Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
×
AI Math Robotics Science

DARPA's Latest Chip Is Designed To Be Bad At Arithmetic (technologyreview.com) 192

Reader holy_calamity writes: Pentagon research agency DARPA has funded the creation of a chip incapable of correct arithmetic, in the hope of making computers better at understanding the real world. A chip that can't guarantee that every calculation is perfect can still get good results on many problems but needs fewer circuits and burns less energy, says Joseph Bates, cofounder and CEO of Singular Computing. The S1 chip can process noisy data like video very efficiently because it doesn't need the extra circuits or operations needed to ensure every mathematical operation is performed perfectly. This summer DARPA will put five prototype computers, each equipped with 16 of the inexact S1 chips, online for researchers to experiment with.
This discussion has been archived. No new comments can be posted.

DARPA's Latest Chip Is Designed To Be Bad At Arithmetic

Comments Filter:
  • by cogeek ( 2425448 ) on Thursday April 14, 2016 @03:52PM (#51910807)
    Just program the chips to use Common Core math
    • Re:Simple Solution (Score:4, Insightful)

      by werepants ( 1912634 ) on Thursday April 14, 2016 @04:26PM (#51911095)

      I know whining about common core is a popular pastime among people who have an irrational fear of change, but have you ever taken more than a few seconds to actually look at how common core teaches things? It has much more in common with real mental arithmetic than the standard method we all learned in grade school, and is very intuitive if you actually take a moment to understand it.

      It's funny, because most of the complaints I've seen cherry-pick examples to intentionally make common core look more complex, but gloss over the convoluted aspects of the standard method (75+22 makes standard method look obvious, but 99 + 99 has many more steps because of the carrying). I for one (and probably most people) never actually carry numbers mentally - in the previous example I would add 100 + 100 and then subtract two, or some other shortcut that fits with human cognition rather than optimizing to be easy to write on a whiteboard.

      • Re:Simple Solution (Score:5, Informative)

        by cogeek ( 2425448 ) on Thursday April 14, 2016 @04:37PM (#51911217)
        Thanks for jumping to conclusions, but I happen to have a BS in Civil Engineering, so I've done my share of math. I also have 4 children currently ranging from HS to College level and all have experienced Common Core to varying degrees. I did my homework when Common Core was first being discussed and all the way through it being shoved down my childrens' and their teachers' throats, have you? It was approved by one individual to the outcry of every single other person on the review board. It's complete and utter rubbish. Anytime you'd like to sit down and solve a complex math problem using your Common Core vs my usage of Common Sense, please let me know. I'll even give you a 10 minute head start.
        • by Dadoo ( 899435 )

          Thanks for jumping to conclusions, but I happen to have a BS in Civil Engineering, so I've done my share of math.

          I guess I'd have to ask what's wrong with you, then. Ever since I started helping my kids on their homework, I've started doing math their way, since it's much easier to do in your head.

          • by Anonymous Coward

            It's easier to solve math problems by drawing a bunch of boxes and shit? And you do that in your head?

            Common core is "math for retards".

        • To be strictly pedantic about the whole thing... I agree that the curriculum foisted upon us under the guise of "common core" has been terrible to date. However, this crappy way to teach math existed prior to "common core" and just happens to have the new sticker on it. The one our school used was "Everyday Math" from the University of Chicago. Horrendous. Common Core does not require this kind of teaching, and sticking with Common Core as a philosophy may actually see this style of kitchen sink mathematics

          • However, this crappy way to teach math existed prior to "common core" and just happens to have the new sticker on it.

            It doesnt have a new sticker on it. Whats new is the penalties for not doing it The Only Way Now Allowed.

            • by pnutjam ( 523990 )
              Bad teachers and admins, not bad systems. However, there is nothing wrong with docking a student who doesn't follow directions. My child has demonstrated time and again he doesn't like peas, I still make him try them. Children need to learn to fail as well as learn to succeed.
              • Bad teachers and admins, not bad systems.

                Common Core, the Solution that attacks the Wrong Problem. Thanks.

