Slashdot Mirror
Obama's New Executive Order Says the US Must Build an Exascale Supercomputer
Jason Koebler writes: President Obama has signed an executive order authorizing a new supercomputing research initiative with the goal of creating the fastest supercomputers ever devised. The National Strategic Computing Initiative, or NSCI, will attempt to build the first ever exascale computer, 30 times faster than today's fastest supercomputer. Motherboard reports: "The initiative will primarily be a partnership between the Department of Energy, Department of Defense, and National Science Foundation, which will be designing supercomputers primarily for use by NASA, the FBI, the National Institutes of Health, the Department of Homeland Security, and NOAA. Each of those agencies will be allowed to provide input during the early stages of the development of these new computers."
... NSA data center and stuff.
It little behooves the best of us to comment on the rest of us.
For most specific problems thrown at supercomputers, you can go 30 times faster with a custom hardware architecture baked into silicon
To go 30 times fast for general purpose supercomputing, you use the latest silicon (2X) and more chips (15X) and come up with a super new interconnect to make it not suck. This would involve making some chips that support low latency IPC in hardware.
They are free to send me a few billion dollars, I'll get right on it and deliver a 30X faster machine and I'l even use some blue LEDs on the front panel.
I should use this sig to advertise my book ISBN-13 : 978-1501515132.
Exactly.
My first thought was the new addition will be tasked by the NSA/FiveEyes to break encryption for intercepted communications.
Why are you assuming they don't already have one doing that, and this is just a public version?
There is a lot of highly secured government data infrastructure out there that I hear about even though not inquiring. The cable in Virginia that gets cut by a backhoe accidentally and guys in a black van show up ten minutes later. The contract for a government data center inside a faraday cage. The government likely already has much more computing power available than we know about.
Those issues will be resolved by a side effect of this being a government order. According to the GAO, on average it takes 4 1/2 years from the time the government orders a computer until it's installed. Right now, multiple government agencies have been told to start thinking about a plan. In two years (2017), each agency will have their plan and they'll start working to to resolve the differences between agencies. In another year (2018), they'll put out some RFPs. Those will go through the federal procurement process and the order will be placed about two years later (2020). That's when the 4 1/2 year average clock starts, so expect installation around first quarter 2025.
The goal is that it should be 30 times faster than TODAY'S computers.
And be operational in ten years. They can pretty much just order a Nexus 47, or an HP Proliant gen 12.
Weather guys want this after NSA's done.
We'll take a side of phased-array weather radar to go with that, too.
The "Civilized World" jumped the shark ca. 1973.
Weather guys want this after NSA's done.
I'm a weather guy - running cloud model code on Blue Waters, the fastest petascale machine for research in the U.S. I don't think we've managed to get any weather code run much more than 1 PF sustained - if even that. So it's not like you can compile WRF and run it with 10 million MPI ranks and call it a day. Ensembles? Well that's another story.
Exascale machines are going to have to be a lot different than petascale machines (which aren't all that different topologically than terascale machines) in order to be useful to scientists and in order to no require their own nuclear power plant to run. And I don't think we know what that topology will look like yet. A thousand cores per node? That should be fun; sounds like a GPU. Regardless, legacy weather code will need to be rewritten or more likely new models will need to be written from scratch in order to do more intelligent multithreading as opposed to mostly-MPI which is what we have today.
When asked at the Blue Waters Symposium this May to prognosticate on the future coding paradigm for exascale machines, Steven Scott (Senior VP and CTO of Cray) said we'll probably still be using MPI + OpenMP. If that's the case we're gonna have to be a hell of a lot more creative with OpenMP.
I'm not a weather guy, but my understanding is that a somewhat fixed weather model (set of calculations) is used to do a kind of finite-element analysis on small areas. With better computing and better radars, smaller and smaller areas can be calculated, which results in more accuracy.
With more computing power, could you not vary the parameters or constants used in the weather model, then run the finite-element analysis over the entire weather area again? You could be running hundreds or thousands of slightly different weather models, then apply some processing to figure out which is most likely- either by averaging together the 50% most similar outcomes, or by some other method. I don't think you could peak out a supercomputer with that method if you kept adding more parameter variations, although you may get to the point where adding more parameter variations doesn't improve accuracy.
Maybe that's an incorrect understanding, but we're getting closer to the point where we can calculate all possible outcomes simultaneously. I wouldn't have expected this to be the case with weather but computing has come a long way in the last 20 years.
Even those who arrange and design shrubberies are under considerable economic stress at this period in history.