root@conexch.UUCP (Larry Dighera) (10/14/88)
Can anyone tell me how to determine which 80386 chips have the erratum 21 problem? This only manifests itself in systems with a '387 coprocessor using the paging mode and activating the PEREQ line. Surely there must be some way of identifing these chips without having to test them. Intel hasn't been too helpful yet. They either don't return my calls or don't know the answer. Larry Dighera -- USPS: The Consultants' Exchange, PO Box 12100, Santa Ana, CA 92712 TELE: (714) 842-6348: BBS (N81); (714) 842-5851: Xenix guest account (E71) UUCP: conexch Any ACU 2400 17148425851 ogin:-""-ogin:-""-ogin: nuucp UUCP: ...!uunet!turnkey!conexch!root || ...!trwrb!ucla-an!conexch!root
mslater@cup.portal.com (Michael MPR Slater) (10/22/88)
Intel claims to now be shipping the D stepping of the 386, which is supposed to fix all known bugs, including errata 21. You can identify a "D step" chip by the marking 80386DX -- the DX means it is a D step chip. Michael Slater, Microprocessor Report mslater@cup.portal.com 415/494-2677
smith@iwblsys.UUCP (Mickey Smith) (10/25/88)
In article <10323@cup.portal.com>, mslater@cup.portal.com (Michael MPR Slater) writes: > Intel claims to now be shipping the D stepping of the 386, which is supposed > to fix all known bugs, including errata 21. You can identify a "D step" chip > by the marking 80386DX -- the DX means it is a D step chip. > > Michael Slater, Microprocessor Report mslater@cup.portal.com 415/494-2677 Mike, The 80386 family has taken on the DX suffix to indicate that the part has a 32 bit external bus (80386DX), it just so happens that the part is also a 'D' step part. The 80386SX is the same 80386 core with and external 16 bit data bus and 24 bit address bus. -- Mickey Smith, Regional Software Specialist | Intel Corporation | 7071 Orchard Lake Road | W. Bloomfield, MI 48332 (313) 851-8096 |
mslater@cup.portal.com (Michael Z Slater) (10/27/88)
>The 80386 family has taken on the DX suffix to indicate that the part >has a 32 bit external bus (80386DX), it just so happens that the >part is also a 'D' step part. The 80386SX is the same 80386 core with >and external 16 bit data bus and 24 bit address bus. I was told by the PR folks at Intel that any part marked DX was a D-step part. Are they confused? Are there any DX parts that are not D-step? Or are you saying that the DX notation was added when the D-step was released, but that it is just coincidence that the letter "D" is used? Michael Slater, Microprocessor Report mslater@cup.portal.com 415/494-2677
root@conexch.UUCP (Larry Dighera) (12/26/88)
In article <245@lakesys.UUCP> steven@lakesys.UUCP (Steven Goodman) writes: > If this bug appears and you happen to have a Compaq, contact your >dealer. Compaq will ( free of charge ) add a small board which plugs into >your 386 socket which will fix the "errata 20" bug. > >Not wanting to repeat what might have alreading been said on this subject but >this bug only accurs when using a 387 chip and you can boot up your system >to ignore the 387 until you receive some kinda fix. From what I understand >this is NOT fixable via software. Intel is now shipping '386s that have the Erratum 21 bug fixed. You can identify the fixed part by the DX suffix in its part number: i80386DX. Bell Technology, Ironwood Electronics Inc., (612) 431-7025, and others I suspect, offer hardware fixes for this Intel problem. I have not used any of these products, so I cannot personally endorse any of them. It would seem reasonable to approach Intel about replacing a buggy chip with one of their new DX parts. How could they refuse? Larry Dighera -- USPS: The Consultants' Exchange, PO Box 12100, Santa Ana, CA 92712 TELE: (714) 842-6348: BBS (N81); (714) 842-5851: Xenix guest account (E71) UUCP: conexch Any ACU 2400 17148425851 ogin:-""-ogin:-""-ogin: nuucp UUCP: ...!uunet!turnkey!conexch!root || ...!trwrb!ucla-an!conexch!root
tanner@ki4pv.uucp (Dr. T. Andrews) (12/26/88)
In article <17383@conexch.UUCP>, root@conexch.UUCP (Larry Dighera) writes:
) It would seem reasonable to approach Intel about replacing a buggy
) chip with one of their new DX parts. How could they refuse?
I suppose that they could refuse in the same way that they refused to
replace the parts which didn't work in "native" mode. If they refuse
to replace an alleged "32-bit" CPU which doesn't work in 32-bit mode,
surely they can refuse to replace for something minor like this!
--
...!bikini.cis.ufl.edu!ki4pv!tanner ...!bpa!cdin-1!cdis-1!ki4pv!tanner
or... {allegra killer gatech!uflorida decvax!ucf-cs}!ki4pv!tannered@imuse.uucp (Ed Braaten) (12/31/88)
In article <7067@ki4pv.uucp>, tanner@ki4pv.uucp (Dr. T. Andrews) writes: > I suppose that they could refuse in the same way that they refused to > replace the parts which didn't work in "native" mode. If they refuse > to replace an alleged "32-bit" CPU which doesn't work in 32-bit mode, > surely they can refuse to replace for something minor like this! Could you please clarify? Are you referring to the 32-bit multiply bug in older 80386's? Who refused you a replacement? ------------------------------------------------------------------------- Ed Braaten | "I will destroy the wisdom of the wise; Intel Semiconductor GmbH | the intelligence of the intelligent I Dornacher Strasse 1 | will frustrate." 1 Corinthians 1:19 8016 Feldkirchen bei Muenchen | West Germany | uucp: ed@imuse.uucp Ph: +49 89 90992-426 -------------------------------------------------------------------------