Author Topic: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?  (Read 17684 times)

0 Members and 1 Guest are viewing this topic.

Offline JV1234

  • Newbie
  • Posts: 4
So I can't get use to this Kibi/Mibi/Gibi-byte thing.  Are they catching on, are is it a failed redefinition by IEC / SI?

I've been reading forums (on this forum, from 2013, and a hijacked post here: http://www.tomshardware.com/forum/282404-32-wanted-backup-strategy), and I don't see a good reason for using KiB/MiB/GiB.

And how many people are using KiB when they use to use KB today?
Comments?
« Last Edit: September 13, 2014, 04:20:51 am by JV1234 »
 

Offline miguelvp

  • Super Contributor
  • ***
  • Posts: 5549
  • Country: us
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #1 on: September 13, 2014, 04:26:26 am »
Plenty of comments here

https://www.eevblog.com/forum/chat/kibibyte-and-mebibyte-wtf/

also it showed up on a metric discussion.

At least in the gaming industry we still use kilobytes as 1024 bytes and I don't see any sign of change other than unix kernel messages to the user.
 

Offline marshallh

  • Supporter
  • ****
  • Posts: 1462
  • Country: us
    • retroactive
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #2 on: September 13, 2014, 04:33:14 am »
Overly pedantic types love it.
And those of us with things to get done just move on.
Verilog tips
BGA soldering intro

11:37 <@ktemkin> c4757p: marshall has transcended communications media
11:37 <@ktemkin> He speaks protocols directly.
 

Offline Rerouter

  • Super Contributor
  • ***
  • Posts: 4621
  • Country: au
  • Question Everything... Except This Statement
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #3 on: September 13, 2014, 05:05:23 am »
Why not argue to have file sizes represented in scientific notation? e.g. 2x2^10

for every extra 10 powers it is equivalent to the argued decimal order or magnitude, e.g. x1024,

It then becomes immediately apparent whether or not your file will fit, at least basing off the discussions from the other thread,
 

Offline miguelvp

  • Super Contributor
  • ***
  • Posts: 5549
  • Country: us
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #4 on: September 13, 2014, 05:10:47 am »
Can someone link the JEDEC standard referred to?

JEDEC still defines 1024 as a kilobyte, we are safe :)
 

Offline JV1234

  • Newbie
  • Posts: 4
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #5 on: September 13, 2014, 05:52:03 am »
You can look here http://en.wikipedia.org/wiki/Kibibyte to see a reference to the JEDEC standard.
 
And miguelvp, thanks for the link.  Yeah I read a bunch of those comments. 

After reading all the comments that I have, I just don't understand why we don't just treat K/M/G/... as Base-10 numbers (1000^1/2/3) only if they are not prefixing a computer term such as bits or bytes.  Otherwise K/M/G/... are Base-2 numbers (1024^1/2/3) for computer terms. 
Seems pretty cut and dry, with no confusion. 
It is just when the marketing people go and use Base-10 numbers to try and make their HDDs look bigger, or perhaps when the Telecom industry uses Base-10 numbers to try and make their data rates look bigger (1.5 Gbps, 44kbps, etc) that things get confusing. 
After all, lots of symbols in science and math have multiple meanings based on context, so why can't K/M/G/... ? http://en.wikipedia.org/wiki/Greek_letters_used_in_mathematics,_science,_and_engineering

Wouldn't this be simple, less confusing, and sound better (compared to speaking KiB)?


 

Offline miguelvp

  • Super Contributor
  • ***
  • Posts: 5549
  • Country: us
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #6 on: September 13, 2014, 06:01:52 am »
because 1000 doesn't have a good binary representation so programmers won't use it.

