• Patrick (unregistered) in reply to AnOldRelic
    AnOldRelic:
    campkev:
    Anders:
    mb = millibit = (1/1000) bit MB = Megabyte = 1000000 bytes

    How exactly do you have a thousandth of a bit?

    Technically, bits are just off and on voltage. Instead of slamming voltage to high, you incrementally increase it until you get 1/1000 of your maximum capacity.

    That's the theory, anyway.

    Sounds like the sort of thing used in error diffusion. Depending on the type of interference, set the mibibit threshold to separate the 1s from the 0s. Typically, you'd want it on 500, but magnetic fields might require it to go higher or long distance to make it lower.

  • Wheaties (unregistered)

    This was a nice read. I've spent the past year explaining that our bottleneck was disk I/O only to have people suggest extra disk I/O to reduce the computational costs of the project. Of course, the computation must be causing the issue. It's not like 90% of the time is spent in I/O with the CPU at 15%...

  • Kevin Thorpe (unregistered)

    Many moons ago I was a lowly junior operator at an ICL site. Despite being the only trained programmer in the building most of the night I wasn't allowed to touch the hallowed console as I might just know what to do with it (stupid auditors). One fateful night the shift manager came in and (IIRC) CHCP jobnumber 100 (I forget proper VME syntax) which boosted one particular low priority job to 100% cpu because the programmer wanted to go home. Quietly I said 'I really think you should put that back'. Before he could say anything the system coredumped and started avery single bloody printer in the room spitting out core. At least we got a good chunk of the 14 hours overtime it took to do a restore.

  • Living at High Data Rates (unregistered)

    A millibit isn't used as a direct unit. It's used in "millibits per second", the most convenient unit for measuring disk I/O in PDP/11s. Lack of millibits-per-second caused the original problem.

  • (cs) in reply to campkev
    campkev:
    Anders:
    mb = millibit = (1/1000) bit MB = Megabyte = 1000000 bytes

    How exactly do you have a thousandth of a bit?

    You gather up 1000 microbits of course. Now to get a microbit you will need to start collecting nanobits, but that is another story.

  • PG4 (unregistered)

    Good god, I've had to put up with idiots that don't understand priority.

    Recently it was a DBA, only app on the multi-processor UNIX machine. He wanted me to use the nice command on his process because he needed it done sooner. His thing was only using 80% of one CPU.

    Or the FEA users that for some stupid reason restricted their NASTRAN jobs on a VAX to use very little memory. So the program was constantly doing disk IO to load and ship out parts of the big matrix it was working on. "Hey my job ran for 8 hours, you need to increase it's priority" Then I showed them charts of CPU usage while their job was running, it peaked at 25%. "But if you doubled my priority it would peak at 50%."

  • Trevor D'Arcy-Evans (unregistered)

    I've upped my priority, now up yours!

  • (cs)
  • (cs)

    I'm glad there are some new authors, but it is kind of annoying when every sentence is put on a new line.

  • AnOldRelic (unregistered) in reply to enfiskutensykkel
    enfiskutensykkel:
    I'm glad there are some new authors, but it is kind of annoying when every sentence is put on a new line.

    What if the author's dictating for a quadripelegic and wanted to accurately portray the pauses in speech? How insensitive of you...

  • DudeWaitWhat (unregistered) in reply to RandomUser423682
    RandomUser423682:
    Scott:
    Patrick:
    leo:
    Anders:
    mib = mibibit = (1/1048576) bit mb = millibit = (1/1000) bit MiB = Mebibyte = 1048576 bytes MB = Megabyte = 1000000 bytes
    ftfy
    ftfy
    ftfy
    Nah. Clearly leo had it right. 1 megabyte = 1048576 bytes, the same way 1 kilobyte = 1024 bytes, and 1 kilometre = 1024 metres.

    Er... Hmm.

    Well, I suppose the span of Niagara Falls would still be, "fractionally over a kilometre," either way.

    Actually, this is a very good representation of the argument against kibi-, mebi-, gibi-, etc.

    You can't have divisions of a bit. A millibit can't exist. A bit is a Binary digIT. A digit is an indivisible unit. The concept it represents may or may not be, but it itself is indivisible. Thus, half of the SI prefixes for bits don't apply. Following that, NONE of the SI prefixes can apply, since they're a set - all or nothing. So the prefixes used by bits are NOT SI prefixes. They simply mimic SI prefixes in context.

    This proves that we don't need kibi-, mebi-, or gibi- or any of their braindead ilk. Kilo-, mega-, and giga- will do just fine in the context of bits, even with their non-base-10 multipliers.

