Thread View: alt.folklore.computers
4 messages
4 total messages
Started by sha... Shannon Nelson
Tue, 10 Apr 1990 16:14
More blinkenlights
Author: sha... Shannon Nelson
Date: Tue, 10 Apr 1990 16:14
Date: Tue, 10 Apr 1990 16:14
29 lines
2218 bytes
2218 bytes
It's really amazing how handy lights can be when debugging some programs. Especially when dealing with many CPUs at once. Our parallel computer, the iPSC/2, has LEDs on every node to show communication (red), numeric activity (yellow), and general activity (green). The green is programmable by the user. Particular algorithms are very recognizable by the light patterns they produce. The first computers we built (iPSC) only had lights on the nodes inside the cabinet, not on the outside where they could be seen. Customers were always leaving the front door open so that they could see what was happening. This was a problem because the door was designed to keep in all the nasty EMI that the machine generated. We couldn't open a glass window in the door, as that would also defeat the EMI sealing. There was no support on the nodes themselves to bring out the individual LED signals, so something else had to be devised. What we finally did was put light sensors in the door so that when the door was closed, there was a sensor over each of the 96 leds in the chassis. These sensors controlled a light panel on the front of the door. Expensive, but as long as the door was closed, this system worked fine. Some customers still liked leaving the door open anyway, which had a tendency to flood the light sensors with too much light. Our latest models now have LED signals routed to the backplane and up to the door from there. They can also be read by software. -- Shannon Nelson "Live, from Intel Scientific Computers..." sha...@isc.intel.com ...!uunet!littlei!intelisc!shannon (503) 629-7607 Intel disclaims all knowledge of my existence.
Re: More blinkenlights
Author: g... gordon k chace x6524
Date: Wed, 11 Apr 1990 17:47
Date: Wed, 11 Apr 1990 17:47
20 lines
1609 bytes
1609 bytes
In my previous life (University of Illinois, 1970's) we had an old IBM 360/75 which had a *beautiful* light panel. We unscrewed the light bulbs (no led's in those bygone days) and plugged in some wires that went into our "heart/lung" machine. This thing had a patchpanel for an array of NAND, NOR, and NOT gates, plus some frequency-divide-by-32768 stuff, finally incrementing mechanical counters (just like the odometer in an automobile). We used this contraption to analyze performance, things like "how often in system more versus user mode" or "how often is the program address up in slow memory instead of fast memory". So they were clickenlights as well as blinkenlights. That computer was too heavy for standard false-floor struts, so we braced critical flooring panels with concrete blocks which were filled. I was given one of these (about 90 pounds) when the machine retired. Gordon Chace Control Data Corp. (612) 482-6524 (fax 482-2791) g...@freddie.udev.cdc.com
Re: More blinkenlights
Author: x04... Butch Kemper
Date: Thu, 12 Apr 1990 14:04
Date: Thu, 12 Apr 1990 14:04
12 lines
1145 bytes
1145 bytes
In my earlier days, I worked for the KFA Julich in W Germany and we had a 360/75 system. The light panel was huge with rows of lights and a big "test" button. I was told that on early model 75s, the "test" button could only be depressed for 5-10 seconds before the power supply tripped which would cause the whole system to power down. Since the peripherals were all connected to the cpu by EPO cables, everything in the room would go down when the panel light power supply tripped. Alas, I never got to test the "test" button because it had been removed and the hole taped over. Butch Kemper
Re: More blinkenlights
Author: holl... The Polymath
Date: Thu, 12 Apr 1990 20:22
Date: Thu, 12 Apr 1990 20:22
19 lines
1498 bytes
1498 bytes
In article <669@intelisc.nosun.UUCP> sha...@isc.intel.com (Shannon Nelson) writes:
}The first computers we built (iPSC) only had lights on the nodes inside
}the cabinet ... Customers
}were always leaving the front door open ...
}a problem because the door was designed to keep in
}all the nasty EMI ... [We]
}put light sensors in the door so that when the door was closed, there
}was a sensor over each of the 96 leds in the chassis. These sensors
}controlled a light panel on the front of the door. ...
Did no one think of using fiber optics? A few large, cheap "designer
table lamp" type fibers per light should have done it. No muss, no fuss
and _much_ cheaper than electronic optical sensors.
--
The Polymath (aka: Jerry Hollombe, M.A., CDP, aka: holl...@ttidca.tti.com)
Citicorp(+)TTI Illegitimis non
3100 Ocean Park Blvd. (213) 450-9111, x2483 Carborundum
Santa Monica, CA 90405 {csun | philabs | psivax}!ttidca!hollombe
Thread Navigation
This is a paginated view of messages in the thread with full content displayed inline.
Messages are displayed in chronological order, with the original post highlighted in green.
Use pagination controls to navigate through all messages in large threads.
Back to All Threads