                • by pnutjam ( 523990 )
                  Education is all about local control, it's a conservative dream. Common Core introduces some non-locally controlled standards for everyone to meet, but your on your own to fix the other stuff.

                  Plenty of localities do it right.
        • Re:Simple Solution (Score:5, Insightful)

          by yodleboy ( 982200 ) on Thursday April 14, 2016 @06:25PM (#51912001)
          "I happen to have a BS in Civil Engineering"

          You do understand that your math skills are probably far greater than the majority of parents sending kids to public school right? Your kids are fortunate enough to have a parent that really gets math and can help them understand. Let's face it, not every kid is going to need to do complex math when they grow up. Common Core seems to be very good at teaching kids practical math that they can do mentally. Something the majority of adults I run into seem to have a lot of trouble with. I have two kids in elementary school, and both of them regularly impress with their math skills. I'm OK at math and my wife is actually good at it, so sometimes we cringe when we see their homework assignments, but the facts are, they are learning math, they can apply math even at this age, and they are not intimidated to try more difficult math later on. I always wonder how many kids with actual aptitude have been turned off at an early age because of math taught in traditional ways.
          • You do understand that your math skills are probably far greater than the majority of parents sending kids to public school right?

            Is that an excuse for teaching the foundations of maths in a way that is most suitable for someone counting change at McDonalds?

            We should be aiming to make engineers and geniuses and then demoting the people who don't understand it to common core, not teaching for the minimum and hoping for the best.

            I always wonder how many kids with actual aptitude have been turned off at an early age because of math taught in traditional ways

            I wouldn't wonder that. People with aptitude typically don't have a problem being given a complicated problem. That turns off the people without aptitude, which is good because let's face it, they're not going t

        • Yes, I have looked at it. My wife taught with it for a few years and observed it working better for many of her students. And, a good friend of mine is a highly acclaimed math teacher (Bill Gates took her out to lunch a few years back to hear her ideas about math education) and she's a fan of common core. Just because something is different than the way you learned it doesn't mean it's worse. Any unfamiliar method will seem more complex at first.

        • My daughter is going through it now. Some of it is very good such as doing long division. It's so good that I've adopted it in the rare times that I need to do long division.

          _ _ _
          40 | 970 -- now we (as adults know that 40 goes into one thousand 25 times and the answer is one less (24) with 10 left over.

          Let's look at it from a child's perspective who doesn't know what we know.

          I had a good ascii representation but Slashdot's filter prevented it.
      • by Kneo24 ( 688412 )
        Well, I guess I'm not normal, as I tend do both methods depending on the situation. If I'm strapped for time, I go the shortcut route. If I'm not strapped for time, I enjoy the small bit of mental work that it requires, which really isn't all that much.
      • by tlambert ( 566799 ) on Thursday April 14, 2016 @04:50PM (#51911335)

        I know whining about common core is a popular pastime among people who have an irrational fear of change,[...]

        Actually, most of the Millennials who learned math via Common Core have an irrational fear of change.

        For example, I tried to give this young woman at Panda Express 12 dollars and 12 cents, because the bill was 6 dollars and 87 cents, so that I could get a $5 bill and one quarter back from the transaction so I wouldn't have to carry around so many separate bills or extra coins, and she looked apoplectic.

        I thought she was going to cry.

        She simply could not cope with the change...

        Because she could not do simple math in her head.

        • by cogeek ( 2425448 ) on Thursday April 14, 2016 @05:00PM (#51911433)
          You should have provided her a piece of paper and a pencil so she could draw out 1,212 little circles and then cross off 687 of them, then count up the remaining circles. Would be a system she's likely to be much more familiar with, unfortunately.
          • That's not common core though, that's a stereotype.

            • That's exactly the method my son was taught for subtraction. For division, they draw lines, circle groups of the lines, and then count the groups. Much of Common Core math is removing the whole "working with numbers" out of math. I'm not sure if the creators (who weren't educators, by the way) thought "numbers are too hard for kids to understand so let's go with lines and boxes", but that's the end result.

