This message is from the T13 list server.
Pat & Co.
I have observed slower drives failing when a faster drive is the
slave. At the time I reached the conclusion that because the master has to
be online for all register accesses, even when the slave is currently
selected, it can get hung because of the faster access speeds. I observed
this behavior for both normal register access and PIO data transfers.
The issue also goes a bit deeper than PIO modes. Some drives
reported support for both PIO 2 and PIO 4, but when you looked at their
cycle times in the ID DEVICE data you found numbers in excess of 300ns; not
the usual 180 or 120... This means that although both drives may report
Mode 4 capability, you still must run at the slower speed. In this case you
were really running closer to Mode 1...
So, the answer to the question is that you really have to run at the
slowest cycle time regardless of the mode capability.
-----------------------
Curtis E. Stevens
Pacific Digital Corp.
2052 Alton Parkway
Irvine, CA 92606
Phone (949) 477-5713
Fax (949) 252-9397
E-Mail: [EMAIL PROTECTED]
WEB: www.PacificDigital.com
Never take life seriously... after all, nobody ever gets out alive!
-----Original Message-----
From: Pat LaVarre [mailto:[EMAIL PROTECTED]]
Sent: Monday, October 07, 2002 9:23 AM
To: [EMAIL PROTECTED]
Subject: RE: [t13] Master and Slave speed
This message is from the T13 list server.
> when DMACK is asserted by the host
> side, the address signals are "disabled" and not used.
Do the public specs carefully require DMACK to be
asserted with enough setup for a merely PIO 0 device
to know to ignore the address signals before the host
begins using them? And on the other end, do the specs
require the address signals to go quiet early enough
before DMACK is deasserted?
Curiously, lazily yours. Pat LaVarre
-----Original Message-----
From: Hale Landis [mailto:[EMAIL PROTECTED]]
Sent: Mon 10/7/2002 9:43 AM
To: [EMAIL PROTECTED]
Cc: Stephane Cattaneo; Alistair LIVESEY; [EMAIL PROTECTED]
Subject: Re: [t13] Master and Slave speed
This message is from the T13 list server.
On Mon, 07 Oct 2002 17:30:11 +0200, Stephane Cattaneo wrote:
>This message is from the T13 list server.
>Thanks for your answer. Anyway, let's say that my two devices
accept pio mode
>4 (a hdd capable to run up to udma mode 4 and a dvd player capable
to run up
>to pio mode 4), so the max speed for pio mode. (this is normally
the case for
>all device present in the market).
In this example you may use PIO mode 4 to talk to both devices (for
both register r/w and for PIO data transfers). The DMA mode of both
device need not match - the HDD can use UltraDMA mode 4 while the
DVD
can use MW or UltraDMA (it need not also use UltraDMA mode 4 is the
point). Why is this true? Because when DMACK is asserted by the host
side, the address signals are "disabled" and not used. While DMACK
is
asserted the device that is not selected shall ignore the address
and
IOR/IOW signals.
Warning (because if I don't say this Pat will correct me): There are
some *VERY* old devices that do not support DMA and don't even
decode
the DMACK signal. These devices may do strange things unless the
host
keeps the address signals deasserted while DMACK is assested. This
should not be a problem on any modern host (a host that supports
UltraDMA).
>Because the commands are sent in pio timings (mode 4 in this case),
I think
>that I can communicate in udma mode 4 with the hdd, and pio mode 4
with the
>dvd player. Do you agree that the read and write with the hdd will
not
>perturb the dvd player ?
Yes.
>Same question when I use the same hdd, plus another hdd capable to
run up to
>mode 2 instead of a dvd player.
In this case you *SHALL* use PIO mode 2 when doing register r/w and
PIO data transfers with both devices. But again if you use DMA with
these two HDD you could use different DMA protocols and/or modes.
*** Hale Landis *** www.ata-atapi.com ***