1000 decimal in binary is 00000011 11101000
1024 decimal in binary is 00000100 00000000

1000000 in binary is 00001111 01000010 01000000
1024x1024 (1MB) in binary is 00010000 00000000 00000000

so 1024 is 1 shifted left 10 positions or 2^10
and 1024x1024 is 1024 shifted left 10 positions or 2^20

Leave metric to base ten stuff and leave our base two stuff alone  >:D

And we don't use KiB , we use KB and it means 1024 like the JEDEC standard says.

Edit, after all a byte is not decimal it's 8 bits with 256 values [0-255], makes no sense to treat it as a decimal like the metric table in that wiki page does.

Maybe they are going to force the byte to only go from 0-99 wasting the rest of the storage so it fits neatly in the decimal metric system.

« Last Edit: September 13, 2014, 06:07:51 am by miguelvp »
 

Offline Artlav

  • Frequent Contributor
  • **
  • Posts: 750
  • Country: ru
    • Orbital Designs
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #7 on: September 13, 2014, 09:45:51 am »
Well, it allows the HDD manufacturers to cheat with disk capacities.
Giga- means 1e9, right? So here you go with 500 Gigabytes HDD, and we keep the extra 36 Gb you were supposed to get.

Even SSD makers start doing such shenanigans nowadays, despite playing it fair in the earlier days.

Seriously, it's like with + and - in electricity now.
It would have been great to make the electron's charge positive as it should be, but there is not enough magic smoke left in the world to survive the change, and confusion would last for generations.
Hacking the universe since 2008
Having a life since 2013
 

Offline Tinkerer

  • Frequent Contributor
  • **
  • Posts: 346
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #8 on: September 13, 2014, 03:53:02 pm »
I think we are safe and that kilo, mega, etc will stand. They have been in use fo decades now and they are in no way changing over night.
If any kind of standard like this is going to change, it would need to basically take an instant worldwide revolution on understanding of something, some sort of massive disaster leaving only a few people to make decisions for the future, or 50+ years(or some long time period like that).
 

Offline Richard Crowley

  • Super Contributor
  • ***
  • Posts: 4310
  • Country: us
  • KE7GKP
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #9 on: September 13, 2014, 04:32:47 pm »
It is a pedantic solution to a pedantic "problem".  Most of us don't care and have better things to worry about.
 

Offline hamster_nz

  • Super Contributor
  • ***
  • Posts: 2286
  • Country: nz
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #10 on: September 13, 2014, 06:49:46 pm »
I have problems with this at work. We have various different storage system, and when I create a 2TB disk I end up with somewhere between 2x10^12 and 2^40 (a 10% difference) depending on which storage system and user interface (GUI, CLI, script...) I use.

The only other time it gets annoying for me is with bandwidth - 1 Mb/s is always 1,000,000 bits per second, never 1,048,576 bits per seconds... so if I need to transfer 1Gb over a 1Gb/s link it will take 1.07 seconds.
Gaze not into the abyss, lest you become recognized as an abyss domain expert, and they expect you keep gazing into the damn thing.
 

Offline saturation

  • Super Contributor
  • ***
  • Posts: 4788
  • Country: us
  • Doveryai, no proveryai
    • NIST
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #11 on: September 13, 2014, 08:28:45 pm »
It will catch on but it will take decades as its taught in schools and used for many academic papers.

However, a practical need today is insure you are aware of the quota on your dataplans contract, if you have one.  In the USA, Verizon will charge $0.25 per mebibyte above your plan quota or escalate you to the next tier once over the quota.   

So to insure you don't truly go over your quota be sure you check what exactly your cellphone dataplan really means in terms of gigabyte or gibibytes.

For example, in those 2 terms the difference is 72MiB or $18.50 in surcharges/month.  Or if your contract calls for a next tier clause, then at 1GiB+ 1 byte you pay an extra $10 up to 2 GiB, another $10 when you exceed 2 GiB etc.,

FWIW Verizon does define what megabyte means, which is a mebibyte.


Best Wishes,

 Saturation
 

Offline JV1234

  • Newbie
  • Posts: 4
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #12 on: September 14, 2014, 04:39:15 am »
Good point "saturation".

