On Sunday 30 November 2003 07:29 pm, Edward Dekkers wrote:
Pete Nesbitt wrote:
On November 28, 2003 04:07 pm, Reuben D. Budiardja wrote:
Hi all, Is it true that secondary IDE is slower than Primary? I have a brand new Western Digital 120 GB 7200 RPM HD, and because I don't have a place on the primary IDE, I unplug the (unused) CD-ROM and use the secondary IDE as the harddrive connection. I checked with "hdparm -tT" and the Timing buffered disk reads is horrible: about 25 MB/sec. This is of course in linux single mode. On my other machine, the same WD HD (80 GB) on primary IDE gives me almost twice the troughput. I checked all options that can be enabled in hdparm to be enabled (multcount, dma, 32-bit I/O) , so I ended up have all the same options enabled for those 2 system, but the one with the HD in secondary IDE has considerably slower I/O.
Anything that I might overlook? Any info is greatly appreciated.
RDB
Hi Reuben, There is no reason the secondary controller would be any slower, if they are both the same. I have seen board where the primary controller supported EIDE or ATA but the secondary only supported regular IDE. You could see if the cable is a 40 or 80 wire (I think all the ata66 and higher use the 80 wire cables). Is it the master or single drive on the chain?
Presuming it is not using a regular older IDE cable like the one the CD would have been using, you should take the new drive, and mount it in place of where your 80 GB is on the other machine. It may be a cabling issue or a bad controller. The best way to test hardware, except power supplies, is to put the suspect part into a known good environment. that should give you some definitive information. If the drive is good, try swapping the cables onto the good machine.
And a side note to that - make sure the big difference you're seeing is
not because the 80Gb is a JB model and the 120 a BB model.
Uhhm..would you care to elaborate what JB / BB means?
Thanks.
The BB models from Western Digital have a 2Mb cache on the hard drive. The JB models have 8Mb cache. Pitting one model against the other will yield a large difference in results, with the JBs being a lot faster.
Regards, Ed.
-- redhat-list mailing list unsubscribe mailto:redhat-list-request@xxxxxxxxxx?subject=unsubscribe https://www.redhat.com/mailman/listinfo/redhat-list