              • Thats a method used to help someone understand what subtraction is, it's something you do in lower grades. It's not a method taught about how to actually do this in real life.

                In the 60s this is how we did it, drew circles on the board, drew diagonals in them to represent fractions. For subtraction it was "if you have 20 pennies and I take away 7, how many do you have?" And so forth. Teach the concepts, then teach how to make use of it.

                • I wouldn't have a problem if this was a method they used for a week or two to teach the concept and then moved onto real numbers, but this is THE method that they are using for solving actual math problems - long after the kids are comfortable with the concept of subtraction (or addition, multiplication, or division). If kids use the traditional method, they are marked as wrong even if they show their work and get the right answer.

          • by pnutjam ( 523990 )
            Why don't you come up with a better way to visualize the mental process this is illustrating?
            • Why don't you come up with a better way to visualize the mental process this is illustrating?

              Hunh.

              I was not aware that all kids were visual learners, as your statement seems to assume, since you want them to visualize the result.

              However, men tend to have brain structures that are, on average, more visually oriented than the brain structures in women.

              Perhaps this method of teaching, to only visual learnes, explains the dearth of women in STEM...

              • by pnutjam ( 523990 )
                Visual, might not be the right words. Since this is new, I don't see how it could be the driving factor behind lack of women in STEM.

                When I say visual, I mean a way the teacher can demonstrate without trying to verbally explain a thought process.
        • by Dadoo ( 899435 )

          For example, I tried to give this young woman at Panda Express 12 dollars and 12 cents, because the bill was 6 dollars and 87 cents, so that I could get a $5 bill and one quarter back from the transaction so I wouldn't have to carry around so many separate bills or extra coins, and she looked apoplectic.

          Like it or not, some people are just stupid. I remember about 20 years ago, I went to a junkyard to get a couple of parts for my car. The guy looked flustered for a few seconds, as he tried to add up the cos

          • by stdarg ( 456557 )

            That's a good point, but it shows the futility of common core. We're not going to make every kid (or even most) a math star. And why should we?

        • That's nothing. I should tell you about what happened that one time I tried to pay with a $2 dollar bill at Taco Bell...

        • by Anonymous Coward on Thursday April 14, 2016 @07:21PM (#51912247)

          For example, I tried to give this young woman at Panda Express 12 dollars and 12 cents, because the bill was 6 dollars and 87 cents, so that I could get a $5 bill and one quarter back from the transaction so I wouldn't have to carry around so many separate bills or extra coins, and she looked apoplectic.

          That's funny, because the fast food restaurants I've visited tend to have this thing called a "cash register" where the employee keys in the *exact* amount that customer wants to pay with. She wouldn't have had to do any math whatsoever if you had handed her the amount you claim you did.

          What probably happened was that you handed her a ten, she posted the transaction, and like a true jackass you said, "Oh wait I got change," and started counting out the remaining $2.12 in nickels, dimes, pennies, and atm receipts. She saw the line forming behind you, rolled her eyes, and your brain registered that as "bitches can't do math."

          But yeah, I feel you: some people have a hard time with change.

        • Yup, kids can't do math today. And that's up into the twenties. Granted, the most brilliant kids aren't taking those jobs, and certainly anyone with any degree of affluence isn't taking any job anymore (I guess it no longer builds character). But when they can't even make change then you gotta blame their education.
          However that faulty education was done BEFORE common core.

        • by radarskiy ( 2874255 ) on Thursday April 14, 2016 @08:40PM (#51912603)

          "so I wouldn't have to carry around so many separate bills or extra coins"

          Sounds like you couldn't cope with change, either.

        • by dave420 ( 699308 )

          One person does not a generation make. You seem to assume that she was having the perfect day, with nothing clouding her abilities. Your desire to overlook common courtesies when witnessing someone having difficulty and leaping to the conclusion of "they're fucked in the head" speaks more of you than the hapless person in your story. I've seen people of all ages choke when making change at some point, but unlike you I don't tar everyone with some convenient, snuggly brush. If that was the case I'd tar y

          • One person does not a generation make.

