Re: 2.4 vs 2.6 USB mass storage device performance

[Alan Stern <stern@xxxxxxxxxxxxxxxxxxx>]

On Fri, 30 Nov 2007, syed khader wrote:

> Alan,
>   I have added these routines to 2.4.35 usb code. I
> get around 50% improvement in the performance in the
> throghput:
> #./sg_dd if=/dev/sda of=/dev/null bs=136k count=10000
> iflag=direct sync=1 time=
> 1
> time to transfer data: 85.233351 secs at 16.34 MB/sec
>   which is close to 18MB/sec reported by 2.6 usb stack
>    
>   I am attaching patch against the 2.4.35 containing
> my changes. 
> Apart from those apis which you mentioned there is
> this API "fill_inquiry_response()" which also uses
> usb_sg* routines. What does this API do?

It makes up a fictitious reply to an INQUIRY command.  Under 2.4 it
shouldn't need to use the usb_sg* routines.

> how does it
> effect the performance?

It doesn't affect performance at all.

> I did change this API too to
> use usb_sg routines but couldn't test it as it never
> gets called while I do scsi xfer.

That's why it doesn't affect the performance!  :-)

>   Could you comment about the patch I sent?

It basically looks okay.  You can take out the parts used only by the 
fill_inquiry_response() routine.

An important test you should make is to see how well the new driver 
behaves when you unplug a USB drive in the middle of a data transfer.

Alan Stern


-------------------------------------------------------------------------
SF.Net email is sponsored by: The Future of Linux Business White Paper
from Novell.  From the desktop to the data center, Linux is going
mainstream.  Let it simplify your IT future.
http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4
_______________________________________________
linux-usb-devel@xxxxxxxxxxxxxxxxxxxxx
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel