> I've downloaded a large file today and needed to move
> it to my Windows partition. Since it's NTFS, no
> available FAT32 partitions and mkisofs for some
> reason messing up the ISO it's supposed to create,
> I've figured to use my USB stick.
> 
> The thing is, it took almost 30 minutes to copy 2
> gigabytes, but under Windows, it copied them back to
> the disk in under two minutes. As there isn't such a
> crass difference of 14:1 between reads and writes on
> flash drives, I'm assuming that the ports are set to
> USB 1.1 (speeds) or something under Solaris,
> explaining the really low performance.
> 
> To be sure however, I have to find out, so I can file
> a bug report about this.

Check the /var/adm/messages file, if you see something like
this, it's using USB 2.0 speeds:

Oct 23 10:35:44 tiger2 usba: [ID 912658 kern.notice] USB 2.0 device 
(usbd49,3210) operating at hi speed (USB 2.x) on USB 2.0 root hub: storage at 
4, scsa2usb0 at bus address 2
Oct 23 10:35:44 tiger2 usba: [ID 349649 kern.notice]    Maxtor Corporation 3200 
604010120544



An USB 2.0 flash memory stick connected on an old USB 1.x hub
would use the slow USB 1.x protocol, and would be reported like this:

Oct 29 14:31:33 tiger2 usba: [ID 912658 kern.notice] USB 2.0 device 
(usb1976,6025) operating at full speed (USB 1.x) on USB 1.10 external hub: 
storage at 4, scsa2usb4 at bus address 7
Oct 29 14:31:33 tiger2 usba: [ID 349649 kern.notice]    Alliance Flash Disk 
17164900F0B50800




Btw. it seems certain low-cost usb 2.0 flash memory devices
are only able to write ~ 3 mbyte/second, and can read
~ 10 mbyte/second (3:1).  And some devices seem to use an
internal block size of e.g. 8kbyte, and when you write using
smaller units, write transfer rates become really bad.

E.g. with one flash memory stick, I get

3000 kb/sec, writing with "dd bs=64k" (or 16k or 8k) to the raw device
 531 kb/sec, writing with "dd bs=4k" !

Of cause, Solaris' pcfs seems to use the small units
when writing the media, so performance on that flash
memory stick is *really* bad (20:1).
 
 
This message posted from opensolaris.org

Reply via email to