Usenet_area_"Cs.I.Pc"@watmath.waterloo.edu (01/07/88)
From Usenet: watmath!codas!karthur From: karthur@codas.att.com (Kurt_R_Arthur) Newsgroups: comp.sys.ibm.pc Subject: Re: 80386 Test Suite? Message-ID: <1506@codas.att.com> Date: 8 Jan 88 15:11:41 GMT References: <2334@tekigm2.TEK.COM> Distribution: usa Organization: AT&T, Altamonte Springs, FL Lines: 33 Xref: watmath comp.sys.ibm.pc:10371 comp.sys.misc:969 In article <2334@tekigm2.TEK.COM> timothym@tekigm2.TEK.COM (Timothy D Margeson) writes: > Having just ordered a QIC 80386 computer, I want (need) a test suite to > verify that the installed 80386 is one without the 16 bit limitations. > > Also, if you know of the bugs within the early 386's, please reply also. If > no one has the test suites that can tell me, I have the new v5.0 MASM so can > write my own if I know what to look for. According the sources I have read: 1. The 80386 bug is an error in 32-bit multiplication operation. This CANNOT arise in real-life unless you are using a '386 32-bit operating system. I'm almost positive it can't show up even if you are using one of the DOS-extenders. 2. Intel has not authorized or validated any software suite to check for the existence of the bug (corrections, please), and uses a hardware test when they wish to check for the bug. 3. Intel had the problems very early in the production cycle, fixed the chip mask in Spring, 1987, and (supposedly) shipped nothing but "good" chips since last April or May. Therefore, the odds are excellent you own a good chip. If you are REALLY worried about the chip, call Intel and get a list of the production numbers affected by the bug, and compare the list to the markings on your cpu. Regards, Kurt Arthur Software Services of Florida, Inc. --- via UGate v1.6 * Origin: watmath (221/163)
fyl@ssc.UUCP (Phil Hughes) (01/14/88)
In article <16211@watmath.waterloo.edu>, 221.162.fido!Usenet_area_"Cs.I.Pc"@watmath.waterloo.edu writes: > From Usenet: watmath!codas!karthur > From: karthur@codas.att.com (Kurt_R_Arthur) > Newsgroups: comp.sys.ibm.pc > Subject: Re: 80386 Test Suite? > Message-ID: <1506@codas.att.com> > Date: 8 Jan 88 15:11:41 GMT > References: <2334@tekigm2.TEK.COM> > Distribution: usa > Organization: AT&T, Altamonte Springs, FL > Lines: 33 > Xref: watmath comp.sys.ibm.pc:10371 comp.sys.misc:969 < < In article <2334@tekigm2.TEK.COM> timothym@tekigm2.TEK.COM (Timothy D < Margeson) writes: < > Having just ordered a QIC 80386 computer, I want (need) a test suite to < > verify that the installed 80386 is one without the 16 bit limitations. < < According the sources I have read: < 3. Intel had the problems very early in the production cycle, < fixed the chip mask in Spring, 1987, and (supposedly) shipped < nothing but "good" chips since last April or May. Therefore, < the odds are excellent you own a good chip. Just yesterday we got in a new 386 system with a mother board make by AMI under license from Mylex (the one with the 64K cache buffer.) Tried to load Xenix and it puked with the multiply bug (which Xenix 386 is smart enough to check for.) I was writing down mask numbers from the chip so we could harass the supplier and the last line on the chip package said: 16 bit S/W only Consider this a warning. Looks like Intel decided to sell the old ones but just mark them. Also, attempting to load Xenix 386 seems to catch the problem. -- Phil Hughes, SSC, Inc. P.O. Box 55549, Seattle, WA 98155 (206)FOR-UNIX uw-beaver!tikal!ssc!fyl or uunet!pilchuck!ssc!fyl or attmail!ssc!fyl