    And bytes follow this by extension, since they're defined in terms of bits.

    In conclusion, those that continue to push for the use of kibi-, mebi-, gibi-, etc. can die in a fire.

  • Zach Bora (unregistered) in reply to Patrick
    Patrick:
    Ren:
    Bluesman:
    Scott:
    Patrick:
    leo:
    Anders:
    mib = mibibit = (1/1024) bit mb = millibit = (1/1000) bit MiB = Mebibyte = 1048576 bytes MB = Megabyte = 1000000 bytes

    ftfy

    ftfy
    ftfy
    ftfy
    ftfy
    ftfy

    The megabyte is a multiple of the unit byte for digital information storage or transmission with two different values depending on context: 1 048 576 bytes (1024^2) generally for computer memory; and one million bytes (10^6, see prefix mega-) generally for computer storage. The IEEE Standards Board has decided that "Mega will mean 1 000 000", with exceptions allowed for the base-two meaning.
    src: http://en.wikipedia.org/wiki/Megabyte

    I don't get it, why would it be different for memory and storage?

  • Anonymous (unregistered) in reply to Living at High Data Rates
    Living at High Data Rates:
    A millibit isn't used as a direct unit. It's used in "millibits per second", the most convenient unit for measuring disk I/O in PDP/11s. Lack of millibits-per-second caused the original problem.
    I thought that unit was reserved for the speed of the C64's tape drive.
  • John Tantalo (unregistered)

    “a control panel straight out of straight out of Star Trek” should be “a control panel straight out of Star Trek”

    http://www.emendapp.com/sites/thedailywtf.com/edits/0

  • (cs) in reply to Kevin
    Kevin:
    IT. Manipulating the perceptions of its constituents since the dawn of the computing era.
    The other meaning of "optimisation".
  • (cs) in reply to AnOldRelic
    AnOldRelic:
    campkev:
    Anders:
    mb = millibit = (1/1000) bit MB = Megabyte = 1000000 bytes

    How exactly do you have a thousandth of a bit?

    Technically, bits are just off and on voltage. Instead of slamming voltage to high, you incrementally increase it until you get 1/1000 of your maximum capacity.

    That's the theory, anyway.

    No. Technically, bits are units of data that exist in one of exactly two possible states. The concept of low/high voltages is related to one method of representing bits. It's an implementation detail, unrelated to the general concept of a bit.

  • Berzerkly (unregistered) in reply to Schlagwerk

    I can completely relate. In one company I worked at, I had a boss who was put in charge of my department (and I was a department of one) simply because he was friends with the CEO. He was one of these guys who thought that any problem was solvable by just working harder, and that if he spent long enough at it, he could do anything, which based on his track record was not even close to reality.

    This demotivator:

    "When you earnestly believe you can compensate for a lack of skill by doubling your efforts, there's no end to what you can't do"

    described him to a T, and even worse, I had it on my wall in my office and he didn't understand what I was getting at by putting it up. He thought it had a typo.

  • Jack (unregistered)
    one megabyte of memory, up to 63 concurrent jobs
    Most of you programmers today couldn't barf up a useless splash screen with only a megabyte of memory, much less run 63 jobs, even if your life depended on it. The GUI fanbois keep promising it will get easy enough for idiots to use it any day now, meanwhile those of us who actually have work to do are getting more and more hindered by your bloat.

    Get off my lawn. Better yet, get off my planet, and take your bloatware with you.

  • Simpleigh (unregistered) in reply to DudeWaitWhat

    Well, the bit started life as a binary digit, but fractions of bits are certainly useful these days. Try looking at http://en.wikipedia.org/wiki/Entropy_(information_theory) for more info.

  • (cs) in reply to Jack
    Jack:
    The GUI fanbois keep promising it will get easy enough for idiots to use it any day now, meanwhile those of us who actually have work to do are getting more and more hindered by your bloat.

    That damn well better have been posted via Lynx. Me, I hand write all my HTTP requests and send them via cURL.

  • Mike (unregistered) in reply to Richard
    Richard:
    "This lasted for months..."

    And Eric was still working there instead of walking out and letting someone else prop up Kevin? Now there's the WTF...

    Stupidity and ignorance abounds everywhere, it's not restricted to just a few companies. What's the point of jumping ship to another pass-the-duct-tape-and-paper-clip outfit, especially if it means giving up decent benefits? In some cases it's just easier to shut up, play the same game and forget about work on the weekends instead of trying to edumacate the masses & cretins.

