EARLIER FEATURES
FEATURES CONTENTS
LATER FEATURES
|
|
|
|
| 7th August 2005 |
I'VE GOT A BIGGER GIGABYTE |
Brian Grainger |
|
How many bytes are there in a gigabyte? How true is what you read on the web? How can you really get me irritated? All three of these questions will be answered in this article and, in passing, you may learn something of the history of computing. A colleague of mine has recently changed his PC. Having got a new SATA disk drive and wanting to install his old ATA (IDE drive) as a spare, he found numerous problems in getting the new system to recognise the old drive and its existing partitions. This necessitated lots of web searching about disk drives, partitioning, tools to perform disk functions and so on. As I was being kept informed of progress I was also doing some web searching and I came across the use of some strange terminology. On a personal web site someone referred to the size of disks in gibibytes, GiB. Now GiB is not something I have heard of in relation to hard disks, (although I have heard about it in another context altogether!), so I got to searching to find something about it. The search was interesting in that it told me what it was all about and it transpired that a lot of people were peddling untruths. Because of their ubiquity, they were now becoming to be believed. I want to give an alternative view so that any future searchers may at least know there is more than one viewpoint. Now there are a number of ways that will guarantee I will get hot under the collar. Rewriting history to support a particular argument is one of them. Denigrating a person, belief, definition, whatever, just because it does not fit in with the denigrator's viewpoint is another. There seems to be elements of both of these in the story of the gibibyte. The best definition I found for the gibibyte was from NIST at: The number or bytes in a gibibyte, according to that reference, was 2 to the power 30 or: Now, this was what I understood a gigabyte to be so I was intrigued to find the gigabyte was defined as 10 to the power 9 or: In this article I am going to use 2-30 to represent the former and 10-9 the latter, with equivalent values for kilobyte (2-10 or 10-3) and megabyte (2-20 or 10-6). The gibibyte definition had been made by an organisation called the International Electrotechnical Commission (IEC) and their information about gibibytes can be found at: Now I work with electrical engineers all the time and I have never heard of the IEC before. I have heard of the IEE and the IEEE, but not the IEC. Since it is based in Switzerland perhaps it is just the Europeans trying to redefine everything again! A foot is - well the length of my foot - and an inch the length from the tip of my index finger to the first joint. Why do we need centimetres and metres, which I cannot measure at all without a ruler! In their justification for the gibibyte the IEC say there is confusion in the various fields, which is true. They then go on to peddle a story about how disk manufacturers have always used the 10-9 version. This is also true, but history suggests that this was not because it was easier for general folk to understand, as implied by the IEC. Disk drives have been available long before the general public had computers in the home. In the early days they were dealing totally with IT professionals and hobbyists, yet they were always trying to fob us off with the 10-3 kilobytes and perhaps 10-6 megabyte. In those days the gigabyte sized disk was just a dream. The story was that the drives WERE measured in 2-10 and 2-20 size but because disk drives at that time inevitably had errors on them it was convenient to market them as 10-3 and 10-6 size and claim the gap between the 2 base and 10 base system allowed for the errors. Convenient for the disk manufacturers but not for the buyers, who were totally confused and being diddled out of the extra bytes. I've just realised why my USB Flash Drive only comes up with 120Mb when it was sold as 128Mb! It was quite clear to those people who dreamt up the drives what a kilobyte was and, like everything in computers, it was a binary value, i.e. 2-10. It may be unfortunate that they chose kilo as their prefix, which did not quite tie in with its use with kilometre, for example, but that was the way it was and it WAS near to 1000. Fast forward to the late nineties and the general public have computers everywhere. Apparently they are confused about the gigabyte. Let's be honest. The general public will be confused about the gigabyte, however it is defined. They don't care. It is just a big disk! It IS important to the professional designers. We need to talk the same language. Along comes the IEC. Instead of trying to educate manufacturers, telecommunications engineers, and others who keep dreaming up different systems that gigabytes are 2-30 they decide to introduce a brand new system to confuse us even more and be damned with what the originators had defined. Stop denigrating IT people. They defined the gigabyte first. There is no reason to say they got it wrong. No reason to change. The 10-9 gigabyte and the 2-30 gibibyte was recommended by the IEC in 1998. Seven years on and it is still not in general use. Some have suggested this is due to the daft sounding names - kibibyte, mibibyte and gibibyte. An alternative suggested on the web was the large gigabyte. I repeat - we don't need new names. We have had them for years; megabytes, kilobytes and gigabytes. In the same way as I don't associate bytes of memory with things I can eat, I do not associate kilobytes with kilometres. No need to change - just a need to educate. In some cases users of the gibibyte even say it is an SI unit. As NIST points out, this is not the case. Not all you read on the web is true. I venture to suggest that more people know what a gigabyte really is than know the definition of a gibibyte, although this may change when all my readers have seen this article! The last nail in the coffin for the gibibyte is that those who now use the GiB definition must spell out what it means - because nobody else does! |
|
|
|
|
|
|
|
|
|
|
