Hard drive cable question -

Hans Kristian Rosbach hk at isphuset.no
Thu Feb 23 11:18:13 UTC 2006


On Thu, 2006-02-23 at 17:03 +1100, Wolfgang Gill wrote:
> On Thu, 23 Feb 2006 13:39:13 +1030, Tim wrote
> > On Thu, 2006-02-23 at 11:12 +1100, Wolfgang Gill wrote:
> > > Nope it doesn't matter. Drive selects are done with jumpers on the
> > > drive. All IDE 40/80 pin cables are straight through, so the drive
> > > itself govens whether it's master or slave. When I only have one
> > > drive, I use the middle connector, and hide the other one out of the
> > > way, to give me better casing air flow.
> > 
> > That is just so much misleading, not to mention outright wrong,
> > information.  Kindly stop leading people down the garden path.
> 
> Misleading?? That wasn't misleading at all!! And I'M NOT leading people up the
> garden path. I've build literally 10,000 of machines (Probably more, lost
> count after the first 1000 or so). And only 1% have failed due to hardware
> faults, and cabling wasn't one of them. It's NO use to explain things into
> GREAT detail to people that have little understanding of the concepts as it
> is, and confuse them even more. And as I understand it, the original poster
> has solved it, so it not necessary to slag everyone that tries to help.

I'd have to agree that you were misleading.

I used to leave the disk on the middle connector and then stuff the last
connector somewhere out of sight. That works really bad.
I detected how bad it was when a linux machine continuously complained
about ECC errors on the cable. After connecting the disk to the last
connector it was just fine (After recommendation from linux-ide ppls).

This also seemed to improve disk throughput on some machines, since
they seemingly had to cut down to udma-33 speeds OR retransmit silently
in order to maintain signal integrity. This does not happen in all cases
but doing it right hasn't failed yet.

-HK




More information about the users mailing list