And I agree it does seem to be a very fussy solution to a problem that didn't really seem to be a problem.

And I still don't understand why they [IEC/SI/Gov.] don't just treat K/M/G/... as Base-10 numbers (1000^1/2/3) only if they are not prefixing a computer term such as bits or bytes.  Otherwise K/M/G/... are Base-2 numbers (1024^1/2/3) for computer terms.
That would cover software, hardware, technical specs / marketing specs... everything.  If it was made law, there would be no confusion and no need for Verizon to "define" their terms.  And no HDD marketing tricks that would result in a law suit.

Seems pretty cut and dry, with no confusion.

Anyway, that is how I currently treat it.. and I :palm: when I run into people not doing this.
 

Offline vk6zgo

  • Super Contributor
  • ***
  • Posts: 5373
  • Country: au
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #13 on: September 14, 2014, 05:06:39 am »
The problem was driven by the obsessive need by IT people to mess with perfectly good definitions .
1024 was "pretty close to" 1000,so why not?--sloppy,but what do you expect from geeks?

It didn't really matter when "everyone knew what was meant".----it just annoyed RF techs & EEs who weren't part of the "in" group.
When the "general public" got their hands on it,the disparity became a problem,& got all the IT guys "knickers in a twist".

I'd like to see the IT people brought to task for their incorrect usage of the term "bandwidth",but that will never happen,the general public love it,as do the "tech savvy" (read "know nothing") Politicians! >:(
 

Offline rs20

  • Super Contributor
  • ***
  • Posts: 2188
  • Country: au
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #14 on: September 14, 2014, 05:19:48 am »
I love all this emotional "awww, people got upset and so they changed it". OK, the metric system is dumb because "awww, people couldn't figure out imperial units, diddums". Use logic, for goodness' sake.

Quoting JEDEC as one's justification for defining kilo- as 1024 a is pretty telling stretch because JEDEC's core focus is memory and flash chips (I can't stress this enough, chips, not HDDs and SSD modules). The chips are obviously addressed and built in powers of 2, so 1024 is so obviously what is meant by kilo- in that context.  There's no ambiguity in that space, so spending an hour at JEDEC's board devloping a resolution use kibi- is obviously a waste of time -- for JEDEC and its members. Where one makes a huge and stupid leap is when they appeal to JEDEC's standard when describing the quota on their 2TB hard drive, or attacking the concept of kibi-. The IEC are far more broadly focussed and relevant to our everyday lives than JEDEC, the unit "has been accepted for use by all major standards organizations, and is part of the International System of Quantities", it's just another option on your plate. Appealing to JEDEC is cherrypicking.

Whenever you use KiB, MiB, etc, you make your reader's life easier because they immediately know what you mean. All you have to do is hit the "i" key, and a cloud of ambiguity vanishes. Of course, if you mean power-of-10 (kB, MB), the same ambiguity still exists (these days), but the hope is that those usages will erode over many years. Change aversion is so cute.
 

Offline miguelvp

  • Super Contributor
  • ***
  • Posts: 5549
  • Country: us
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #15 on: September 14, 2014, 05:24:58 am »
Don't blame IT, our IT department knows what 1KB is and what 1Kbps is and what 1MB means on drives compared to Memory.
IT never messes with things they just work with what it is.

It's always been a marketing ploy.

For me IT is there for a reason, they know more than you think they do.

Edit: And don't get me started with mega pixel where they treat each RGB component separate, so if a display had say 4leds per pixel for some kind of better color rendition they will multiply each pixel times 4 instead of 3.
Meh, it's all marketing, that's why on displays you gotta look at resolution, same with cameras where all the Mega Pixel started when 1 Mega Pixel is really 1 Million pixels divided by 3 instead of 1024x1024 full color.
« Last Edit: September 14, 2014, 05:31:44 am by miguelvp »
 

Offline Fsck

  • Super Contributor
  • ***
  • Posts: 1157
  • Country: ca
  • sleep deprived
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #16 on: September 14, 2014, 05:26:01 am »
Good point "saturation".

