On Sunday 06 May 2007 17:51, Donn Washburn wrote: > Rajko M. wrote: > > On Sunday 06 May 2007 13:15, Donn Washburn wrote: > >> Has anyone noticed this problem? It maybe a problem with hdparm dealing > >> with the sda drive and not hda > > > > Can you give output of hdparm that is making trouble to you. > > The transfer rate of 66 interface is lower than the 66 MB/s. > > How much lower depends on many factors. > > Just trying to change the to 32 (1) bit from the default 16 (0) > # hdparm -c1 /dev/sda > /dev/sda: > setting 32-bit IO_support flag to 1 > HDIO_SET_32BIT failed: Invalid argument > IO_support = 0 (default 16-bit) > # > > It seems to be a "sda" related issue and not hdparm. > It happens on three different machine running 10.3Alpha2 > > Any more questions let me know > > I did also upgrade hdparm in a effort to see if it was hdparm or the sda > issue
I guess that you found Sid's mail. The sdparm is not applicable. You said that they are IDE drives on older 500 MHz system, so hdparm should be right application. I'm not familiar with hdparm internals so I ran test on my older computer with 10.3 alpha 1 and it shows the same error, although I used symlink hda that points to sda ie. solving naming: hdparm -c1 /dev/hda That might be because the device ID major is changed from 3 to 8. Which probably confuses hdparm. The hdparm -i /dev/hda claims that parameter is not applicable for the device. The same works fine in 10.2. The speed test hdparm -t /dev/hda shows speed of 55 MB/s for ATA 100 drive, which seems to be right. For details on transfer rates http://www.realworldtech.com/page.cfm?ArticleID=RWT011701000000 The difference between 16 and 32 transfer is non existent on older machine, both show 55 MB/s. The reason is probably oldfashioned parallel cable that has 16 lines for data transfer and using 32 bit brings no improvement, except on very old systems where CPU load is 1/2 if it can send 32 bits to controller at once. -- Regards, Rajko. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
