. The problem is that Alexey needs to use the data returned by the previous transfer in order to compute the data for the next transfer. Therefore the transfers must be made sequentially; they cannot be done concurrently. > This discussion gets me thinking of once i stress-tested some USB code, by > hacking into the EHCI DMA schedule, repeating each control request 128 times > or more. This allows for submitting more than 8000 control requests per second > for the EHCI controller and can be useful to find timing related issues in the > device side. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