            It was an anecdote about an incident that was not isolated. It's rather common.

            You seem to assume that she was having the perfect day, with nothing clouding her abilities.

            OK, I'll cop to that. You're right: she could have been high on weed, or something similar, making her unable to do simple math.

            Your desire to overlook common courtesies when witnessing someone having difficulty and leaping to the conclusion of "they're fucked in the head" speaks more of you than the hapless person in your story.

            And now, you are projecting your own bias. I in fact did tell her what the change would be, after she looked upset, and told her to "just type the amount into the cash register".

            But you are right to call her "hapless"; she was definitely "unfortunate, unlucky, luckless, out of luck, ill-starred, ill-f

        • Actually, most of the Millennials who learned math via Common Core have an irrational fear of change.

          Anybody old enough to work at Panda Express probably wasn't educated via Common Core. It has only really taken off in the last few years (at least in my area) so current teenagers probably learned the standard method. So that's really an indictment of math prior to Common Core, and sounds like exactly the kind of outcome it's trying to address: students who can endlessly work long division problems out on paper but have no real intuition for numbers or ability to apply their knowledge outside of the classro

        • by pnutjam ( 523990 )
          sure, because that's a new problem. I've seen this for decades. Most people can't do math in their head, and most cashiers today didn't learn common core, which, as established above, started around 2009. I would posit that common core might make it easier for people to do simple math in their head.

          Save the complex paper techniques for people who really need to do hard math.
          • sure, because that's a new problem. I've seen this for decades. Most people can't do math in their head, and most cashiers today didn't learn common core, which, as established above, started around 2009. I would posit that common core might make it easier for people to do simple math in their head.

            Save the complex paper techniques for people who really need to do hard math.

            You realize that for an 18 year old cashier, that would mean 6.5 years ago at the shortest, which would have made her 11.5 years old at the time, which would have put her in 6th grade at the time "it really caught on", according to you, right?

            So yes, she was taught "Common Core" methods.

            • by pnutjam ( 523990 )
              I clearly said it started in 2009. I know it didn't catch on before then. I have an almost 15 year old who has had no exposure to common core. I remember noticing the different way my younger kids were doing math problems.
              To be honost, I didn't always agree with the way my older daughter was taught math. I had to show her alternative methods and assure her that I would back her up with the teacher if she used the method that worked best for her. That's pretty much what the Common Core (for math) does today
      • What I've observed (and I have a BA in Math and a MS in Electrical Engineering) is that Common Core attempts to teach the concepts of math and not necessarily the most efficient algorithms for calculating an answer. And I may be a bit uncharitable here, but the complaints I generally see about Common Core seem to be from people who are either not so good at math so they can't see doing arithmetic any way other than how they were taught, or people who are good enough at math but see it as a tool rather than

        • That's basically the same observation I've had - people who actually teach this stuff, or who understand the problems Common Core was created to address, are generally in favor of it. People who are very comfortable with the standard method see something that isn't instantly familiar and become suspicious. I tend to think building intuition about numbers will serve students far better than memorizing an efficient system for longhand arithmetic. Because honestly, nobody ever needs to do that anymore, so we o

      • You are a fool or just plain naive. The problem with common core is that it tries to teach a way of thinking and not the material at hand. For example, some people are just inherently good at numbers. They can add two 6 digit numbers in their head on the fly. Common core requires this person and the kid struggling in math to tackle the problem in the same way. It ignores the fact that people think differently.

        Common core is equivalent to you filling out a document top to bottom at the DMV and bringing it
        • It ignores the fact that people think differently.

          It's actually the opposite - Common Core provides multiple ways to think about and approach a problem, rather than forcing every student into a one-size-fits-all approach. It focuses on building intuition about numbers, helping kids understand how math relates to real objects and quantities in the world, rather than just trying to mold kids into human calculators. And I can tell you, that's something sorely missing - there are plenty of kids that come out of the traditional method able to do algebra manipul

  • by gurps_npc ( 621217 ) on Thursday April 14, 2016 @03:54PM (#51910823) Homepage

    As in, in order to get a real AI, it will need to have this fuzzy logic.