And I agree it does seem to be a very fussy solution to a problem that didn't really seem to be a problem.

And I still don't understand why they [IEC/SI/Gov.] don't just treat K/M/G/... as Base-10 numbers (1000^1/2/3) only if they are not prefixing a computer term such as bits or bytes.  Otherwise K/M/G/... are Base-2 numbers (1024^1/2/3) for computer terms.
That would cover software, hardware, technical specs / marketing specs... everything.  If it was made law, there would be no confusion and no need for Verizon to "define" their terms.  And no HDD marketing tricks that would result in a law suit.

Seems pretty cut and dry, with no confusion.

Anyway, that is how I currently treat it.. and I :palm: when I run into people not doing this.

networking processors/ASICs is where your terminology is truly useful for absolute clarity.

your interface has a transfer rate in base 10 and you have memory in base 2. you're going to be using both so might as well state them in no uncertain terms.
"This is a one line proof...if we start sufficiently far to the left."
 

Offline rs20

  • Super Contributor
  • ***
  • Posts: 2188
  • Country: au
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #17 on: September 14, 2014, 05:44:02 am »
Meh, it's all marketing, that's why on displays you gotta look at resolution, same with cameras where all the Mega Pixel started when 1 Mega Pixel is really 1 Million pixels divided by 3 instead of 1024x1024 full color.

You mean 1000x1000 full colour  :)
 

Offline miguelvp

  • Super Contributor
  • ***
  • Posts: 5549
  • Country: us
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #18 on: September 14, 2014, 05:46:58 am »
Meh, it's all marketing, that's why on displays you gotta look at resolution, same with cameras where all the Mega Pixel started when 1 Mega Pixel is really 1 Million pixels divided by 3 instead of 1024x1024 full color.

You mean 1000x1000 full colour  :)

No I don't :)

Edit: but it seems a 640x480 camera is 0.3 Mega pixel so I guess they did stop doing that even they are rounding down 307,200 pixels would be 0.3072 Mega Pixel base 10, or 0.29296875 base 2 using 1024 instead of 1000.



« Last Edit: September 14, 2014, 06:03:35 am by miguelvp »
 

Offline rs20

  • Super Contributor
  • ***
  • Posts: 2188
  • Country: au
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #19 on: September 14, 2014, 06:09:45 am »
You mean 1000x1000 full colour  :)

No I don't :)

Edit: but it seems a 640x480 camera is 0.3 Mega pixel so I guess they did stop doing that even they are rounding down 307,200 pixels would be 0.3072 Mega Pixel base 10, or 0.29296875 base 2 using 1024 instead of 1000.

Have megapixels ever meant 2^20 pixels?
 

Offline miguelvp

  • Super Contributor
  • ***
  • Posts: 5549
  • Country: us
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #20 on: September 14, 2014, 06:27:40 am »
Nope but it should have been in my opinion.

Now 1K on TVs is 1920x1080 and 4K is 4 times that.
How is 2073600 pixels 1K?

Edit: That's actually 1024*2025 it's like IBM times HAL.

Marketing just confuses things, well not really since it doesn't really matter, we can adjust to any new form of counting easily.


« Last Edit: September 14, 2014, 06:30:27 am by miguelvp »
 

Offline rs20

  • Super Contributor
  • ***
  • Posts: 2188
  • Country: au
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #21 on: September 14, 2014, 06:51:54 am »
Nope but it should have been in my opinion.

Now 1K on TVs is 1920x1080 and 4K is 4 times that.
How is 2073600 pixels 1K?

Edit: That's actually 1024*2025 it's like IBM times HAL.

Marketing just confuses things, well not really since it doesn't really matter, we can adjust to any new form of counting easily.

