Slashdot Mirror
Scientists Using Supercomputers To Puzzle Out Dinosaur Movement
Nerval's Lobster writes "Scientists at the University of Manchester in England figured out how the largest animal ever to walk on Earth, the 80-ton Argentinosaurus, actually walked on earth. Researchers led by Bill Sellers, Rudolfo Coria and Lee Margetts at the N8 High Performance Computing facility in northern England used a 320 gigaflop/second SGI High Performance Computing Cluster supercomputer called Polaris to model the skeleton and movements of Argentinosaurus. The animal was able to reach a top speed of about 5 mph, with 'a slow, steady gait,' according to the team (PDF). Extrapolating from a few feet of bone, paleontologists were able to estimate the beast weighed between 80 and 100 tons and grew up to 115 feet in length. Polaris not only allowed the team to model the missing parts of the dinosaur and make them move, it did so quickly enough to beat the deadline for PLOS ONE Special Collection on Sauropods, a special edition of the site focusing on new research on sauropods that 'is likely to be the "de facto" international reference for Sauropods for decades to come,' according to a statement from the N8 HPC center. The really exciting thing, according to Coria, was how well Polaris was able to fill in the gaps left by the fossil records. 'It is frustrating there was so little of the original dinosaur fossilized, making any reconstruction difficult,' he said, despite previous research that established some rules of weight distribution, movement and the limits of dinosaurs' biological strength."
SGI IS the dinosaur.
It seems there might be a typo in the article. 320 gigaflops/second is not that much, and can be gotten on a GPU for maybe $100 now. According to this page (http://n8hpc.org.uk/about/facilities), 320 gigaflops/second is the peak performance of each 2-cpu, 16-core sandy bridge node in the cluster, while the entire cluster has a peak performance of 110 teraflops/second. It is possible that this simulation just uses a single node on the cluster, but then it would be hardly worth mentioning the use of the cluster at all.
So, starting with a few partial chunks of data, and guessing the shape of the others based on overall dinosaurs knowledge, we got a very high precision simulation.
At least when we get an election poll, it usually is delivered with a calculated uncertainty margin.
You're telling me that SCIENTISTS, actual scientists, are turning to COMPUTERS, to do their work?
What is the world coming to?
Extrapolating from a few feet of bone, paleontologists were able to estimate the beast weighed between 80 and 100 tons and grew up to 115 feet in length.
A few feet of bone... like what? A femur, perhaps? So, from a couple of feet of femur, they can determine the size and shape of the entire animal? How do they know it was a brontosaur*-style vs a tyrannosaur style animal? How do they know this is the bone of a separate species and isn't the bone of an animal suffering from a glandular defect? Extrapolating the characteristics of a species from one very incomplete sample... I sense a large percentage of handwaving, guesswork and artistic license taking place here.
*Yes, I know it's not called a brontosaur anymore. Poor things went extinct twice.
When our name is on the back of your car, we're behind you all the way!
N8 HPC currently operate Polaris, an SGI High Performance Computing cluster, with 332 compute nodes. Each node has two of the latest Intel Sandy Bridge processors and these nodes have a capacity of 320 GigaFLOPS/second. By using all of the nodes of Polaris together, a peak performance of 110 TeraFLOPS/second. Unlocking this peak performance has been helped by our use of a fast InfiniBand interconnect joining the nodes together.
- http://n8hpc.org.uk/about/facilities
So my GPU isn't three times faster. Oh well. Can't win 'em all.
All rites reversed 2010
I am always amazed at how much under-estimation of dinosaurs there is. Remember when we thought Tyranosaurus Rex was slow? Now, we estimate that it runs at what? up to 35 mph?
