The Top 10 Supercomputers, Illustrated 68
1sockchuck writes "The twice-a-year list of the Top 500 supercomputers documents the most powerful systems on the planet. Many of these supercomputers are striking not just for their processing power, but for their design and appearance as well. Here's a visual guide to the top finishers in the latest Top 500 list, which was released this week at the SC11 conference."
Re:Every supercomputer should look nice . . . (Score:5, Informative)
Way back in '03, Virginia Tech built a cluster of 1,100 Mac G5's. It came in at #3 on the Top 500 list that year, and at $5.2M, it was a fraction of the cost of the next cheapest supercomputer in the top ten. And it was assembled by students in 3 weeks, using stock G5 towers fitted with InfiniBand cards.
It was later upgraded to G5 xServe boxes, and as of 2008, was still ranked 281 on the Top 500 list [top500.org].
Here's a short promo film that VT produced: http://www.youtube.com/watch?v=vLujLtgBJC0 [youtube.com]
Re:Visual guide? (Score:3, Informative)
Good question (Score:5, Informative)
The Tegra 3 chip that's showing up in phones this spring and Transformer Prime tablet now is about 7.2 GFLOPs [anandtech.com]. That's more than enough to be top 10 in 1993. Current ARM architectures might go all the way up to fast enough to take that number one spot in reference sample designs now but they consume too much power to go in your pocket on retail shelves as yet. Maybe in a year or two.
Mali T658 [hothardware.com] and PowerVR [imgtec.com] are two to watch here. Mali is supposed to go up to 350 GFLOPs. It still amazes me that in 1993 that machine cost about $70 million [chrisvernon.co.uk] in today's money and you can almost match it today for under $500.
Re:Nice rack. (Score:3, Informative)
Though, when you buy a system like that, the cost isn't the hardware, it's the field and support engineers available 24/7, customer support, projects and power consumption that are the big costs. There used to be a joke, "Buy a super-computer from us, and we'll throw the building in for free"
Wrong. Actually, current systems (e.g. Blue Waters) easily cost $200 mio. to procure, and that is just the hardware and support for 1 year, excluding staff, power etc.
Modern day supercomputer systems use a standardized rack frame system and intercommunication fabric so that the oldest and slowest nodes can be pulled out, while the newest and fastest ones can be slotted in straight away. That removes the overhead of having to construct a new building, power supply system, air conditioning and network infrastructure just to do a simple upgrade.
Sorry, but wrong again. Modern supercomputers quite often use custom interconnects (e.g. Cray's Seastar or Gemini or Fujitsu's Tofu). Also, as K and Jaguar show, the cooling solutions are commonly custom, too. This is because node density is growing exponentially and off-the-shelf interconnects and cooling can't keep up with this.