Using an alternative geometry

[robh]

Some large disks (like WD10EARS) have sectors of 4KB, but report themselves as having sectors of 512B. The disk functions properly with DANIS506. Documentation mentions that for optimal performance a track must have a multiple of 8 (512B) sectors. With DFSee I could change the logical geometry to H=252 and S=120, but this seems to be ignored by DANIS506, although the BPM/BIOS column shows the changed value for H, but not for S! It would be nice to have the possibility to specify the geometry.

Unfortunately without thie feature it is impossible to say if it would have any effect on the disk (write) performance!

Output of 'type IBMS506$' for this unit below:

Controller:1  Port:9000 IRQ:17  Status:OK BusMaster Scatter/Gather
  Intel ICH7 SATA host (8086:27C0 rev:01) on PCI 0:31.2#1
 Unit:0 Status:OK SMS:16 LBA NL:126 BusMaster UltraDMA6/PIO4 BIOS
  Model:WDC WD10EARS-00Y5B1                     80.00A80
   OS2:log   phys   BPB/BIOS   IDE:log   phys     Total Sectors
  C  60321  65535                65535  16383   Avail 1953525168
  H    255     16        252        16     16   OS2   1953495585
  S    127     63         63        63     63   % Used     99.99

[Steven Levine]

Grumble. No TRAC e-mails or I would have replied sooner.

Danis506 should use the BPB geometry if there are no conflicts. Exactly how did you adjust the geometry?

[robh]

Forgot to check the follow up of this request...

As far as I remember, this is the sequence with DFSee:

• select disk
• wipe beginning of disk
• reboot
• select disk
• geo ? 252 120 (255 heads would be OK too, I think)
• fix CHS values to geometry
• set LVM geo to dfsee L-geo
• reboot

In October 2010 I discussed this scenario with Jan van Wijk and he agreed (although he couldn't check because he didn't own a disk of over 500 GB).