    Which by the way will end up making our new Robotic overlords require human slaves to do math for them.

    Which we will do incorrectly, causing their entire robotic empire to fall in a matter of hours.

    • by Shawn Willden ( 2914343 ) on Thursday April 14, 2016 @04:47PM (#51911313)

      As in, in order to get a real AI, it will need to have this fuzzy logic.

      Which by the way will end up making our new Robotic overlords require human slaves to do math for them.

      Which we will do incorrectly, causing their entire robotic empire to fall in a matter of hours.

      I've thought this for a long time (well, not the humorous bits), that it's entirely possible that general intelligence fundamentally requires fuzziness and imprecision, and that by the time we succeed at creating really smart artificial intelligences, we'll find that they're just as error prone and fallible as people are. Which isn't to say we can't make them smarter than we are or that they won't be incredibly useful.

      I also wonder if we will find we need to make systems that are deliberately crippled for many of the tasks we want them to do. I mean, imagine installing Marvin the Paranoid Android's brain in a car and requiring it to spend all of its time driving people around. It would likely soon drive off a bridge just to end the boredom. Yeah, Marvin is a fictional character, but it's believable to me that true artificial intelligences will get bored, distracted, make mistakes, etc., just like people, making them perhaps not much better than people at many of the moderately mindless tasks we'd like them to take on. So we'll have to limit them to make them good at what we want them to do... which may make them not so good at what we want them to do.

      I think we may also have to deal with the equivalent of mental illness in AIs, and be unable to fully diagnose and/or fix the problems because the system is complex enough to be opaque to us, the same way we don't (and may never) fully understand our own brains and their malfunctions.

      Intelligences general and flexible enough to do anything may do everything somewhat badly, and systems sufficiently specialized to do a task extremely well may be unable to cope with the unexpected. Or not. The next few decades are going to be very interesting.

      • Yeah, Marvin is a fictional character, but it's believable to me that true artificial intelligences will get bored, distracted, make mistakes, etc., just like people, making them perhaps not much better than people at many of the moderately mindless tasks we'd like them to take on. So we'll have to limit them to make them good at what we want them to do... which may make them not so good at what we want them to do.

        You would need to implement emotions. Deep learning is still completely void of emotions, and that makes it better at what it does than humans.

        • Yeah, Marvin is a fictional character, but it's believable to me that true artificial intelligences will get bored, distracted, make mistakes, etc., just like people, making them perhaps not much better than people at many of the moderately mindless tasks we'd like them to take on. So we'll have to limit them to make them good at what we want them to do... which may make them not so good at what we want them to do.

          You would need to implement emotions. Deep learning is still completely void of emotions, and that makes it better at what it does than humans.

          Is boredom an emotion? I don't think so, and I could see a self-aware general-purpose AI deciding it just doesn't want to do what it's been asked to, because it's boring.

    • As in, in order to get a real AI, it will need to have this fuzzy logic.

      Which by the way will end up making our new Robotic overlords require human slaves to do math for them.

      Which we will do incorrectly, causing their entire robotic empire to fall in a matter of hours.

      Well, I'm pretty sure it was Hofstadter. One of the discussions on AI and computers I read a long time ago posited (quoting someone else) that an AI would quite possibly not be very good at math. Or rather, would only be about as good at basic math as a human with a calculator would be (ie, good, but not perfect, due to residual/external issues).

      Intelligence operates at a different level.

    • As in, in order to get a real AI, it will need to have this fuzzy logic.

      I think the key is that sloppy arithmetic is much faster, while still being good enough.

  • Finally, a chip that's close enough for government work!

    • Maybe the next, even more imprecise version will be able to do copyright math for the MPAA and RIAA.

      • They're way ahead of you. Where do you think they get the figures they use for damages?

