Subject: Re: [linux-audio-user] Re: OT: ide drives
From: Mark Knecht (markknecht_AT_gmail.com)
Date: Fri Oct 22 2004 - 01:21:24 EEST
On Thu, 21 Oct 2004 23:47:11 +0200, Robert Epprecht <epprecht_AT_solnet.ch> wrote:
> Mark Knecht <markknecht_AT_gmail.com> writes:
>
> > On Thu, 21 Oct 2004 19:43:16 +0100, anahata <anahata_AT_treewind.co.uk> wrote:
> >> On Thu, Oct 21, 2004 at 07:44:11PM +0200, David Baron wrote:
>
> >> > was told that the slower CD device will slow down the HD.
>
> >> It used to be true. It was often misrepresented as the CD slowing the HD
> >> down to its speed
>
> > No - this actually is true. An EIDE controller that is compliant with
> > the ATAPI specs will recognie that one drive operates with the cable
> > going 100MHz, while the controller recognizes that the CDROM only
> > works when the cable goes 33MHz or 66MHz. since the two drives are on
> > the same cable the EIDE controller sets the cable speed at 33MHz and
> > therefore you get less throughput from the hard drive. I wasn't aware
> > until about 2 years ago that the controller had to do this. There are
> > some good white papers somewhere on the Seagate or Maxtors sites if I
> > remember correctly.
>
> So you say, that it is the *controller* who does it?
> It can not be influenced (set) by the kernel, by hdparm or something?
No - it's done in hardware. If one drive REQUIRES that the cable run
at 33MHz, and the other drive can operate at 33/66100MHz, then the
controller sets the cable at 33MHz and leaves it there. The EIDE cable
cannot change frequency while operating.
I think the good news is that your equipment looks new enough that you
will likely not have this problem. It's more an issue when mixing new
and old hardware.
>
> This sounds like it could be easy to figure it out with dmesg, hdparm
> or something comparing results with and without connecting the optical
> drive? What exactly should i look for?
Sure - run:
hdparm -tT /dev/hda
with and without the optial drive attached. (Optical should be /dev/hdb)
Do it again with the second drive using
hdparm -tT /dev/hdc
>
> Robert
>
This archive was generated by hypermail 2b28 : Fri Oct 22 2004 - 01:23:05 EEST