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


Forgot your password?

Research Reveals Mislaid Microprocessor Megahertz 99

slashdottit! tm
ransom1982 noted a new article on The Register that says "Not only are chip companies regularly releasing ever-faster microprocessors, but new research has revealed that modern CPUs actually lose megahertz over time." This makes it even more complicated to compare the performance of Intel and AMD CPUs since you have differing architecture, clockspeeds AND the year of manufacture to consider. Buyer beware!
This discussion has been archived. No new comments can be posted.

Research Reveals Mislaid Microprocessor Megahertz

Comments Filter:
  • by Khaed ( 544779 ) on Sunday April 01, 2007 @01:06PM (#18567395)
    This one wouldn't surprise me if it turned out to be true, because sometimes I certainly feel like my processor is degrading...
    • Just look at the references to Einstein's work on relativity as an explanation (!?!).
    • by ColdWetDog ( 752185 ) on Sunday April 01, 2007 @01:16PM (#18567483) Homepage
      You computer guys are all alike. No balls. We car guys don't whine about our wheels going slower, we just get out the drill and polish and start working. The trick is to take weight off the vehicle by drilling out any unnecessary metal. That's called "adding lightness".

      Even geeky guys can do this. And you don't need a real drill, just a dremel tool. Just start drilling holes in those IC's, you soon see the flames roar out! Smoke! Speed!

      Works like a charm. Even works for Windows.

      • Re: (Score:3, Funny)

        by hahiss ( 696716 )
        I just added pin-stripes and a big-ass wing instead. . . .

        Oh, and speed holes. Lots and lots of speed holes.
        • Don't forget you have to add a big Dell sticker (or your computer brand) on your computer to increase speed. It works for the ricers when they put a big Honda sticker on their windshield or an Injen sticker on their window. I'll probably give you an extra 100MHz!
          • by Miseph ( 979059 )
            Dell isn't 31337 enough to get extra MHz, you'd have to slap at least an Alienware badge on.

            Also, computers with both "Intel Core 2 Duo Inside" and "Powered By Athlon64" can get up to a 150MHz boost. For best results, use with a 32 bit Athlon or Pentium; Just don't do it with a Celeron, only posers do shit like that.
      • by EmbeddedJanitor ( 597831 ) on Sunday April 01, 2007 @02:33PM (#18567875)
        I thought parent was posting bullshit, but though what the heck let's try this.... except I didn't want to screw up hardware so I thought I'd take a stab at lightening Windows. I went into the Windows directory and truncated a few DLLs to make them lighter.

        Well who would have expected it... Windows BSODs much faster than it used to!

      • Works like a charm. Even works for Windows.
        I didn't drill anything on my Windows box, but this has already been happening for years!
    • by Sigma 7 ( 266129 ) on Sunday April 01, 2007 @01:41PM (#18567621)

      This one wouldn't surprise me if it turned out to be true, because sometimes I certainly feel like my processor is degrading...

      It is true. Processors don't live forever, and will eventually die from entropy - hence some retailers and manufacturer's selling extended warranties to cover the period between 1-3 years where it's not likely to have noticable problems.

      The rate at which Mhzs are lost does vary - if you run DOS, you can be sure that MHzs would be working at optimum capacity as there's isn't more than 1 task wearing away the ability of the processor.

      When the processor malfunctions or finally dies, that means it's lost too many MHzs to contine operating. It may simply stop working, develop the Pentium division bug (or some other interesting event), or some other symptom. Some people compensated by overclocking the chip to recover these MHz - but removing the heat sink caused these MHz to spew out so rapidly that it blew a hole in the mainboard [].

    • by Smidge204 ( 605297 ) on Sunday April 01, 2007 @01:48PM (#18567655) Journal
      Yeah, they just don't build 'em like they used to. I remember when I got my first Pentium system back in the late 1990s, the cache pipeline burst and I spent that whole evening picking bits out of the carpet.

