home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.ibm.pc.hardware
- Path: sparky!uunet!mcsun!sunic!fuw.edu.pl!cocos!jt
- From: jt@fuw.edu.pl (Jerzy Tarasiuk)
- Subject: Re: What's the deal? My chip says "SX-25"; Norton says "SX-33"
- In-Reply-To: khor@bnr.ca's message of Tue, 5 Jan 1993 19:04:34 GMT
- Message-ID: <JT.93Jan12222733@fizyk1.fuw.edu.pl>
- Sender: news@fuw.edu.pl
- Nntp-Posting-Host: fizyk1
- Organization: Warsaw University Physics Dept.
- References: <1ht90eINNei0@hpscit.sc.hp.com> <C0Cos2.2Fw@usenet.ucs.indiana.edu>
- <1993Jan5.152504.9786@bmers95.bnr.ca>
- <1icegqINN7vl@savoy.cc.williams.edu>
- <1993Jan5.190434.12413@bmers95.bnr.ca>
- Date: Tue, 12 Jan 1993 21:27:33 GMT
- Lines: 40
-
- >>>>> On Tue, 5 Jan 1993 19:04:34 GMT, khor@bnr.ca (Kyle Hor) said:
- Kyle> In article <1icegqINN7vl@savoy.cc.williams.edu>, 93gke@williams.edu (oldwanda) writes:
- Kyle> |> If this is true, then how can vendors push the 25Mhz chips to 33Mhz?
- Kyle> |> Wouldn't they fail at the higher speeds since they were slow in
- Kyle> |> testing?
- Kyle> |>
- Kyle>
- Kyle> 1st there is a guard band. I.e. a 25MHz grade are usually tested to pass
- Kyle>
- Kyle> 2nd, there are numerous test on different parts of the chips switching
- Kyle> speed. Again, not knowing the 486 intimately, let's assume there are
- Kyle> 100 such tests. Maybe 99 of them meet the 33MHz grade but one of them
- Kyle>
- Kyle> 3rd, semiconductors are sensitive to temperature. The chip is
- Kyle>
- Kyle> Finally, if Intel has a huge order of 25MHz and have a huge stock of
- Kyle> 33MHz, they would simply re-mark the chips as 25MHz and ship it. This
-
- In this case why Intel doesn't sell 33MHz CPUs at 25MHz's price ?
- (for the vendor who gets larger amount of them only, of course)
-
- But I can say also 2 reasons for which 25MHz chip can work at 33:
- 1st, a reason to say 33MHz is too much for it may be heating - higher
- clock rate means more heat and it may destroy chip at full speed
- and maximum ambient temperature simultaneously; but a vendor can
- add heatsink to it or give good cooling to prevent destroying.
- 2nd, and seems no one said it: test is done at some output load
- (a capacitance connected to every pin), and a chip can be to slow
- with this load (for example, the output current from it is not
- sufficient to load the capacitance in required time); but a vendor
- can put chip on board where capacitances are less than those used
- by Intel's test (the capacitances also affect chip heating).
- Anyway, to make sure computer made this way is reliable must measure
- a lot of parameters or consult Intel to know e.g. reducing load
- capacitances by 40% or reducing ambient temperature by 35 degrees
- allows to use 25MHz chip at 33MHz. But the only reason for vendors
- to do it is honesty. Or, more precisely, might be honesty if there
- were honest businessmans (but I don't know about any).
- Jerzy Tarasiuk
- "Cannot serve to two masters - God and money - simultaneously."
-