From tw45639@vub.ac.be Thu 22 Jun 95 12:18:59
From: eric@telebit.com (Eric Smith)
Newsgroups: comp.sys.hp48
Subject: Re: HP48 vs. TI92
Date: 22 Jun 1995 00:01:47 GMT
Message-ID: <ERIC.95Jun21170147@marble.telebit.com>
References: <3rof35$s3a@zipper.zip.com.au> <3rqifk$fdq@crow.cybercom.com>
To: Matthew Carlson <carlson@rahul.net>

In article <3s2mlu$3gj@hustle.rahul.net> Matthew Carlson <carlson@rahul.net> writes:
> First thing: Putting a faster chip in the HP48 series would not be a 
> problem cost wise but it would drain the batteries to hell.

This claim is ridiculous.  The power consumption of CMOS digital circuitry
scales almost linearly with frequency, and in low power designs the quiescent
power comsumption is negligible.  The HP 48 stops its CPU clock when it is
idle.  Therefore, even if there was an 8 MHz Saturn chip in place of the 4 MHz,
you would get essentially the same amount of computation from a single set of
batteries.  You would just get your answers quicker.

On the other hand, it is unlikely that they will use an 8 MHz Saturn.  The
Saturn architecture was designed in the early 80s to meet a different set of
design goals than would be desired for an HP 48 successor.  A low power RISC
chip (or RISC/CISC hybrid) such as an ARM, Hitachi SH, NEC V8xx, AT&T Hobbit
(RIP), or even the new Motorola "Cold Fire" (which is now called something
else) would all be suitable for a new handheld calculator.  Remember that
clock speed isn't the whole story.  How much useful work you get per clock
cycles is very significant, as is number of signal transitions per clock
cycles.

RPL could be easily and efficiently implemented on any of the processors I
listed above, and would probably be more energy efficient than the Saturn.

For more details you can refer to a copy of my earlier posting:
        http://www.telebit.com/~eric/hpcalc/rant.html

And to support my claim that low clock frequency isn't essential to energy
efficiency, check out the paper "Energy Efficient CMOS Microprocessor
Design" (Burd and Brodersen '95, Proceedings of the 28th Annual HICSS
Conference, Jan. 1995; Vol. I, especially section 5.2.  This paper is
available on the web from
        http://infopad.eecs.berkeley.edu/CIC/gpp.html

Cheers,
Eric


