The Cell Matrix
Monday, October 31, 2005
Monday, September 12, 2005
Monday, September 05, 2005
Hardware Page Camera OfflineIt appears that the online MOD 88 camera is offline, as of 2 Sept. The interface to the MOD 88 board itself still works, and you can send commands and receive results. But the camera is frozen. This will not be fixed until later on Sept 8, when I can reboot the box running the camera (an old Windows machine that sometimes freezes up. Unfortunately, camera software seems to work better under Windows than Linux, but the long-term reliability is not there).
Apologies for the inconvenience.
Tuesday, August 09, 2005
Hardware Fault InjectionSome people have expressed interest in being able to simulate faults on the MOD 88 hardware system. The MOD 88 (Version 6.1) now has that capability. The upgrade from 5.1 is a single socketed chip, available on request if you already have V5.1 There is also a software upgrade in the API and the Command Line Interface.
A single API call, set_fault(row,col,output,value) is used to SET (value=1) or CLEAR (value=0) a STUCK fault in any cell [row,col] and any output (dn,ds,dw,de,cn,cs,cw or ce). When SET, a STUCK fault will freeze the indicated output, causing it to remain at its present output value, regardless of subsequent cell activity. When a stuck fault is set to CLEAR, the fault is removed, and the output operates normally.
Additionally, when a cell contains any such faults, its LED will turn orange.
From the Command Line Interface, a new command "E" is available. It takes several forms:
E row col output value
behaves identically to the set_fault() API call.
E row col
clears all faults in cell [row,col]
clears all faults in the MOD 88.
Note that faults are NOT cleared by the reset() call or the R (reset) command. This more accurately models the behavior of persistant faults which would remain following a system reset. Note also that faults can only be set in the board directly connected to the PC. In the online 2x2 setup, only the cells in the upper-left board can have faults applied.
The online MOD 88 has also been upgraded to V6.1, and the E commands are available from the user interface.
Below is a picture of the online system, configured as a binary counter.
After a fault has been injected into cell [5,5], the corresponding LED turns orange, and the higher counter bits freeze.
Re-bootstrapping the Cell Matrix with a faulty cell can lead to unexpected results, as shown below. This is because cell [5,5] is used in the configuration of other cells, but since it is faulty, those configuration operations have failed.
Send us a comment...
Wednesday, August 03, 2005
Back Online!It may appear that this BLOG has died, but we're still here! Been swamped since the DC EH 2005 conference, and spending a lot of time on non-postable work. But I wanted to give a brief update on recent events.
The NASA/DoD EH 2005 conference (Evolvable Hardware workshop) was big fun. Met a lot of people, saw a lot of familiar faces, and generally had a good time demoing the MOD 88 hardware setup. We had a 2x4 setup (16x32 cells) connected to an old laptop running Linux. Had a set of self-contained demos available, for people to run themselves, or that could be left running for others to look at when they walk by.
Jason Lohn was a huge help (actually everyone involved in setting up and running the conference was a huge help!), he had a big table for us right at the entrance to the poster area, so we had the standard easle, plus a table, plus our own easle for displaying the hardware. The whole thing took around an hour to setup the first time, but we were down to 20 minutes by the end of the conference. Here's a picture of the whole setup:
That's Murali Raju, one of the original designers of the Cell Matrix. He spent much of the day helping set things up, take things down, move things when the poster location changed, and interacting with curious on-lookers.
Here's a closeup of the boards themselves:
Some people had never heard of the Cell Matrix, and this provided a graphical, tangible way to introduce them to the architecture. Others were interested in the educational potential, which is one of the main reasons we built it. Others were interested in possibly using it for Evolvable Hardware research. Arturo Hernandez-Aguirre from CENTRO DE INVESTIGACION EN MATEMATICAS(CIMAT) bought one of the boards I'd brought to demo, and will hopefully have students working on it this fall.
Pauline Haddow from Institutt for datateknikk og informasjonsvitenskap expressed interest in the possibility of introducing hardware faults into the MOD 88 in a controlled way. More on this later...
Tuesday, April 19, 2005
Attending EH 2005 in Washingon, DCCell Matrix Corporation will be at the upcoming NASA/DoD Evolvable Hardware Conference (EH 2005) in Washingon, DC June 29-July 1. See http://ic.arc.nasa.gov/projects/eh2005/index.htm for conference details.
We will have a spot in the poster session, where we will have MOD 88 hardware available for demonstration, etc. We will also have a short paper in the proceedings, describing the MOD 88, its relevance for education and research, evolvable hardware uses, and so on.
This is always an interesting and fun conference!
Wednesday, April 06, 2005
MOD 88 Back OnlineOur DSL problem has been resolved, and we have a new light. Everything is back online now.
MOD 88 Server DownOur Verizon DSL line has gone dead. Not sure what the problem is yet, but we'll get this back online as soon as possible...
New MOD 88 Pages
We have created a new set of demo pages for the MOD 88 online hardware. The URL is still http://cellmatrix.go.dyndns.org. The pages have a cleaner look, more links to related pages, and include a new walkthrough that demonstrates some basic features of the MOD 88.
Note that our light has burned out! Therefore, the “Flip the Light Switch” button has no effect. We'll get a new bulb on Wednesday...