Imagine you were able to solve a problem 50 times faster than you can now. With this ability, you have the potential to come up with answers to even the most complex problems faster than ever before. Researchers behind the U.S. Department of Energy's (DOE) Exascale Computing Project want to make this capability a reality, and are doing so by creating tools and technologies for exascale supercomputers – computing systems at least 50 times faster than those used today. These tools will advance researchers' ability to analyze and visualize complex phenomena such as cancer and nuclear reactors, which will accelerate scientific discovery and innovation.
Developing layers of software that support and connect hardware and applications is critical to making these next-generation systems a reality.
Researchers in Argonne's Mathematics and Computer Science Division are collaborating with colleagues from five other core ECP DOE national laboratories -- Lawrence Berkeley, Lawrence Livermore, Sandia, Oak Ridge and Los Alamos -- in addition to other labs and universities.
Their goal is to create new and adapt existing software technologies to operate at exascale by overcoming challenges found in several key areas, such as memory, power and computational resources.
Summary from ScienceDaily
Additional coverage: Phys.org, Newswise
Source: How to build software for a computer 50 times faster
Why only 50 times faster, why not 100 times (or more) faster ??
(Score: 2, Funny) by aristarchus on Saturday June 17 2017, @08:15AM (2 children)
Was this in the queue? Or did it just magically appear here, because it was 50 times faster than a normal submission? Oh, Tempore! Oh Mores! Oh Huge Manatee? I come here to praise SoylentNews, not to bury it, because it seems to be doing that all by itself, with non-community submissions and random -10 mods to those the powers that be do not agree with. OK, I have some ideas: can a supercomputer eliminate the Mightly Buzzhard bias factor? Can it? Proof needed!
(Score: 2, Informative) by takyon on Saturday June 17 2017, @10:13AM
You seem to have some kind of grievance against what subs make it to the front page. Are you talking about "bot" (Arthur T Knackerbracket) submissions? Because those are used when there is almost nothing left in the submissions list except political trash that is too toxic to even be run here. If you are talking about Mr Plow/exec, those can be submitted from the IRC channel by anyone using a chat command.
If you're talking about some submissions being accepted within minutes rather than hours/days, then I see nothing wrong with that. Some subs are better than others, and some subs stick around until they can be put at a low priority time like the weekend, or until some editor decides to finally delete the sub. Submissions also might happen to be submitted around the times editors are active, around morning or early evening U.S. time, leading them to be considered and accepted very quickly.
If you're calling this particular story a non-community submission, then I invite you to explain how AnonTechie [soylentnews.org] is not part of the community.
Spam mods [soylentnews.org] have been around since 2014. Apparently you got hit by one and couldn't post any more comments anonymously. If your subnet karma dips too low, the anonymous posting restriction will go into effect immediately, even if the Spam mod(s) have not been vetted yet. That can stop a flood of spamming in its tracks, although it is trivial to resume spamming. I can't say whether or not you deserved the -10s or -1s because I haven't checked them. But given your habit of railing against TheMightyBuzzard, Runaway, etc. in off-topic comments in places such as this story, I can't say that I'm surprised you got blasted by downmods.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 0) by Anonymous Coward on Saturday June 17 2017, @04:47PM
Actually your post looks like spam to me. But, maybe I'll be nice, and not mod you down today.
Protip: quityerbitchin
(Score: 0) by Anonymous Coward on Saturday June 17 2017, @09:55AM (2 children)
Talk of building software differently means it's not that fast at some things...
(Score: 0) by Anonymous Coward on Saturday June 17 2017, @10:24AM
(Score: 2) by Dr Spin on Saturday June 17 2017, @05:32PM
I seem to remember using a Cray that was 50 times faster than the PDP8 I had used previously.
The answer was Write the programs in Fortran
Warning: Opening your mouth may invalidate your brain!
(Score: 2) by gringer on Saturday June 17 2017, @10:40AM (3 children)
I frequently make things on a current computer 10 times faster by changing the way a problem is approached, usually by excluding human interaction as much as possible. For some specific extreme cases, (e.g. someone who is copy-pasting single cells in order to transpose a table), 50 times faster is also achievable.
Ask me about Sequencing DNA in front of Linus Torvalds [youtube.com]
(Score: 4, Informative) by fyngyrz on Saturday June 17 2017, @12:28PM (2 children)
You could just use a language that is many times faster than the others: c or c++.
Of course, this would require the hiring of competent programmers, for instance, people who have skills to manage memory and write secure code.
Never mind. Guess you're going to need a 50x faster computer... :)
(Score: 0) by Anonymous Coward on Saturday June 17 2017, @02:33PM
Why not? The computer would be cheaper and someone will probably code the C++ anyway for a hobby project ;)
(Score: 2) by Dr Spin on Saturday June 17 2017, @05:29PM
this would require the hiring of competent programmers,
Which might cost 50 times more the the people they currently use.
Warning: Opening your mouth may invalidate your brain!
(Score: 2) by RamiK on Saturday June 17 2017, @10:51AM
Ahaa the classics... [youtube.com]
compiling...