    • by noamsml ( 868075 )
      Maybe you should try cleaning your fan.
      • Re: (Score:3, Funny)

        by Khaed ( 544779 )
        (AOLer) I don't know what my ceiling fan has to do with it, but okay... (/AOLer)
    • While this is clearly a joke, the problem is real. There are plenty of reliability effects like electromigration [] andNBTI [] that depend upon device aging. These issues wouldn't manifest as in the TFA (although it is conceivable that the clock signal could degrade too), but as overall performance degradation.
    • Basically, the technology press should review a CPU by taking it's throughput under typical conditions and comparing that with the price. This is in contrast to the reviews of the early P4s which basically said: "slightly slower than the AMD and much more expensive - looks like Intel has another winner on its hands with this great product!"
    • If you re-install windows or Linux, your processor will speed up again. Probably because of electricity buildup that gets flushed out. Just make sure you change your electron filtration system every 100 trillion cycles.
  • by andy314159pi ( 787550 ) on Sunday April 01, 2007 @01:07PM (#18567403) Journal
    I'm too dumb to figure out if this is an April fools joke or not.
    • Re: (Score:3, Funny)

      by metlin ( 258108 ) *
      Well, who cares.

      I think we should celebrate - this is the only way we'd see the word laid in any on the front page of Slashdot.
      • Re: (Score:2, Funny)

        by skoaldipper ( 752281 )
        "Research Reveals Mislaid Microprocessor Megahertz"

        For a brief second there, I thought MIT introduced their first prototype for the Fembot Miss Universe competition in 2010.
    • Intel was unavailable for comment. AMD, however, claimed to be well aware of the problem and to be already designing processors capable of working beyond the limits of Einsteinian mathematics - with a negative temporal displacement facility built in.
      I'd say that it's probably a joke. This seems a little goofy to me. IANAE or anyone else important, though.
    • Reading the article, the explanation offered makes it obvious that it's an April Fools' joke. "it's like the megahertz are just leaking away" - what kind of scientist would say that? However, this would actually be an interesting study and it could be that this is true, but not for the reason offered here.
    • Yeah. Me too. But the way I see it, it's actually funny, so there's no way this could be an April fool.
    • Re: (Score:3, Informative)

      by Technician ( 215283 )
      I'm too dumb to figure out if this is an April fools joke or not.

      Let me update you. In simplistic terms on each clock cycle, the CPU performes an operation. Using just that metric, it would not be the chip that changed. If the clock slowed down, so would the chip. Now lets get just a little more complicated.. On each clock cycle, the chip requests information and sends requests, computes results, stores results etc. Sometimes the chip has to wait for several clock cycles such as requesting data from m
      • So there is some truth to the story. I sorta thought that there might be, and some others musing here did also.
      • by tuxicle ( 996538 )
        More than drying heatsink grease, I would think dust accumulation on the heatsinks makes more of a difference. I've seen this happen on a laptop of mine, once I cleaned out the dust, temps dropped and I'm guessing SpeedStep wouldn't have worked against me.
    • by gangien ( 151940 )
      considering the rating of 6*10^23 which i think is close to that chemistry number of molecules that make up a mole or something stupid like that. I forget :P, although i thought the common approximation was 6.22*10^23
    • D00dd itz 4 rl, Im stealin ur megahurtz [].
    • At least you are honest. Thumbs up for that.
  • Slashdottit (Score:3, Insightful)

    by Kryptonian Jor-El ( 970056 ) on Sunday April 01, 2007 @01:10PM (#18567431)
    Ever notice how the slashdottit number is 1 mole [] ? 6 * 10^23 thats great
  • by jd3nn1s ( 613014 ) on Sunday April 01, 2007 @01:14PM (#18567461)
    To me the giveaway was El Reg posting new articles at the weekend :)
  • OMG!!! Ponies!!! (Score:4, Interesting)

    by nbritton ( 823086 ) on Sunday April 01, 2007 @01:29PM (#18567551)
    1. Download: 5 []
    2. Tools > Add-ons > Slashdotter > Options.
    3. Click on styles tab, than change 'Default style'.
    4. ???
    5. OMG!!! Ponies!!!