Now, I wonder what their model would predict the top speed of an Elephant to be? If you look at an Elephant and it's gait you'd think it slow. But it can reach top speeds of 25 mph. Guess what, if you compare an elephant to most other animals in a simulation. You would not conclude that they could hit 25mph. Why? Because elephants don't really run as other animals. They always have one foot on the ground. Their movement doesn't match. So if you said, take an elephant and model it to see if it can run. It would return back that it was incapable of running. When in fact, elephants have a modified gait that they use for higher speeds.
I'd wager the movement is akin to pogo'ing affect. If you've ever run with your knees locked, you'll understand what I mean. It's clearly not as fast as "running" but it can be faster than walking. But whatever, however, you want to describe it. An elephant exceeds the speeds a typical model would show it capable of.
Now, if you take these premises, and create a model. Your conclusions are going to be "fail". Why? Because you have limited knowledge. Take a large saurapod and model it after a running gait. It will show negative results. So you say, hey, let's see how it does with a more elephant like gait. Cause elephants don't really run, but they can charge at up to 25mph. Results fail. You now conclude that said giant beastie is only capable of 5mph walk.
YOU ARE WRONG!!!! Because you've predicted based on a model without understanding other forms of movement. Would 5mph be sustainable for the achievement of accessing food? avoiding predators? etc. In other words, can a 5mph speed sustain a giant saurapod's lifestyle?
Consider for a moment that many lizards when they run seem to not just use their legs, but rather bend their elongated bodies left and right creating a wiggle affect. Rather similar to what a snake does. It is not just the reach of the legs, but the shortening of the torso that affects the speed and movement of some lizards.
And it may very well be likely that a similar body function would allow greater speed. Most extremely slow moving creatures such as turtles, have a tendency to be armored and protected.
My gutt feels like they're missing an unknown aspect of mechanics. And that such creatures could probably reach at least 8-12mph
Has anyone figured out how much food these things had to eat to grow that big, how long they had to live?
My questions are about the likelyhood of these things being "common" at those sizes.
Agent K: A *person* is smart. People are dumb, stupid, panicky animals, and you know it.
God said his feet hurt when he stepped.
For more context: elephants can run up to 25 km/h
http://www.newscientist.com/article/dn3583-elephants-need-for-speed-defies-definition.html
...which actually sounds pretty good
*stumbles sideways falls over stegosaurus*
crazy dynamite monkey
Extrapolated from one bone???
They once thought these animals couldn't even raise their tales.
The length was too long. They were too big.
Yet look at giant Anacondas. Neither slow nor weak and with a vascular system running blood up and down its length. The truth is, nearly every under-assumption of these creatures has been proven wrong.
Look at our large land animals: elephant, rhino, moose, giraffe, polar bear....NONE OF WHICH ARE SLOW
Now, they're never fast as pinnacle predators cats, wolves, etc. But they're never slow. Rather, it is not speed they tend to lack, but agility. They often can charge forward at a brisk pace, but most are not very good at high speed turns.
I'll still wager $5 they're completely mis-analyzing due to mis-conceived models. A penguin fails at a flight model. But it would be pretty darn stupid to then conclude that a penguin's top speed is a mere 2-3mph. Simply because you never thought to consider an aquatic model.
The really exciting thing, according to Coria, was how well Polaris was able to fill in the gaps left by the fossil records.
And we know it did so correctly because ... ?
"Extrapolating from a few feet of bone..." There's certainly a place for modeling in science, but there are definitely scientists that are using these models based on very little actual data to give some semblance of authority to what is really a wild guess.
... it didn't happen.
...and the cluster consists of 332 nodes. So according to the lab's homepage the whole cluster is able to deliver 110 TFLOPS (Tera Floating-point operations per second). You'd need to buy a couple of GPUs to equal that.
I don't understand what you mean by acceleration unit. Each node delivers that performance instantly. There is no change over time.
Computer simulation made easy -- LibGeoDecomp
I imagine they tried to let the machine figure our how to press the gazillion of buttons to make the dinosaur go by testing thousands of combinations.
Computer simulation made easy -- LibGeoDecomp
I'd just like to add that dinosaurs are awesome.