The Ware for August, 2005 is shown below. Click on the image for a much larger view.
This month’s contest should be a little more difficult than last month’s contest, although I don’t rule out the possibility that someone may have worked on this machine and could recognize this board outright. Remember, answers are judged partially upon the depth of the analysis of the board that leads you to your conclusion.
This board demonstrates a fatal flaw in its implementation that can be observed with casual inspection. Bonus points to anyone who can figure out what this is! Keep in mind that I have a proclivity for computer architecture that colors my outlook on circuit boards like this one.
Hmm. Looks like a quad-Intel i860XP RISC processor running at 50 Mhz setup (less a heat sink). Kinda unusual to see two boards screwed together, I’d say that is a good design idea if the processor architecture needed to be modified for the specific logic test. But a bad design fo structual integrity. This looks a lot like the discontinued Agilent E2412A for ‘reverse engineering’ or ‘inverse assembly’. It is also unsual seeing two wire bridges in the 600-region of the left ‘processor’ board. Well, that’s my two cents. PS: We all love you in Xbox-land! Truely Pioneers of your time! :)
I wish that submissions for the first while (week?) were kept private. That way I could go research supercomputers of the early 90s a few days from now and not have to worry that I’d get scooped.
I have a SWAG with an SHA1 sum of
70009e2532a96b57d112690664afdbce6b76eb85
Intel Paragon XP/S CPU board hooked up to the comms unit, with Emulex SCSI chip. Bug is probably memory latency, related to your thesis.
I’m hurt that I got dinged for “not enough info” last time!
For what it’s worth,
echo “Intel Paragon” | sha1sum
70009e2532a96b57d112690664afdbce6b76eb85 –
I really wanted it to be an SGI geometry engine board, but I couldn’t
find a matching SGI board out there, and I latched on to the Emulex as
Nate did. The Paragon was an massively-parallel machine based on the
then-super-hot i860 or “cray-on-a-chip” as the hype machine called it.
You too can own one of these boards (sans the I/O daughterboard):
http://www.ion4.com/product.php?product_id=62
Each node in this generation of Paragon had 2 i860’s devoted to
computation and a third CPU devoted to communication. In my little
academic part of the world, people seemed to think that the functional
division didn’t work out so well—I have vague memories of claims
that on two-processor boards you were better off using both for
computation and communication. Or maybe the claimant wished that they
could use both for computation and couldn’t.
Here are all the clues I saw:
“SSD” stamped on lower left of board indicates Intel’s Supercomputing Systems Division (now disbanded). This would have given it away even if the CPU labels had been blurred out. This sticker plus the one in the upper right, plus all the EPROMS have the standard 6-digit serial number scheme still in use today: http://developer.intel.com/design/pentiumXE/datashts/306831.htm
There are 3 CPUs and an LSI “northbridge”. The LSI copyright of 1993 indicates the board was one of the later ones, hence XP/S or XP/E. The 50 Mhz designation indicates this is an XP/S (although I could be wrong here).
Other highlights:
Xilinx FPGA “southbridge” — guess volumes of the comms boards were too low to do a custom ASIC for each model
Multiple crystals — haven’t they heard of clock dividers? 80 Mhz crystal top left, 10 Mhz bottom right (label not visible but guessing from its closeness to the ethernet chip)
10Mbps Ethernet controller lower right, Intel N82586-10
MAC address sticker on comms board
Emulex makes SCSI controllers although the 2400121 didn’t show up in a parts search. I think it may be dual-channel, judging from the headers for connectors on right side of board
May I suggest in the future upping the ante by blurring out all major part numbers (CPUs, north/southbridge, RAM)? This is similar to a lot of production boards where the numbers are milled off all custom parts. Perhaps anyone who won the regular contest could only enter the difficult one afterwards.
Megan
Did you consider any other viewpoints before you wrote this??
Megan
I wish more people had the guts to say that
subj
?
Good Night webmaster!
Only for professional webmastersOne hundred percentpartnerku, all without cheating money list on the second day after reaching the limit, the result is 100%, there are bonuses to 120,000 dollars, register for free-http://germanfriendfinder.com/go/g1111998-brk
P.S love planet-there is everything! Germanfriendfinder: nothing lost …
See you!