    So if your lose megahertz's wouldn't that screw up the RTC calculations?
  • Now it is my turn (Score:5, Insightful)

    by houghi ( 78078 ) on Sunday April 01, 2007 @01:33PM (#18567579)
    I should be modded insightful for no reason.
  • I saw this years ago (1990-4?) happen on early SPARC processors. I think I remember 2 or 3 Sparcstations (1,1+, 2's) out of 40 or so we had, suddenly getting pokey. And in about a week or two becoming unusable. This would have been 20-50 Mhz CPUs droping to 5-10 Mhz (or effectivly feeling like that) - I remember them being replaced by SUN for free...Not thats this has any bearing on this story....
    • I have seen this effect three different ways, in real life:

      1. The new laptops and even some desktops have heat pipes. If the heat pipes loose coolant, or become unbonded, then the processor automatically throttles back to avoid overheating. The result is that processors "slow down" or loose MHz.

      2. On some older designs, the crystals have been known to go, and the computers really do loose MHz.

      3. The "Anti-Virus" effect. Some antivirus packages are extremely slow. They use lots of CPU power. After

  • There's no way this story is more interesting than Julianne Moore!


  • by Inda ( 580031 )
    I'm glad to see that you still don't get April Fools day. Maybe next year you'll understand and just post one or two fake stories. I might even crack a smile then.

    I think this was my shortest daily visit to Slashdot.
  • Processors are getting fast, so, it looks like the processors which do not, relatively are getting slower over the time.
  • This explains why my C=1541 drive is so damned slow.
  • I have Google to show the main news, etc and the Slashdot story was linked there, so good job with the jokes...
  • This 'slowing' of cpu's with ageing in years must be the culprit! But hang on, it seems this trend may reverse and some may be enjoying a free clock crank from day one!

    Introduction to Quartz Frequency Standards - Aging wPapers/IntroQuartz/vigaging.htm [] "Aging can be positive or negative [17]. Occasionally, a reversal in aging direction is observed."

  • As we all know part of the reason why we want a new machine every few years (or sooner) is obviously because we see our friends with newer machines and we desire to have something better... Furthermore it is apparent that software tends to get more complex from newer versions of a given piece of software. Today's desktops and laptops are getting so quick, that well, there really isn't any delay with loading any sort of basic business software, especially if you have a fast storage system for your computer (
  • What happens if Chuck Norris punches you?

  • but no one wants you to know about it. Some guys at Ars dipped a mobo in that Tahitian Noni Fruity Juice stuff, and it totally cured the problem. it even increased the mhz 20-30, and thats no lie. Run your system in a bath of that Noni stuff, and Windows won't even crash. I swear.
  • They're testing US made CPUs in the EU. They're measuring in metric seconds, while the CPUs were designed to run on English seconds. The difference is very slight, but it accumulates. This is why we have to adjust for it periodically, by having a leap second. You'll notice these leap seconds are announced in the US by the Naval Observatory, and not at Europe's equivalent, the Greenwich observatory, which runs on the more constant metric time and serves as the background against which we measure the drift*.
    • by cbacba ( 944071 )
      cute april 1 gunk - but sorry the metric second was never instigated. Probably had to do with dividing the day up to 10 hours of daylight, 10 hours of night and a 10 day work week. Also, the metric angle stuff didn't make the cut either - converting those 90 degree right angles into 100 degree right angles just didn't win over much support for some reason. IMO, it was the 10 day week that gutted the whole time thing.

      This is written on apr 2.
      • You're talking about decimal time as proposed by Poincaire. Although the entire system was not adopted, we do use decimal days See the Near Earth Object Impact Hazards Table for instance: [] Look at any of the objects' data and observe the date/time stamp used. Time as date point xx more compact than date/hour/minute/second

        The "metric angle stuff", most commonly known as gradians (400 per circle, 100 per right angle) is commonly used in surveying. Theodolites have 100 divisions
        • by cbacba ( 944071 )
          You are referring to Julian dates, in some form or another as there's half a dozen varieties. These are astronomy related and are not part of the crud attempted at the beginnings of the metric system. And yes they are more condensed than typical date time indicators although I prefer my days to end at midnight, not at lunch time.

          Nothing like a calculator getting screwed up while under time pressure. As pointed out, the change of degrees from 360 to 400 was not something that happened in a wide spread fas

Any sufficiently advanced technology is indistinguishable from a rigged demo.