When you have companies such as IBM and Intel looking to destroy your business, it's nice to have a fella like Ivan Sutherland stored away in a back room. Sutherland, considered the father of computer graphics, developed a method of linking processors together called face-to-face computing, as part of his work at Sun …
It's child's play
They should seek patent-swaps with the LEGO(tm) folk, and just put the chips together like so many LEGO(tm) bricks. Perhaps some nano-nubs will help the alignment.
Again, some of this is being done at MIT. In fact, there used to be an interactive exhibit option that used programmable bricks. It was cool.
Del has the solution right there
Lego manages to achieve pretty good alignment with little effort because of the interlocking pegs. Sun should adopt the same idea and stop making their chips flat, the pegs will help guide alignment.
Such pegs could even work as connecting pins and form the electrical contacts, with a better surface area to volume of metal than a flat surface contact it will reduce resistance/capacitance problems and so the heat requirements as highlighted in this article.
wires are too slow, consider QPCT
Perhaps they should consider quantum proximity communications technology.
Quantum entanglement could be done in silicon with quantum chipsets 'tuned' to interface with each other.
That way, you could put chips where ever you want and concentrate on the lowest production cost for everything else.
http://www. csmonitor. com/2001/1004/p15s1-stss.html
Maybe Sun should give up on chips
Sun continuously posts "world record benchmarks" which show their processors(chips) vs. other vendors processors(cores).
"The Sun Fire E6900 (24 processors, 48 cores, 48 threads) with UltraSPARC IV+ 1.95GHz set a new world-record for the SAP-SD 2-Tier Standard Application benchmark for systems with 24 or fewer processors as of 04/02/07, achieving 6160 users."
Check out www.sap.com/benchmark
First of all they had 48 cores (some people would call that a processor)
Second of all they had 3 times the number of cores than the IBM p570 (16 cores) and only had 12% more performance.
And they are proud of this "world record"?
Btw...the p595 64core system had 23,456users.... the 32 core system would be at least 11,728 users
again...why is Sun proud of 6,160 user with 48 cores???
Quite old news !!!
Sun posted this info 3 or 4 years ago !
it's a form of democracy....
intel has itanic,
IBM has zseries,
Microsoft has windows,
Sun has near proximity....
near proximity is a 25 $ word for "close, but not there"
which is unfortunately the case to often with SUN's inventions and ideas; good, but very often 2 years late......
Anyway, let's not give up hope; who does not fear the day we have to live with IBM-sales guys only.....
new News isn't any better.....
quite recently (last month or so ? ) HP managed to "beat" IBM's tpc/c record (4 mln tpm/c, 64-way power box ) with
with a margin of 0.3 % or so..... ( itanic superdome )
unfortunately, in the process the # of cpu's doubled, ( 64==>128 ) and so did the software bill...... ( 480K ==> 1000K $ )
re: it's a form of democracy....
Quote [Sun has near proximity....
near proximity is a 25 $ word for "close, but not there"]
maybe that should have read
near proximity is a $0.019999 word for "close, but not there"
not quite my 2c worth
re: Maybe Sun should give up on chips
Recent Sun systems have been designed with a large number of cores/threads. You are right that each thread has less performance than a chip optimized for single thread performance. However, this design makes sense for servers, since the system as a whole has very high throughput. Also, the system cost and power consumption of a single chip 32 thread system is far less than competitors' systems constructed from multiple chips.
For servers, "performance per core/thread" is often not an important metric. Customers are more interested in price/performance and performance/Watt. Sun is doing very well on both fronts.
Quantum Entanglement for Beginners
Imagine I have two playing cards, one red and one black. (These represent the two entangled particles with opposite spins.) You take one of the cards without looking at it, and travel halfway across the universe. Now you look at your card (representing measuring the spin of one of the particles). And you instantly know what colour -my- card is, half a universe away (representing the spin of the other particle). But because you have carried out an observation, the two particles are no longer entangled and so do not necessarily have opposite spins anymore. (Scratching the ink off your card and repainting it won't change the colour of my card.)
That is why Quantum Entanglement cannot be used for communication: no information is actually being sent anywhere. You -only- know the colour of my card, or the spin of the other particle, because they were opposites all along.
ANNNNCIENT NEWS: TORONTO COMPUTES SEPTEMBER 1990!!
Talk about dusting covers off old ideas! Another example of Canadian know-how creeping into the limelight light years after they're thought forgotten: I remember reading a front page article in 1990 about two Ryerson University CompSci students debating the merits of this very idea and lamenting the lack of funding to really kick things into gear. That was circa 1990-91! I wonder what happened to Messrs Attah-Poku, Karamuna and Mcnair: hopefully they're working on ideas even more exotic these days! Remember hearing of something called a MOACH (Modem On a Chip) back in 1993? Yessir - the same Attah-Poku dude's ideas. I used to really hate them; I was a York U student back then!
- Leaked screenshots show next Windows kernel to be a perfect 10
- Amazon warming up 'cheapo web vid' cannon to SINK Netflix
- Windows Phone will snatch biz No 2 spot from Android – analyst
- Something for the Weekend, Sir? I need a password to BRAKE? What? No! STOP! Aaaargh!
- Episode 13 BOFH: WHERE did this 'fax-enabled' printer UPGRADE come from?