home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.os.ms-windows.misc
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!saimiri.primate.wisc.edu!ames!pacbell.com!att-out!cbfsb!cbnewsg.cb.att.com!rnichols
- From: rnichols@cbnewsg.cb.att.com (robert.k.nichols)
- Subject: Re: 32 bit multiplication not reliable ???
- Message-ID: <1992Nov7.185433.20112@cbfsb.cb.att.com>
- Summary: Early 80386-16 devices bad in OS/2 2.x or Windows Enhanced mode
- Keywords: 386/16, processor faults
- Sender: news@cbfsb.cb.att.com
- Organization: AT&T
- References: <kaustell.80@viikki.Helsinki.FI>
- Date: Sat, 7 Nov 1992 18:54:33 GMT
- Lines: 64
-
- In article <kaustell.80@viikki.Helsinki.FI> kaustell@viikki.Helsinki.FI (KIM OLAVI KAUSTELL (MATEK)) writes:
- ...
- >I recently got my hands on a used WYSE 386/16 PC with 4M RAM, 1M Trident,
- >68+20M hard disk and Intel 387/25. Now that I installed Win 3.1, I get this
- >very annoying WARNING from WIN (not exactly these words...):
- >
- > "Your Intel 80386 processor is not reliable on 32 bit
- > multiplications. Usually Windows can handle these situations,
- > but You might want to replace the processor.
- ...
- >I can imagine there would exist some system.ini -setting like "disable 32
- >bit multiplications=1" or something.
- ...
-
- Some history:
- (Source: Mueller, "Upgrading and Repairing PCs," 2nd Edition,
- Que, 1992)
-
- Some early Intel 16MHz 386DX processors had a bug in the chip's 32-bit
- multiply routine. The problem manifests itself only when you are
- running true 32-bit code, such as OS/2 2.x, MS-Windows in Enhanced
- mode, Unix/386, or some specialized 386 memory management systems. The
- bug usually causes the system to lock up, but is difficult to diagnose
- because it is intermittent and software-dependent.
-
- Only Intel, with proper test equipment, can determine whether your chip
- has the bug. Some programs can identify the problem in some, but not
- all, defective chips. If a program indicates a bad chip, the chip is
- certainly defective; if the program passes the chip, it could still be
- defective.
-
- Intel requested that its 386 customers return possibly defective chips
- for screening, but many vendors did not. The known defective chips
- were later sold to bargain liquidators or vendors that did not require
- chips capable of running 32-bit code. The known defective chips were
- stamped with a "16-bit SW Only" logo.
-
- Chips that passed the test, and subsequently produced chips, were
- marked with a double sigma logo. 386DX chips not marked with either
- the "16-bit SW Only" or the "EE" logo (sorry, my keyboard doesn't have
- a Sigma key) have not been tested by Intel and might be defective.
-
- This problem was discovered and corrected before Intel officially added
- the "DX" to the part number. If you have a chip labeled as "80386DX"
- or "386DX" it does not have this problem.
-
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
-
- It would appear from your description that Windows 3.1 has code that
- attempts to detect the problem and has determined that your chip has
- the bug. Your options are:
-
- (a) Replace the CPU chip (difficult if it's not socketed),
-
- (b) Run Windows in Standard mode only.
-
- If you run Windows in Enhanced mode, you will probably experience
- intermittent, unexplained lockups.
-
- Sorry about that.
-
- Bob Nichols
- AT&T Bell Laboratories
- rnichols@ihlpm.ih.att.com
-