home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.sys.ibm.pc.hardware:23854 comp.arch:9319
- Newsgroups: comp.sys.ibm.pc.hardware,comp.arch
- Path: sparky!uunet!iWarp.intel.com|ichips!ichips!glew
- From: glew@pdx007.intel.com (Andy Glew)
- Subject: Re: Does a 487sx shut down the 486sx??
- In-Reply-To: tdbear@tandon.com's message of 4 Sep 92 17:21:31 GMT
- Message-ID: <GLEW.92Sep9195211@pdx007.intel.com>
- Sender: news@ichips.intel.com (News Account)
- Organization: Intel Corp., Hillsboro, Oregon
- References: <1992Aug26.173519.22421@unislc.uucp> <1992Aug27.183804.8605@tandon.com>
- <GLEW.92Sep1091842@pdx007.intel.com> <1992Sep4.172131.27778@tandon.com>
- Date: Thu, 10 Sep 1992 03:52:11 GMT
- Lines: 52
-
- In article <GLEW.92Sep1091842@pdx007.intel.com> glew@pdx007.intel.com (Andy Glew) writes:
- >Test vectors are cheap. But tester time isn't. Testers are often the
- >manufacturing bottleneck.
-
- Andy, since you are in a good position to find out... could
- you post how long it takes to test the 486DX and how long it
- takes for the 486SX? Then could you provide an approximate
- number of chips which go through one particular tester each
- day (on the average). I know I am curious about this, and I
- am sure others are, too.
-
- You're testing me, right?
-
- Seriously, I very much doubt that I could ever post such information.
- The microarchitecture of the chip that I am working on is moderately
- interesting and relatively secret --- but details about process, time
- per stage, and yields are EXTREMELY secret. Process details would
- allow a competitor to estimate cost of manufacture, and hence
- anticipate our pricing strategies. I think that you will observe that
- all papers on such manufacturing details have graphs expressed in
- relative numbers, not absolute values.
-
-
- I truely cannot believe that the 486sx PGA (which probably
- still has the numeric) would not be tested thoroughly to
- ensure that all of the silicon is working (even though it
- won't be used) since there is a slight chance a damaged
- numerics area might cause contamination to the rest of the
- chip if (say) something overheats.
-
- I am not 100% certain, but I am reasonably sure that the original
- i486sx was *not* a CPU that passed integer tests but failed FP tests.
-
- The present i486sx, of course, doesn't need to pass FP tests because
- it has no FP on chip.
-
- There would be far too much risk involved in chips that failed any
- part of their functionality tests. Imagine that FP tests failed
- because of an intermittent short in the FP side - a short that would
- eventually close, even with integer only use...
-
-
- --
-
- Andy Glew, glew@ichips.intel.com
- Intel Corp., M/S JF1-19, 5200 NE Elam Young Pkwy,
- Hillsboro, Oregon 97124-6497
-
- This is a private posting; it does not indicate opinions or positions
- of Intel Corp.
-
- Intel Inside (tm)
-