Slashdot Mirror

In between 5.25" and 3.5" floppies - I had both in my machine when I decided to migrate my data to 3.5". No problem - after all my 5.25" floppies were copied over, I retired the lot. Programs were in BASIC to begin with - which was directly translatable (as long as you weren't poking and peeking memory locations - in which case you also had to do a bit of integration/modification work to make it edible by the new hardware). As the years moved on, and DOS and Windows, and then other binary formats took hold - the strategy turned to emulators or virtual machines as the solution. I have VMs running Windows and DOS that I use to run binaries for those old applications and games.

As for data, early on I decided that I wanted to keep critical files in a common format that I knew would be readable in the future. I started off with ASCII - and had no problems moving files from one machine to the next in that format, then I moved to .RTF format. Today I've taken that a step further and new files are kept in XML format. In any case, I have files from my early years still readable today.

Vint Cerf has been championing the idea of 'Digital Vellum' - technologies to preserve our software and data for the future so data doesn't loose meaning because there is no machine to run its program on - what he refers to as the Digital Dark Age. Here is Mr. Cerf talking about it on youtube, and here is a document and abstract on the concept.

TFA doesn't seem to mention it by name, but it sounds like this is an attempt to build a computer based on rapid single flux quantum principles.

Basically, you replace transistors with things called "Josephson junctions", and use short (picosecond-range) bursts of electricity instead of continuous DC current. Josephson junctions are a quantum phenomenon that happens in superconductors, hence the Q in RSFQ, but the computation itself is traditional logic, not quantum weirdness. That's why it needs to be cryogenic - we don't have room-temperature superconductors.

The main benefit is that Josephson junctions are extremely fast - a 2005 study by the NSA (apparently not classified) planned an initial test computer at 50GHz, boosting to 250GHz by 2010. This seems to be referenced obliquely in TFA, which mentions 100GHz clockspeeds. They're also very power-efficient - the bursts of electricity can be absolutely tiny, since it's working in a superconductor.

The 2005 paper seemed to go nowhere - there are a lot of issues to sort out, from "how do we make RAM for this thing?" to "how do we make power cables that run at these temperatures?". But I've suspected that they secretly continued development, and had functional RSFQ computers, since it's actually a pretty neat idea and the price tag in the paper was pretty low ($46M).

Oops, got in a hurry there. My FAIL. While parts of that bill were passed in other legislation, that bill ultimately failed. What I was thinking of was actually:

High Performance Computing Act of 1991
http://www.nitrd.gov/congressional/laws/pl_102-194.html [nitrd.gov]

I don't need a citation for common knowledge, but since knowledge evidently is not too common:

Exhibit A: DARPANET

http://en.wikipedia.org/wiki/ARPANET

Exhibit B: TCP/IP

http://en.wikipedia.org/wiki/History_of_the_Internet#TCP.2FIP

Exhibit C: High Performance Computing Act of 1991
http://www.nitrd.gov/congressional/laws/pl_102-194.html

Almost every major problem in computer science is the result of our infatuation with the Turing machine. The problem with the Turing computing model is that time is not an inherent part of the model. Timing is the key to solving the cyber security and reliability crises. Turing is the problem, not the solution.

Check out this short discussion at the new Federal Cybersecurity R&D Forum.

I did some Google searches, hoping to find some historical info on NSA's cryogenic computing efforts, and found this, a 2005 plan out of NSA to build a 50-100GHz computer by 2010.

They want faster CPUs, not more CPUs. The commercial world isn't even trying any more. After reading this paper, one can see why. By throwing a few hundred million, and liquid helium, at the problem, they might get a 20x performance gain over commercial microprocessors. The CPU has to run at 4 degrees Kelvin, liquid helium temperature. And it has to be kept at 4K while dissipating about a kilowatt.

The technology is totally nonstandard. The basic components are Rapid Single Flux Quantum devices running at 4K. The logic voltage power voltage is 3-5 mV. Signals are around 200 microvolts. This stuff requires custom semiconductor fabs to make.

Getting data out of the low-temperature zone is a very tough problem, and optical interconnects have to be used. The proposed memory bandwidth is huge: "For example, a particular architecture may require half a million data streams at 50 Gbps each between the superconducting processors and room-temperature SRAM." Developing devices to drive the output data links from the low temperature zone, without causing too much heating in the cold part of the system, is a big part of the problem.

The justification for all this is in Appendix E, and sounds totally bogus. Either there's some desperate need for this technology they don't mention, or it's a boondoggle. There must be something important for which parallelism won't work. It's surprising to see this from NSA, because most signal analysis and crypto problems parallelize well.

Please read here

Basic CS research ought to be funded, IMO, but there's no reason completely open-ended CS research should be funded by DARPA---that's what the National Science Foundation is for.

Let's be frank, there are certain things in basic long term CS research that DARPA is going to be a lot more interested in than the NSF. It makes sense for DARPA, then, to bother to make sure that research is getting done. The best way to make sure that research is getting done is to pay for it.

What sort of research should DARPA be interested in? Anything related to software security and assurance is going to be of more interest to DARPA than the general public (yes the general public is interested, but they aren't quite as motivated as DARPA). There's plenty you can do in that field, from new security architectures in the OS (like, for instance, what the NSA did with SELinux etc.), through to new protocols, better fault tolerance, intrusion detection etc. Having your military computer networks secure is just good practice. You should be interested in being at the cutting edge of of that. If you want a nice list of things DARPA could be doing, along with a reccomendation that more money ought to be invested in long term research at DARPA, you could try this report to the President from a month ago by the President's Information Technology Advisory Committee.

Jedidiah.

Jedidiah.

Well actually it is particularly amusing in the light of this article posted here not 2 weeks ago in which the President's advisory committee provided a report that specifically cited declining DARPA focus on long term research as a fundamental problem for cybersecurity. The particular details can be found in the report on page 19 (with other references spread throughout).

Collect together an advisory committee that reports directly to the President. Have them do a study of the state of the nation's IT security, both from a civilian and from a military standpoint). A month after the report is released, do the exact opposite of one of its primary reccomendations.

I know the government is big (perhaps that's part of the problem?) and that the left hand rarely knows what the right hand is doing, but this is quite ridiculous.

Jedidiah.

I located two other government sources here and here.

Another poster also found it here.

I'd like to point out that while there is no direct mention of Trusted Computing, it calls for a "fundamentally different architecture", some sections mostly later in the paper apprear to describe Trusted Computing functionality, the experts they cite all appear to be Trusted Computing speciallists and proponents (in particular David Spafford was the author of the semi famous WHY_TCPA and TCPA_REBUTTAL papers), at least some of the committee members appear to have Trusted Computing ties, and an earlier Cyber Security Advisor gave a speech at the Washington D.C. Tech summit calling for Trusted Computing and for ISPs to eventually make it a mandatory part of terms of service for internet access. A call to fight worms and viruses and to Secure the National Information Infrastucture against terrorist attacks, to defend against Osama bin Laden himself. Yes, he actually cited bin Laden by name. chuckle.

-