  • PITA (unregistered)

    Did someone forget to adjust the speedup loop? http://thedailywtf.com/Articles/The-Speedup-Loop.aspx

  • sino (unregistered) in reply to DOA
    DOA:
    Roger the dodger:
    TRWTF is that Eric didnt just ask Kevin if he was a total retard (like we do in my company-even the MD hasnt escaped-although he finds it funny)
    Wait... you can do that over there? Where do I sign up? I work for peanuts. Hell I'll work for free if I get to tell the boss how mind-numbingly stupid his latest idea is.
    Problem is, at a company with that kind of policy, the boss isn't mind-numbingly stupid, and doesn't generally have mind-numbingly stupid ideas. Just occasionally, or rarely, like the rest of us. Cuz were all sew smrt.

    Of course, that's also not really a problem...

  • mango (unregistered) in reply to Remy Porter

    You mean telnet, surely? Why else would the line terminator in the standard be CRLF instead of just LF?

  • Name Here (unregistered) in reply to Zach Bora
    Zach Bora:
    The megabyte is a multiple of the unit byte for digital information storage or transmission with two different values depending on context: 1 048 576 bytes (1024^2) generally for computer memory; and one million bytes (10^6, see prefix mega-) generally for computer storage. The IEEE Standards Board has decided that "Mega will mean 1 000 000", with exceptions allowed for the base-two meaning.
    src: http://en.wikipedia.org/wiki/Megabyte

    I don't get it, why would it be different for memory and storage?

    Because slimy disk drive vendors wanted to make their product look bigger than it really was, so they would claim it would store a megabyte when in fact it only had room for 1,000,000 bytes (and less than that after overhead). But you can't make RAM in decimal quantities, it is an inherently binary product. So when you buy RAM, a meg is a meg and a gig is a gig.

    Disk drive vendors can suck eggs and die. They're the reason for this push to define "MiB" etc. since they stole the original meaning of "MB".

  • Grig (unregistered)

    Yeah, I worked with a guy who had a boss like this. Wanted "high uptime" on the UNIX boxes. Rebooting was never an option, and saw no reason to have uptimes of a year or more. Eventually, the guy I worked with replaced the "uptime" command to show "uptime + 365days" script (or something), and made an alias with the boss' login to the script. Of course, this didn't affect /proc/uptime or uptime reported by other tools, like top, but the boss was all about "uptime."

  • Jack (unregistered) in reply to Remy Porter
    Remy Porter:
    Jack:
    The GUI fanbois keep promising it will get easy enough for idiots to use it any day now, meanwhile those of us who actually have work to do are getting more and more hindered by your bloat.

    That damn well better have been posted via Lynx. Me, I hand write all my HTTP requests and send them via cURL.

    $ telnet thedailywtf.com 80
Trying 74.50.106.245...
Connected to thedailywtf.com.
Escape character is '^]'.
POST /Comments/Im-Givin-Ye-All-Shes-Got!.aspx?pg=2 HTTP/1.1
Host: thedailywtf.com
...
  • Jo Diggs (unregistered)

    LOL, thats classical stuff dude. Well done.

    Lou www.vpn-privacy.us.tc

  • Mike (unregistered)

    I once had a boss that liked lights and motion. He thought those things meant the equipment was functioning properly. Well, after I had successfully pleaded for a RAID-5 NAS unit with the ability to store months of nightly backups, I had it installed and operational a few days later. Enter the boss.

    "So what does this thing do again?"

    "It stores our nightly backups. We still back up mission-critical data to tape for redundancy. But with this little box I can keep backups much longer and restore them quicker if there's a failure. It's a lot easier to log in to the NAS and restore the missing data than it is to retrieve the tapes from the fire safe and go through the whole backup looking for the missing data."

    "Oh. So what's it doing right now?"

    Seeing as how I had worked with this man for 4 years, I knew what was coming. Thinking quickly, I turned to a terminal and did the first thing that came to mind, created a batch file with a never-ending loop to copy a small JPG file from the web server to a share on the RAID, delete it, and recopy it once every minute. The network and drive activity light immediately responded.

    "OK I see it now. Well, keep me posted. Good work."

    I changed jobs soon after, but as far as I know, that batch file runs 24/7 to this day.

  • Patrick (unregistered) in reply to DudeWaitWhat
    DudeWaitWhat:
    Actually, this is a very good representation of the argument against kibi-, mebi-, gibi-, etc.