        "Let's see... this person shared 12 songs with 37 people... OK that's $17,092,259 in damages."

      • The "holy grail" of imprecise arithmetic research is to automate Hollywood accounting. We're still a long way off from that.

  • I have a bad feeling about this.

    Maybe its intended for political calculations!

  • So this is just lossy compression implemented in hardware?

  • They invented a Pentium 4?? Wait...no they said it was power efficient...i'm confused now!

  • by Gravis Zero ( 934156 ) on Thursday April 14, 2016 @04:23PM (#51911051)

    Does it run Linus... cause you know, close enough.

  • I'n tryimg onc uf thees ch1ps nnow. Is't rael1y co0l!

  • by Daniel Matthews ( 4112743 ) on Thursday April 14, 2016 @04:41PM (#51911253)
    Why spend all that money on research when Microsoft already had the perfect product for their needs.
  • So, how much of my taxpayer dollars were spent creating a computer that works about as well and reliably as a stoner highschool dropout? Shall we call it the Spicoli Chip? If you criticize it's answers, does it just take a solid hit off it's digital doobie and say to you "..that's just, like, your opinion, man"? Should we refer to this as 'Millennial processing'?
    • by hey! ( 33014 )

      Actually, this reminds me a lot of the way the human brain works, particularly in the area of visual perception, where it processes exceedingly messy data to come up with useful, timely, but not particularly reliable results.

    • If you criticize it's answers, does it just take a solid hit off it's digital doobie and say to you "..that's just, like, your opinion, man"? Should we refer to this as 'Millennial processing'?

      I'm pretty sure "that's just, like, your opinion, man" would "boomer processing" not "milennial processing". I know millenials get blamed for a lot of stuff, but they didn't invent hippies getting stoned.

      • If you want to be factually accurate then sure, but if you're going for humor, then more current-events-oriented works better. Or are you criticizing me for being less than accurate? Please, you'll make DARPA cry! XDDD
  • This processor will have eight common cores and be capable of simulating such real-world situations as a table of foodies fighting over how to divide a restaurant check.

  • A machine that can be replaced by ME!
  • Remember: (Score:5, Funny)

    by Locke2005 ( 849178 ) on Thursday April 14, 2016 @06:05PM (#51911883)
    At Intel, quality is job number 0.99998643!
  • Many CD players, for example, use 1-bit DACs. These turn the PCM signal into a stream of bits where the average density corresponds to the signal. I would imagine you could construct circuits to process things like images of video where the average 'pressure or density' of bits output would be the meaningful output.

    • I for one cant wait for 3-bit RGB and the 3072000 FPS needed to make it work as well as 24-bit RGB.

      (a 1-bit DAC suitable for 48khz audio operates at 3.072mhz)
      • why have a fixed clock cycle for display in the first place? maybe you could exploit the physical characteristics of the display to cover up whatever fluctuations there might be in the video stream.

  • For what? So Washington can come up with flawed budget proposals faster so they don't have to sit around Washington as much as they did when they had to come up with flawed budget proposals by hand?

  • ...for government work.

  • When using floating point math you always have an imprecise representation of the actual number. You might be lucky that the finite number of bits will be able to represent the intended number exactly, but when you start with analog values and convert them to digital, you always add noise.

    This "new" technology sound as if they move from double precision back to single precision. Of course it needs less circuitry and power.

    • but when you start with analog values and convert them to digital, you always add noise.

      Yesbut.

      I mean yes you do. However, each stage of analog processing adds noise too, and you can often make the quantization noise sit well below the analog noise floor.

  • Will it have "BRME" (BRanch if Maybe Equal), "MOVMO" (MOVe Most Of), "ADDFS" (ADD Forgetting Sign), or other interesting instructions?

    Personally, I've always wished there was an "if almost equal" statement available instead if having to code "if ( abs(actual - desired) eh_close_enough )" but to have the hardware do something like for me? Sweet!

It is now pitch dark. If you proceed, you will likely fall into a pit.

Working...