Wow.
Just wow. The folks working on this are scary-good at Maths.
Not sure if it's a good thing, or a bad thing, but well done anyway.
This round's on me :-)
Japanese computing giant Fujitsu is claiming a world record after successfully breaking a 278-digit (978-bit) pairing-based cryptography system, providing useful data on how far this next-generation encryption system can be trusted. The company’s R&D arm, Fujitsu Laboratories, worked with Japan’s National Institute of …
When an ellipsis is used at the end of a sentence, the period (or full stop) results in what appears to be a four-dot ellipsis. Sorry 'bout the typing lessons from Herb Caen; he wrote his columns for the San Francisco Comical on a manual Royal typewriter and had an assistant to proofread and correct his copy--probably including the additional dot in a sentence-ending ellipsis.
Under my desk at work sit two boxen. One is my workstation, which 'only' has four cores. The other is my server, an HP ProLiant series machine, which has two hexacore Xeons. Fine, that's only 12 physical cores, but each CPU is capable of hyperthreading, which means that the OS sees (and can use) 24 logical cores. That's just a dual-socket motherboard, so let's extend it forward to something that Fujitsu Labs might even be able to afford.
Make that motherboard quad or even octa-socket, and you're looking at potentially 96 physical cores, 192 logical ones. Put in a dodecacore CPU instead of a hexacore one, and then you have anything up to 192 physical cores, 384 logical ones. And that's in just one machine! Isn't technology amazing?
And also, what sort of tech illiterates do we have reading El Reg these days?
Some might call it a workstation. But there's no exact definition of what is and isn't a PC (Personal computer). I'd say that if it's a system that can sit on or under a desk without making too much noise or heat for an office environment, then it's a PC. Maybe also require that it contains an Intel-x86-compatible CPU and/or can run MS windows if you want it to, if you want to rule out a Sparcstation or a Mac.
Out of interest, what do you call those souped-up gamer systems with overclocked water-cooled CPUs and humumgous GPUs? (Apart from insane, of course).
That doesn't rule out Macs at all. Macs use standard Intel parts and can certainly run Windows on them. My work issued MacBook ran windows 7 on it naively, the hard disk died in transit so I just slapped a better disk in it and installed Windows.
Oh and there was a copy of Windows NT 4 that would run on a sparcStation, also NT 4 could also run on a Power-PC based machine, so it was possible to run on pre-intel macs after some hacking.
As El Reg just reported:
"A 10U chassis holds eight half-width nodes, with up to 128 cores and 4TB of memory. A single rack has four of these, for up to 512 cores and 16TB of memory; and a fully loaded UV 2000 has eight racks for a total 4,096 cores and 64TB of global shared memory."
http://www.theregister.co.uk/2012/06/19/sgi_uv_2000_xeon_super/
Now picture a private or government IT lab filled with racks of racks...
Note that they cracked a 978-bit code. That suggests they matched the target to the available hardware. For any sensible cryptographic algorithm, the amount of CPU needed to crack it rises *exponentially* with the number of bits. 978 might be indicating that 1024 isn't enough bits to be safe against a government agency (don't know enough about the maths of this algorithm).
If in doubt add some more bits. The only trouble is that the time to encrypt and decrypt rises when you do that (but far less so than the time to crack! )
Yup, as far as I was aware, the key space is 70 (US) trillion times bigger when you move from 978 to 1024-bit. And it took nearly 5 months to break* a single instance. Having said that, most HTTPS sites use 256-bit keys...
* not really considered a break or a crack given its brute-force.
This is a function-field-sieve discrete logarithm over GF(3^582); it's asymptotically equivalent difficulty to special number field sieve factorisations, which casual groups have managed to do for 1061-bit (320 digit) numbers. As the Fujitsu paper http://www.nict.go.jp/en/press/2012/06/PDF-att/20120618en.pdf points out, it involved a fair amount of implementation work but no more computing than finding a single DES key.