Very Computer

by **Bill Campbe** » Wed, 16 Aug 1995 04:00:00

There was an article in the Info World I received yesterday telling of
problems with IDE controllers sold with many Intel main boards, and
that this would cause data corruption PARTICULARLY WITH 32 BIT
MULTITASKING OPERATING SYSTEMS.

I don't have the article here at the office so I can't post the name
of the particular IDE controller vendor.

The article claimed that Intel admits knowing about this for over a
year now and hasn't said anything. Let's see now. First we hear of
the floating point math bugs (which Intel claims really weren't bugs).
Then it's the Triton chipset which doesn't allow parity checking
memory (which Intel says we don't need because errors are rare (and
not parity checking saves them money)). Now we have data corruption
with IDE controllers. All this sure makes Sparcs look better to me.

This is Yet Another Reason to use SCSI for Unix systems.

Bill
--

UUCP: camco!bill 2835 82nd Avenue S.E. S-100
FAX: (206) 232-9186 Mercer Island, WA 98040; (206) 947-5591
http://www.celestial.com/
SPEED COSTS MONEY -- HOW FAST DO YOU WANT TO GO?

by **Lawrence Kirb** » Thu, 17 Aug 1995 04:00:00

Quote:>There was an article in the Info World I received yesterday telling of
>problems with IDE controllers sold with many Intel main boards, and
>that this would cause data corruption PARTICULARLY WITH 32 BIT
>MULTITASKING OPERATING SYSTEMS.

>I don't have the article here at the office so I can't post the name
>of the particular IDE controller vendor.

PC Tech RZ1000 used on Intel Plato/Premiere motherboards (possibly others but
not the latest Triton based ones).

Quote:>The article claimed that Intel admits knowing about this for over a
>year now and hasn't said anything. Let's see now. First we hear of
>the floating point math bugs (which Intel claims really weren't bugs).
>Then it's the Triton chipset which doesn't allow parity checking
>memory (which Intel says we don't need because errors are rare (and
>not parity checking saves them money)). Now we have data corruption
>with IDE controllers. All this sure makes Sparcs look better to me.

This is being discussed on various newsgroups like comp.sys.intel. There
was a list posted of OSs which did and didn't exhibit problems. Unixware
was listed but not SCO. I hope SCO are looking into the problem.

A possible workaround is to disable the IDE prefetch buffers in the system
BIOS. According to the OS/2 guy who posted this it results in about a 1%
performance decrease so the bug may not be all that disasterous.

Quote:>This is Yet Another Reason to use SCSI for Unix systems.

It looks that way!

--
-----------------------------------------

-----------------------------------------

by **Dave Edmondso** » Fri, 18 Aug 1995 04:00:00

after reading information about the problem we have
determined that neither open desktop 3.0 nor open server 5.0
are affected by the intel ide chipset bug.

dave.

Very Computer

Beware the IDEs of Intel

Beware the IDEs of Intel

Beware the IDEs of Intel

Beware the IDEs of Intel