This message is from the T13 list server.

Pat,

People have shipped and used a billion UDMA ports for several years now.
Yes, this is primarily with hard drives connected to them - our industry
alone adds another 150 million products (and thus ports) per year to the
experience base, PCs twice as much.  While used with 512 byte blocks (which
is not "outside the standard" but is clearly documented in the ATA
standard), the protocol itself is block size independent.  My point is
simply that UDMA hardware works quite well.

I don't see how people screwing up their bridge implementations is an issue
- people are always screwing up product implementations.  There are
perfectly functional implementations shipping today using UDMA for CD-ROM
devices, so there is not a fundamental protocol problem.  I can't quote
volumes, but that's just because its not my industry.

The problems you keep on bringing up appear to be located in areas other
than ATA - like driver software.  I'm not saying that CD-ROM driver software
does not have any problems.  Just that they are not ATA problems.  And in
particular changing the ATA standard is not the right path to take, and
indeed would just make a further hash of things (if the driver writers
cannot design to something with this sort of shipping history, then I can't
see how changing things does anything but set the clock back further).  The
worse thing a standard can do is to provide the industry with a moving
target.

Jim



-----Original Message-----
From: Pat LaVarre [mailto:[EMAIL PROTECTED]]
Sent: Friday, February 15, 2002 1:33 PM
To: [EMAIL PROTECTED]
Subject: [t13] a billion UDMA ports just use Pio?


This message is from the T13 list server.


> Subject: [t13] but we say the disagreeable hosts are broken 
> From: "Mcgrath, Jim" <[EMAIL PROTECTED]> 
> Date: Wednesday - January 30, 2002 8:00 PM
...
> Undoubtedly a lot of people are filtering this input
> through their own experience of a billion UDMA ports
> in use and not ... actual problem.

Funny how well things work until you turn them on.

Familiar to me are specific examples of AtapiUDma devices that work only
because the driver in question forces the use of Pio for anything less
normal than read/write of blocks of a large power of two size agreed in
advance.  The driver authors in question learned to "just use Pio" back when
they saw SwDma break.

What was new about AtapiDma in late 2001?  Streams above the 17e+6 byte/s of
Pio4 became common enough that Atapi people began to care about UDma.  Yea,
sure, Marketing people have cared for years - but that was just a bullet.
You didn't have to make UDma work always, you just had to let it move blocks
of the fixed size (x200) agreed in advance outside of the standard.

Life is changing now.

People have already shipped 1394a/UDma and Usb2/UDma bridges that don't just
plain work, not when viewed from this light.

Pat LaVarre


Subscribe/Unsubscribe instructions can be found at www.t13.org.
Subscribe/Unsubscribe instructions can be found at www.t13.org.

Reply via email to