For most of us, the 1970s is remembered as the decade the world let its hair down, smoked some drugs and got really into the electric guitar.
But while that was happening, some more camera-shy folk were industriously defining the future of computing.
Seymour Cray, a US electrical engineer known as the "father of supercomputing", was the first architect of supercomputers, behemoths of vast processing power able to crunch numbers at speeds previously unheard of.
Covering five of the best supercomputers, most are fairly recent entries – but considering the wieght of Cray’s achievement and foresight, CBR has to start with one of the inventor’s own, dating from the mid-seventies (Cray had a pretty normal hairstyle, despite the popularity of Led Zeppelin by this point).
Cray-1 (1976)
This machine remains one of the most successful supercomputers of all time, with 85 sold for netween $5m and $8m each. A revolutionary machine at the time, companies and research laboratories snapped up the Cray-1, which boasted speeds of 80MHz.
Content from our partners
At this point no microprocessors existed, so each individual integrated circuit on the Cray-1 motherboard performed distinct and certain functions, together acting as one huge microprocessor, performing
Cray ensured cooling systems were built into the design to stop the machine melting, and there’s a few tales of companies using the machine to heat their offices in cold months.
Specs
80MHz
136 Megaflops (one megaflop represents one million floating point operations per second)
What was it used for?
Often for modelling nuclear weapons, it being built during the Cold War and all.
Tianhe-2 (MilkyWay-2) (2013)
China’s supercomputer retained the top spot on the Top500 list this year, boasting 1.024 GB of memory.
Built at the National University of Defense Technology, the machine was sponsored by the Chinese government’s 863 High Technology Programme, which aims to make China technologically competitive with other countries.
Specs
33.86 Petaflops
3.12m processor cores
1.4m Petabyte memory
What’s it used for?
Simulation and analysis in government security applications, apparently, as well as predict earthquakes, develop new drugs, design cars and control traffic lights – not to mention creating special effects for films.
ASCI Red (1996)
This supercomputer, built in 1996, is significant for being the first to be built from off-the-shelf CPUs and other commercially available parts.
It was still pricey, mind, coming at a cost of $46m, so putting in those 6,000 200MHz Pentium Pros wasn’t that much of a saving.
ASCI stands for Accelerated Strategic Computing Initiative, a US government scheme to help maintain America’s nuclear arsenal, and the machine met the demand to have a supercomputer capable of surpassing the Teraflop benchmark.
In fact it was later upgraded to 3.1 Teraflops, and stayed in use until as recently as 2005, becoming the world’s fastest supercomputer for four years.
Consisting of 104 cabinets and taking up 2,500 square feet, the ASCI Red was also Intel’s last supercomputer to date.
Specs
3.1 Teraflops
1,212 GB memory
9,632 Pentium II Over-Drive processors, each 333 MHz
What was it used for?
Simulations to ensure America’s nuclear arsenal was safe, secure and reliable without the need for underground testing.
Titan (2012)
Seymour Cray finds his way posthumously back into our list with Titan, produced by his own company’s successor, Cray Research.
Number two in the Top500 list only to the previously mentioned Tianhe-2, boasts a whopping memory of 710TB, and was built at the US government’s Oak Ridge National Laboratory, in Tennessee.
Ten times as powerful as the Jaguar model it was based on, the Titan contains almost 19,000 processing nodes, each of which contains a GPU (graphics processing unit) accelerator designed for video gaming, which consume far less power while producing more processing power.
Specs
710TB memory
17.59 Petaflops
What’s it used for?
Lately it’s simulated the behaviour of 140,000 atoms in E.coli cells, discovering that an amino acid known as Phe396 may be the key to stopping disease-causing cells making people sick. Not bad.
Blue Gene/L (2007)
IBM’s BlueGene project is aimed at designing supercomputers with low power usage, leading to the design of some of the most power efficient supercomputers in history.
Its 2007 model, Blue Gene/L, had more than 100,000 compute nodes and a top performance of 600 Teraflops. But it is perhaps more deserving of its place on the list thanks to its designers’ innovative use of low-power cores instead of fast, power-sapping chips
Specs
213,000 cores
73,728 GB memory
What was it used for?
Blue Gene supercomputers have been used to model climate change effects on flooding risks, crop yields, rainforests and renewable energy.
Titan (2012)
Seymour Cray finds his way posthumously back into our list with Titan, produced by his own company’s successor, Cray Research.
Number two in the Top500 list only to the previously mentioned Tianhe-2, boasts a whopping memory of 710TB, and was built at the US government’s Oak Ridge National Laboratory, in Tennessee.
Ten times as powerful as the Jaguar model it was based on, the Titan contains almost 19,000 processing nodes, each of which contains a GPU (graphics processing unit) accelerator designed for video gaming, which consume far less power while producing more processing power.
Specs
710TB memory
17.59 Petaflops
What’s it used for?
Lately it’s simulated the behaviour of 140,000 atoms in E.coli cells, discovering that an amino acid known as Phe396 may be the key to stopping disease-causing cells making people sick. Not bad.
Blue Gene/L (2007)
IBM’s BlueGene project is aimed at designing supercomputers with low power usage, leading to the design of some of the most power efficient supercomputers in history.
Its 2007 model, Blue Gene/L, had more than 100,000 compute nodes and a top performance of 600 Teraflops. But it is perhaps more deserving of its place on the list thanks to its designers’ innovative use of low-power cores instead of fast, power-sapping chips
Specs
213,000 cores
73,728 GB memory
What was it used for?
Blue Gene supercomputers have been used to model climate change effects on flooding risks, crop yields, rainforests and renewable energy.