    You can't have divisions of a bit. A millibit can't exist. A bit is a Binary digIT. A digit is an indivisible unit. The concept it represents may or may not be, but it itself is indivisible. Thus, half of the SI prefixes for bits don't apply. Following that, NONE of the SI prefixes can apply, since they're a set - all or nothing. So the prefixes used by bits are NOT SI prefixes. They simply mimic SI prefixes in context.

    This proves that we don't need kibi-, mebi-, or gibi- or any of their braindead ilk. Kilo-, mega-, and giga- will do just fine in the context of bits, even with their non-base-10 multipliers.

    And bytes follow this by extension, since they're defined in terms of bits.

    In conclusion, those that continue to push for the use of kibi-, mebi-, gibi-, etc. can die in a fire.

    The units represented by kibi-, mebi-, etc. may have had a reason to exist at some point. In fact, that reason was probably to do with constructing the chips. However, the common user doesn't care what the chips look like. Kilo means thousand, every schoolchild knows that.

    Whoever thought using kilo and mega for those values just because they were close to what their real values represent was an idiot. And probably american. Using commonly accepted and understood terminology to represent something completely different is a much better example of you-need-to-die-in-a-fire. Changing the names of the values that don't conform is the only way to fix this mistake.

    And I'm sure many people will appreciate being able to double something like allocated memory without having to pull out a calculator and hitting "=" over and over until the number reaches something that is just over what common sense tells them it should be.

    Also, just because the other end of the spectrum (the fractions) doesn't apply under normal circumstances, doesn't mean the entire system doesn't work. Suggesting it does means you're either a troll, or too stupid to figure out that you just don't use them. It's like someone gives you a thousand matches and asks you to light a candle. You don't use all of them, just one.

    In conclusion, those that continue to push for the use of the values indicated by kibi-, mebi-, gibi-, etc. can do so at their own inconvenience.

  • Anonymous User (unregistered)

    This is all too familiar. I had an engineer on a PDP-11/70 running DEC RSX-11M who raised the priority of his big batch job to higher than the system console. It was already getting all the CPU, and now we lost access to the system until the job finished a couple hours later. Rebooting wasn't an option.

  • Izhido (unregistered)

    Can somebody just frickin' explain to me what a frickin' "ftfy" is?

  • Franz Kafka (unregistered) in reply to DudeWaitWhat
    DudeWaitWhat:
    In conclusion, those that continue to push for the use of kibi-, mebi-, gibi-, etc. can die in a fire.

    Amen to that.

    Megabyte = 2^20 Gigabyte = 2^30

    See how simple that is? Not my fault some people are all butthurt over us CS guys coopting SI units.

  • PITA (unregistered) in reply to Izhido
    Izhido:
    Can somebody just frickin' explain to me what a frickin' "ftfy" is?

    http://www.internetslang.com/FTFY.asp

  • Franz Kafka (unregistered) in reply to Zach Bora
    Zach Bora:
    I don't get it, why would it be different for memory and storage?

    It isn't. Some HD marketing guys decided to start lying about it so they could claim a larger number on the box and everyone else went along so they could do so too.

  • (cs) in reply to Izhido
    Izhido:
    Can somebody just frickin' explain to me what a frickin' "ftfy" is?
    It's very simple... http://lmgtfy.com/?q=ftfy&l=1

    It amazes me how many people don't just do that when they don't understand something...

  • paul (unregistered) in reply to AnOldRelic

    Didn't you know a gigabyte was actually a bit powered by a gigavolt?

  • jgm (unregistered) in reply to campkev

    Maybe bits are like atoms, and we just have yet to split them.

  • Lego (unregistered) in reply to Izhido
    Izhido:
    Can somebody just frickin' explain to me what a frickin' "Fixed that for you" is?

    ftfy

  • Mel (unregistered) in reply to Mike
    Mike:
    I once had a boss that liked lights and motion. He thought those things meant the equipment was functioning properly.
    Been there. Start with a microcomputer the size of two refrigerators. Add 8 "glass ttys" basically a teletype with a TV screen instead of paper. One person types a command, hits enter, system says "Processing" and a couple minutes later you can enter another command.

    But once we got past the demo and the install and started using it, we discovered that when two people entered a command at about the same time, the first one would run to completion before the second one even said "Processing". So, naturally, with all 8 users doing their jobs as fast as they could, the system was "down" 95% of the time. Because, of course, if it didn't say "Processing" within 1/3 of a second, they'd enter the command again. Five times, just in case.

