Date: Sun, 26 Dec 2004 22:43:37 -0800 From: Julian Elischer <julian@elischer.org> To: ticso@cicely.de Cc: freebsd-usb@freebsd.org Subject: Re: How to mount usb2 devices Message-ID: <41CFAF19.8090502@elischer.org> In-Reply-To: <20041226131607.GJ58581@cicely12.cicely.de> References: <41CD83E6.1010304@gmail.com> <20041225181725.GE58581@cicely12.cicely.de> <41CDEDE2.5030900@gmail.com> <20041226002721.GG58581@cicely12.cicely.de> <41CE5B7F.3090805@elischer.org> <20041226131607.GJ58581@cicely12.cicely.de>
next in thread | previous in thread | raw e-mail | index | archive | help
Bernd Walter wrote: > On Sat, Dec 25, 2004 at 10:34:39PM -0800, Julian Elischer wrote: > >>Bernd Walter wrote: >> >>>On Sat, Dec 25, 2004 at 11:46:58PM +0100, Darksidex wrote: >>> >>> >>>>Bernd Walter wrote: >>>> >>>> >>>> >>>>>It is running high speed, because it is attached to your ehci >>>>>controller. >>>>>Having high throughput is another story - msdosfs is not very efficient >>>>>with some devices - you might want to check a dd with big block sizes >>>>>to verify if higher speed can be reached. >>>>> >>>>> >>>> >>>>How can I do it? >>> >>> >>>dd if=/dev/da-whatever-number-it-is of=/dev/null bs=64k >>>You may want to test higher values for bs= too. >> >>I think that there is a lot of work that can be done to speed up >>USB2 devices.. > > > Likely - there is still a lot that can be done even for full speed > devices. > > >>for example Ugen devices can be sped up by an order of magnitude >>with a simple set of changes. (that I will commit when I clean >>them up) I have to guess that the Umass driver may have similar >>performance problems. > > > So you can speed up ugen devices - sounds interesting. > What is the basic idea behind? ehci allows only 4000 interrupts per second. ugen packages data in 1024 byte chunks and does one per interrupt. Previously it allowed only 1000 interrupts per second, limiting you to 1MB/sec.. Now you can get 4MB/sec. however allowing internal transfers of 64KB per transfer allows 64KB*4000 = a lot more. :-) In my original example, throughput went from 1MB/sec to 16MB/sec which was the limit the device would send. With a faster device you should be able to get more throughput.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?41CFAF19.8090502>