(Score: 2) by takyon on Saturday June 17 2017, @11:07AM (7 children)
Thomas Sterling: ‘I Think We Will Never Reach Zettaflops’ [hpcwire.com]
The Tiny Chip That Could Disrupt Exascale Computing [nextplatform.com]
Optical Processing Pioneer wins Project with DARPA [optalysys.com]
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by kaszz on Saturday June 17 2017, @01:24PM (4 children)
Optalysys is interesting:
The founder of Optalysys in 2013 is a renowned world expert in optical pattern recognition.
(Score: 2) by takyon on Saturday June 17 2017, @01:42PM (3 children)
It is very interesting but still in the vaporware stage.
https://www.hpcwire.com/2014/08/06/exascale-breakthrough-weve-waiting/ [hpcwire.com]
I have seen no evidence of those two products being released in 2017.
Furthermore, their petaflops may not be real because it might not be general compute but very narrow types of math problems.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by kaszz on Saturday June 17 2017, @02:11PM (2 children)
The key problem with optical processing has been to find a "optical transistor". If they have solved that then it's indeed interesting.
(Score: 2) by takyon on Saturday June 17 2017, @02:26PM (1 child)
If you look at their marketing video from 2014 [youtube.com], they claim to shine lasers at a tunable liquid crystal grid. These lasers can be made to hit multiple grids by design and they eventually recombine to form an image that is captured by a digital sensor. The voltages of the liquid crystal grid cells are the inputs and the image represents the output. You could measure the output in FPS/Hz... getting the Hz higher would require fast switching of the grid cells and the ability of the sensor to capture each output (so a slow motion camera sensor may be useful here).
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by kaszz on Saturday June 17 2017, @03:48PM
Voltages still imply they are hamstringed by the movement of electrons.
(Score: 2) by Dr Spin on Saturday June 17 2017, @05:34PM (1 child)
depending on the amount of pain we are prepared to tolerate.
Don't know how to break this to you but:
You don't actually do the "flopping" yourself. That is what the computer is for!
Warning: Opening your mouth may invalidate your brain!
(Score: 2) by takyon on Saturday June 17 2017, @08:18PM
Come on mate don't be spinnin
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by c0lo on Saturday June 17 2017, @12:05PM (4 children)
How to build software for 50x faster computers?
Well, that's simple: just install Windows and McAffee and you program it as an usual computer.
https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
(Score: 0) by Anonymous Coward on Saturday June 17 2017, @10:24PM (3 children)
Remove Windoze and the associated pseudosecurity cruft.
Of course, the crowd has beaten me to that notion.
Operating systems used on top 500 supercomputers [wikimedia.org]
Usage share of operating systems [wikipedia.org]
(Windoze doesn't scale technically or economically.)
-- OriginalOwner_ [soylentnews.org]
(Score: 2) by c0lo on Saturday June 17 2017, @10:33PM (2 children)
You weren't jesting, then.
https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
(Score: 0) by Anonymous Coward on Saturday June 17 2017, @11:55PM (1 child)
...nor a link.
You -might- have been referencing the rumored abundance of Windoze-only apps.
...though that tends to thin out quickly as we get into the supercomputer realm.
...and there are those who still believe that an anti-virus app actually improves security.
-- OriginalOwner_ [soylentnews.org]
(Score: 2) by c0lo on Sunday June 18 2017, @12:33AM
Mmm... I thought that 50x -> "normalspeed" was a clear giveaway of my /sarc.
Was it that subtle?
https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
(Score: 2) by kaszz on Saturday June 17 2017, @01:28PM
Some lessons from the article subject is that caches and virtual memory translation will impede speed. Seems the key to take advantage of that is to use programming that will not need it.
(Score: 4, Insightful) by choose another one on Saturday June 17 2017, @02:01PM (1 child)
Just make the current software 50x slower, that usually does the job.
Add a few layers of abstraction, because no one can understand the layers that are already there.
Plug in a few more libraries and frameworks, because no one understands the ones it uses already.
Make all UIs scale to super-ultra-thingy resolution by default, add UI scaling to normal resolutions by using a 3D graphics compositing engine because you know it makes sense,
Make all user interaction indirect via cloud servers on the other side of the world to provide auto-complete/correct which will speed things up for users, then add code to make it all async for the times it slows things down.
Continually persist everything to the cloud and local storage in-sync using mulit-phase commits.
I think that should do it...
(Score: 0) by Anonymous Coward on Saturday June 17 2017, @02:39PM
More [OK] [Cancel] dialog boxes as well. Change the order and the wording each time.
[Save] [Cancel] [Done] [Don't Save]
[Don't Cancel] [OK] [Not OK] [Accept All]
And a command line "Press any key to continue" buried in a text box deep, deep in the background.
(Score: 2) by bzipitidoo on Saturday June 17 2017, @02:47PM
I snickered at the article taking the tone that 50x faster was such a miracle. All that more speed does is make it practical to crunch larger data sets. They must mean algorithmic improvements, and reporters dumbed it down.
(Score: 0) by Anonymous Coward on Sunday June 18 2017, @07:26AM
Use bubble sorts everywhere.