    The supervisor insisted the system was "down", and nothing the vendor reps could explain would convince her otherwise. Luckily one of the refrigerators had 8 LEDs, you know, the ones you use when you are entering the bootstrap code one byte at a time, with toggle switches? You don't? Well it doesn't really matter to the story, infant.

    Anyway, one of the vendor's tech guys came up with a tight assembler loop that would make those LEDs light up from left to right and then right to left, like a Cylon scanning. What? Cylon? Oh cheese!

    Now, as long as Mrs. Overpaid Technophobic Supervisor could see the blinky-lights, she knew the system was not "down" no matter how slow it got.

    The really nice part about this was the assembly routine was so low in the system that even when the application did crash, the lights kept blinking! I'm afraid we permanently bent Mrs. Supervisor's mind shortly after that...

  • NotaDBA (unregistered) in reply to anon
    anon:
    you must be one of the people who prefer using imperial units to metric

    Ease of conversion between submultiples of units does not outweigh the arbitrariness of scale of those units. Imperial units are based on day-to-day activities on a human scale. Why would you want to fix something that's not broken?

  • lucusloc (unregistered)

    Well, since a binary digit is discreet and cannot actually be less than one, a fraction of a binary digit actually represents the uncertainty of the value. if you have 99/100ths of a bit, that means that you are only 99% certain of the value of that bit.

  • burntangel (unregistered) in reply to DudeWaitWhat

    [quote user="DudeWaitWhat"][quote user="RandomUser423682"]

    Actually, this is a very good representation of the argument against kibi-, mebi-, gibi-, etc.

    You can't have divisions of a bit. A millibit can't exist. A bit is a Binary digIT. A digit is an indivisible unit. The concept it represents may or may not be, but it itself is indivisible.[/quote]

    That comment is the real WTF. It is perfectly possible to have divisions of a bit. Take a look at binary decision tree construction (or information theory) for examples of the use of fractions of a bit, or even millibits.

  • downfall (unregistered) in reply to grizz
    grizz:
    This is the second recent article where a techie decides that the solution to a problem is to secretly rewrite an OS command so that it is silently disabled or produces outright lies in it's output. I've been in the data processing field in a variety of positions, including several where I've had the ability to make changes like the ones described, since the mid 70's and in all that time it's never crossed my mind to do something like this. Is this a common practice that I've been blissfully unaware of all this time?

    Beyond that, it's actually standard practice in embedded systems, because there's no file system to store truthful information.

  • mike (unregistered)

    Process priority (n) pra'w-sess pry-ohr-it'tee : A tool for causing process starvation on an otherwise smoothly-running operating system.

  • fjf (unregistered) in reply to lucusloc
    lucusloc:
    Well, since a binary digit is discreet and cannot actually be less than one, a fraction of a binary digit actually represents the uncertainty of the value. if you have 99/100ths of a bit, that means that you are only 99% certain of the value of that bit.
    Actually more like 99.91% (see the entropy article already posted).
  • fjf (unregistered) in reply to Anonymous
    Anonymous:
    ctw:
    Might it be more accurate to say that the PDP-11, and other tech of the era, inspired the design of Star Trek?
    Maybe, if the PDP-11 could travel through time. Did it do that?
    No, but Star Trek did.

    The picture reminds me of the good old days when computer had cats rather than mice.

  • fjf (unregistered) in reply to NotaDBA
    NotaDBA:
    anon:
    you must be one of the people who prefer using imperial units to metric

    Ease of conversion between submultiples of units does not outweigh the arbitrariness of scale of those units. Imperial units are based on day-to-day activities on a human scale.

    [citation required]

  • (cs)

    A bit is the amount of information gained by observing the outcome of a uniformly distributed binary random variable. If the variable is NOT uniformly distributed, then the information gained by observing its value is, on average, less than one bit per observation.

    Information is defined as the negative logarithm of likelihood. When the base of this logarithm happens to be 2, the resulting unit is called a "bit." If it's base e, it's called a "nat."

  • mike (unregistered) in reply to Zach Bora
    Zach Bora:
    I don't get it, why would it be different for memory and storage?

    Memory is manufactured on silicon, where it almost always makes sense to manufacture in power-of-two units. Further, CPUs logically address memory using bits to count, which makes a power-of-two range more sensible than an arbitrary one based on the number of fingers humans have. Hence 1024.

    Storage (and bandwidth for that matter) has no such manufacturing nor addressing constraints, so 1000 is used because it's easier for us 10-fingered people to understand.

Leave a comment on “I'm Givin' Ye All She's Got!”

Log In or post as a guest

Replying to comment #:

Log Out

« Return to Article