#9593 LOW Future : XO-1 F11 slow USB transfer rate

[Zarro Boogs per Child]

#9593: XO-1 F11 slow USB transfer rate
--------------------+-------------------------------------------------------
 Reporter:  Quozl   |                 Owner:  dsaxena       
     Type:  defect  |                Status:  new           
 Priority:  low     |             Milestone:  Future Release
Component:  kernel  |               Version:  not specified 
 Keywords:          |           Next_action:  reproduce     
 Verified:  0       |   Deployment_affected:                
Blockedby:          |              Blocking:                
--------------------+-------------------------------------------------------
 A SanDisk Cruzer Colors 4GB USB key (3926083 blocks) was tested on several
 builds, using a command in Terminal to time the read of the whole device
 while writing the data to /dev/null.

 ||'''model'''||'''build'''||'''time'''||'''rate'''||'''shell'''||
 ||XO-1.5||os33||4min 9sec||15.4MB/s||Terminal||
 ||XO-1.5||os33||4min 7sec||15.5MB/s||Terminal||
 ||XO-1.5||os33||4min 6sec||15.6MB/s||Console||
 ||XO-1.0||os8||6min 10sec||10.4MB/s||Terminal||
 ||XO-1.0||os8||6min 20sec||10.1MB/s||Terminal||
 ||XO-1.0||802||3min 57sec||16.1MB/s||Terminal||
 ||XO-1.0||802||3min 46sec||17.0MB/s||Console||
 ||XO-1.0||debxo 0.4||4min 6sec||15.6MB/s||Console||
 ||Desk 3GHz P4||Debian 2.6.26-1-686||2min 58sec||21.5MB/s||xterm||

 The difference between the rates for 802 vs os8 suggests a USB performance
 regression between the F9 and F11 kernels on the XO-1.0.  If the
 regression is due to kernel, then there is an opportunity to increase
 performance on the XO-1.5.

 (incidentally, the internal SD on two XO-1.5 achieved 16.2MB/s over 3min
 52sec, an external SD on an XO-1.0 achieved 16.2MB/s, and on an XO-1.5
 achieved 16.1MB/s ... no issue)

-- 
Ticket URL: <http://dev.laptop.org/ticket/9593>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system