View allAll Photos Tagged eee;
EEE Magazine
August 1970
Volume 18, Number 8
p37
Nick DeWolf Of Teradyne Speaks Out: Speed Kills (cont'd)
____
cases that's still plenty fast.
Slow consumes less power. That dollar's worth of power supplies I mentioned could come down to about a cent. Not to mention all those fans and filters (both air and noise-suppression) you don't have to use.
I really have no axe to grind on TTL, ECL, and other high-speed families. Where you honestly need all that speed, you need it, and that's that. But a good designer ought to know what he's paying to gain those nanoseconds. I winder if most TTL users are aware of the egregious current transients TTL inflicts on the power supply, for instance. When a TTL gate bombs from one state to the other, it draws from six to eight times the current it draws at either one of those states. Moreover, when the TTL gates pulls up, it derives the current to pull up the output capacitances from the harassed power supply, whose glitches are transmitted to all the "up" outputs. Of course, the power supply has to be bigger in the first place to support the extra current. As if all this were not bad enough, TTL also does you out of any chance of using the "wired-OR" connection, which greatly simplifies so much circuit design.
The push for speed has been a dominant influence in the whole IC business, and it hasn't been a healthy influence. IC vendors who don't have to worry about the marketability of their transistors, gold-dope them to death and get away with it. Having proved that point, they should take a fresh look at the situation and consider other approaches. We've found, for instance, that we can achieve excellent noise rejection through the use of hole storage to utterly stall the gates - a pretty appalling idea to a confirmed "dope fiend."
If deliberately going slow sounds crazy, consider this crazy thought: If bipolars were designed with pinch resistors to go as slow as MOS devices, they could handily complete with MOS for packaging density, "real-estate" advantages, etc., revolutionizing the whole IC technology. The hang-up has been that the bipolar people, absurdly overplaying the importance of resistor tolerance, have been trying to make precision resistors.
How slow is slow?
This whole device-speed argument really involves the order of magnitude on either side of 50 nanoseconds. I feel comfortable with 500 nanoseconds, but 5 nanoseconds is a killer. So I leave 5 nanoseconds to those who are bent on suicide. In an IC test system that takes 10,000 nanoseconds per test, for example, I sure as blazes ought to be able to get along with 50-nanosecond propagation delays (and I'd prefer 200 nanoseconds, for noise rejection).
That leads to my next point, which is that
37
(Photo of Nick by Kemon Tashioglou)
part of an archival project, featuring the work of nick dewolf
© the Nick DeWolf Foundation
Requests for use are welcome via flickrmail or nickdewolfphotoarchive [at] gmail [dot] com
'
Serias tan amable de convidarme uno de tus Lucky Stike ?..
{ si te cagas los pulmones,qe sea con una weá buena.. :$ }
yo.
Thanks UPS!
My Asus Eee PC 900 (didn't want to wait for the 901/Atom proc one anymore) and Metal Gear Solid 4. I guess I have my weekend cut out for me! Sadly, my 2gb RAM module for the Eee didn't come in at the same time so I guess that'll have to wait until Monday.
For all the Eee fans, I got the 20gb Linux version and am ready with my nLite'd install disc with WinXP SP3. I'm a big Linux fan, but I gotta support my brand so Windows, here I come.
edit: why is the MGS4 game in this photo? 2 reasons: I got them on the same day from UPS and thus, it's an easy way to track when exactly I got the Eee - the same week MGS4 came out! That, and it's a good size comparison, too.