home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!noc.near.net!mars.caps.maine.edu!maine.maine.edu!ree700a
- Organization: University of Maine System
- Date: Monday, 27 Jul 1992 14:01:36 EDT
- From: <REE700A@MAINE.MAINE.EDU>
- Message-ID: <92209.140137REE700A@MAINE.MAINE.EDU>
- Newsgroups: comp.sys.intel
- Subject: Re: 386/486 clock (1x or 2x)?
- References: <RFowler.288@kelvin.jpl.nasa.gov> <3713@svin02.info.win.tue.nl>
- <TMH.92Jul22194404@keks.first.gmd.de> <scott.711851090@labtam>
- Lines: 37
-
- Scott Colwell states (correctly) that quartz crystal oscillators (TTL-crystal o
- scillators) run out of steam (read: price-performance benefits) around 80 MHz.
- On the other hand, clock division is very cheap, even if a little ECL is neede
- d, and SAW-based quartz oscillators start around 300 MHz (i.e. garage door open
- er xmitter oscillators. So, the availability of oscillators doesn't limit the
- CPU speed. Furthermore, having a 100MHz oscillator and 2x divider in a sheilde
- d box will not cause RFI any more than the inherent harmonics of a square wave
- at most of the popular CPU speeds.
- What really limits CPU speed is heat dissapation and the 5V standard. Curre
- nt research in the microwave generation of diamond thin films from methane will
- be a big factor in future high speed cpu's. Since crystalline diamond is the
- best thermal conductor/ electrical insulator (OK ultra-wide bandgap semiconduct
- or), such a film would be a big help. also, the drive to low voltage logic (se
- e the IEEE spectrum article "High on Low Voltage" recently) will allow faster c
- lock speeds at the same power dissapation and also allow smaller transistors.
-
- So, my question is: Just because your board manufacturer uses a 2X clock s
- cheme and divider to provide the 486 with its 1X clock, why do you insist on th
- inking that that clock (oscillator) is directly fed to your chip? The 486 is a
- 1X chip, which is neither an innovation nor a big advance. It probably has li
- ttle or no impact on RF interference and only a slight psychological ecffect on
- the dope at the FCC that licences it (Hey, this is a computer, not an FM trans
- mitter!) since many mainframes and workstations have CPU clocks of 100 - 200 MH
- z. (covers most of the TV VHF, UHF and cable TV bands, better stop that trend,
- too).
- The only technical reason for the 486 being 1X that I can arrive at is that
- a fiberglass, multilevel circuit board is one hell of a capacitive load for a
- 100 MHz TTL signal to drive. The vendors knew it and Intel knew it. So, Intel
- made the 486 a 1X clocked chip so they wouldn't need to use low-dielectric cir
- cuit board like teflon at 5X the price. My only shock at this is that they did
- n't rate the previous 2X chips at the input clock rate (Hey, its a 20 MHz 8088!
- Try AMD's 80 MHz 386! got a 32 MHz 386SX.. Real cheap!) rather than the eff
- ective clock rate...
-
- So, can we end this useless bickering over whether Intel's data book is corr
- ect (wrt the input clock speed at least) & get on to the important things in li
- fe)
-