Stories
Slash Boxes
Comments

SoylentNews is people

This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 1) by pTamok on Sunday July 01 2018, @10:57AM (3 children)

    by pTamok (3042) on Sunday July 01 2018, @10:57AM (#700927)

    Oh, and just to add, as far as telecomms is concerned a kilobit is 1 000 bits, NOT 1 024 bits, and a Megabit is 1 000 000 bits, NOT 1 048 576 bits. Some applications report capacities in Mebibytes per second (with a byte assumed to be 8 bits) which is just plain RONG.

  • (Score: 2) by Arik on Friday July 06 2018, @06:25AM

    by Arik (4543) on Friday July 06 2018, @06:25AM (#703403) Journal
    Yeah, nah, hang yourself mate.

    :D

    /aussie

    A megabit is and has always been 1024 bits. If you decimal weenies want to have a 1000 bit unit then you need to invent your own.

    There are only 10 types of people. Those who think in binary, and those who don't. Think.
    --
    If laughter is the best medicine, who are the best doctors?
  • (Score: 2) by bitstream on Sunday July 15 2018, @01:41AM (1 child)

    by bitstream (6144) on Sunday July 15 2018, @01:41AM (#707401) Journal

    When handling memory, base 2 works very good. Telecomms has been in base 10 for a long time..
    However HDD capacity going from base-2 to base-10 sucked.

    Oh and the units KiB, MiB, GiB *yuck*..

    • (Score: 1) by pTamok on Monday July 16 2018, @11:09AM

      by pTamok (3042) on Monday July 16 2018, @11:09AM (#707853)

      Well, it comes down to whoever chose the 8000 samples per second sampling rate for PCM audio on voice telephony calls.

      Experimental testing shows that most of the intelligibility of the human voice is maintained if you transmit only the frequencies between roughly 340 and 3400 Hz. Allowing more frequencies above about 3400 Hz improves the perceived quality, but doesn't add a great deal to the intelligibility.
      Application of the Nyquist-Shannon sampling theorem tells you that to be able to digitally sample a signal of up to 3400 Hz so you can reconstruct the sampled frequencies without error, you need to sample at at least 6800 samples per second. Someone decided to round that up to 8000 samples per second, using 8-bit sample resolution, so a single voice channel would require 64000 bits per second - i.e. not 65536 samples per second. As a result, telephony carrier circuits are specified in kbit/s, where the kilo is 1000, not 1024. Since the SI prefixes are well defined, it means that kilo- mega- giga- and tera- in telecommunications are powers of 10, not powers of 2.
      This means that applications that measure throughput on a telecommunications circuit in kibi-, mebi, gibi- and tebi- bits/second are not helping, as actual circuit capacities don't fall naturally on those boundaries. It would be nice if applications allowed one to display either, as an option, to give the end user the opportunity of choosing the option that works best for them in the context of the throughput they are measuring.

      Telecommunications circuit capacities also have irritating effects - like the differences between 10 Gbit/s Ethernet LAN-PHY and 10 Gbit/s Ethernet WAN-PHY [wikipedia.org].

      (I've not linked to lots of references, partly due to my own laziness, and partly because searching on the Internet for a few key terms will give you the background, where you can learn about things like robbed-bit signalling, and a-law and mu-law companding. Wide-area telecommunications has a long and rich history, and it's own standards and vocabulary.)