 :o Why would you endorse even more confusion by saying that megapixels should be 2^20 pixels? Pointing out that marketing people make up crap does not justify you making up crap, that's just terribly perverse.
 

Offline miguelvp

  • Super Contributor
  • ***
  • Posts: 5549
  • Country: us
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #22 on: September 14, 2014, 07:00:44 am »
It was a tongue in cheek comment, I really don't mind any of it at all. No matter what I'll adapt.

But I would like my pay to be in 1024 K not those decimal 1000 K, wait it's thousands an millions, hmm they better get into changing those as well

stop using milions billions and trillions, they mean different things in different countries, use Mega Giga and Tera and be consistent!

Yup the units are all weird but somehow we just cope with what ever it is.
 

Offline vk6zgo

  • Super Contributor
  • ***
  • Posts: 5373
  • Country: au
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #23 on: September 14, 2014, 07:40:52 am »
They stuffed up "pixels" too!

Pixels date back to photographic practice,where resolution  was measured with "resolution lines"

These were the smallest detail which the medium could produce.
If a standard "3x4"picture could resolve say,1000 vertical lines & 750 horizontal lines it was said to be able to display  750,000 "picture elements" (hence "pixels"),as the horizontal & vertical resolution of photographic film is equal.

TV with CRTs messed things around a bit,as the vertical resolution (how many horizontal resolution lines it could display) was limited by the number of "scanning lines" making up the picture,whereas the horizontal resolution was limited by the bandwidth (occupied spectrum) of the system used.

Ideally,horizontal & vertical resolution should be equal,but that was not often the case,so "pixels" instead of being nice squares like in film,became rectangles.

Up to this point,pixels were really just a concept,not something real,but with the advent of modern camera & display technology,the individual elements of the display or camera sensor,in fact,became the pixels.so you could,in theory,if you have the time,a good magnifying glass & memory,count the pixels on your display.

So far,so good,but then you have the "tech savvy" halfwits of the popular media referring to a TV as having "so many" horizontal pixels & "some other number" of vertical pixels.

To technically informed Greybeards this is gibberish!

 

Offline JV1234

  • Newbie
  • Posts: 4
Re: Will Kibibytes catch-on or can we turn things back to the JEDEC standard?
« Reply #24 on: September 14, 2014, 09:04:16 pm »
Good point "saturation".

And I agree it does seem to be a very fussy solution to a problem that didn't really seem to be a problem.

And I still don't understand why they [IEC/SI/Gov.] don't just treat K/M/G/... as Base-10 numbers (1000^1/2/3) only if they are not prefixing a computer term such as bits or bytes.  Otherwise K/M/G/... are Base-2 numbers (1024^1/2/3) for computer terms.
That would cover software, hardware, technical specs / marketing specs... everything.  If it was made law, there would be no confusion and no need for Verizon to "define" their terms.  And no HDD marketing tricks that would result in a law suit.

Seems pretty cut and dry, with no confusion.

Anyway, that is how I currently treat it.. and I :palm: when I run into people not doing this.

networking processors/ASICs is where your terminology is truly useful for absolute clarity.

your interface has a transfer rate in base 10 and you have memory in base 2. you're going to be using both so might as well state them in no uncertain terms.

Yeah, I guess some group would have to change to make things consistent (or perhaps we just needed to inform everyone about the inconsistencies).  So either:
1) Use K / M / G ... in a context sensitive way consistently (or at least in an informed inconsistent way)
2) Or introduce a new term Ki / Mi / Gi ...

I would have preferred the 1st option, instead of IEC's 2nd option, as it would have caused less confusion since we were already pretty much doing this.  The general public that were not in the 'know' in either case would have to learn something new. 

Btw, it took a while for me to figure out the bandwidth was not 'supposed' to be reported in base-2 values.  For a while, in my beginning career, I was reporting my bandwidth values in base-2 K/M/G values.  I bet I confused some people...  :rant:

(btw the rant emoticon is pretty funny)
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf