I hate the new MiB, GiB etc "binary" prefixes

[baja_yu]

They are profoundly stupid. Invetned by some idiots because of even bigger marketing idiots that misused the standard prefixes. [/rant]

[JuggaloBrotha]

what's an MiB or a GiB?

[baja_yu]

Because the very wise people in marketing, mostly hardware/memory manufacturers thought it would be cool to misrepresent the capacity by using 1KB = 1000 B method. So when they say your hard disk is 320 GB they actually mean it's 320.000.000.000 B... and when you put it in your PC you see that you actually have less than 320GB. So some "smart" shoppers mostly in US thought they were scammed, and as they love suing everyone for everything they did so with this. And some even smarter people, instead of forcing the manufacturers to use real binary representation, they thought of inveting "binary" prefixes.

So when you want to say 1 binary gigabyte you are supposed to write 1GiB... They call it Gibibyte. And MiB, KiB etc.

So now 1GiB is 1024 MiB... but 1GB is 1000 MB... So who will now use the decimal prefixes when it comes to computer memory? only those memory manufacturers and noone else?! What's the point then???

People should be smart enough to know that when the standard quantifiers are used (mega, giga, kilo etc) in context of memory, they mean 1024 and not 1000. But it seems people are not smart. The civilization is getting dumber by the hour.

[JuggaloBrotha]

But 1 KB is 2^10 which is 1024, which means 1 KiB must be 10^3 which is 1000

Wait, now I'm confused because Wikipedia says 1 KB is 2^10 and 1 KiB is 2^10 and 2^10 is 1024 so w-t-f is the difference between the two??

Edit: I'll just stick with what's been established since long before I was born, where 1 KB = 1024 Bytes.

[FireXtol]

But 1 KB is 2^10 which is 1024, which means 1 KiB must be 10^3 which is 1000

Wait, now I'm confused because Wikipedia says 1 KB is 2^10 and 1 KiB is 2^10 and 2^10 is 1024 so w-t-f is the difference between the two??

Edit: I'll just stick with what's been established since long before I was born, where 1 KB = 1024 Bytes.

The Standard SI, and Ratio SI/Binary columns.

[wossname]

Yeah but which is it, "nybbles" or "nibbles"?

[mendhak]

The Standard SI, and Ratio SI/Binary columns.

What is the relevance of that ratio column?

[NickThissen]

People should be smart enough to know that when the standard quantifiers are used (mega, giga, kilo etc) in context of memory, they mean 1024 and not 1000. But it seems people are not smart. The civilization is getting dumber by the hour.

I don't agree with this. The terms mega, giga, kilo, etc, are used in nearly all SI measurement units, and they all mean exactly the same thing. Whether you are talking about a kilogram or a kilometer, it means the same thing: 1000 grams or 1000 meters.

Why change this 'in the context of memory'? Why should a kilometer be 1000 meter, and a kilogram be 1000 grams, and a kilowatt be 1000 watt, but a kilobyte be 1024 bytes? That is simply inconsistent, and it has nothing to do with being smart enough to realize the difference or not. Things like this should be consistent, and having exceptions is only likely to cause more trouble, not less.

It is simply inconvenient that one kilobyte is 1024 bytes, and it is even more inconvenient that this has grown into common usage. Inventing new prefixes for 1024 instead of 1000 seems the logical thing to do to me. The problem is that it is difficult to change something this common, so we should have done it much earlier.

Yes, I realize it is also inconvenient to use the usual system for memory sizes, because memory is stored as binary numbers instead of decimal numbers. But that doesn't mean we need to use the same prefixes and then simply make them mean something else...


At least now, when you read KiB, you know it means 1024 bytes, and there is no possible way it could mean 1000 bytes. If you see 1 kB (or even KB which is technically wrong), there might be confusion because in SI units it would mean 1000 bytes, while in 'common usage' it would mean 1024 bytes. In most cases, common usage will win, but in some cases it won't, and it's this ambiguity that should be avoided. If only we adopted this convention (KiB) earlier, we would also know for sure that kB stood for 1000 bytes and never for 1024 bytes. Unfortunately, due to common usage, you always have to ask yourself whether kB means 1000 bytes or 1024 bytes...

What is the relevance of that ratio column?

Since we are talking about powers of 2 vs powers of 10, the ratio between them is not constant. The ratio between 1 KiB and 1 kB (1000 bytes) is nearly 1 (so they are nearly equal), while the ratio between 1 YiB and 1 YB is only about 0.8, which means there is a 20% difference between them. This 20% is "only" about 2E²³ bytes... I can't imagine how many harddisks you could fill with that amount of data.

[baja_yu]

While I agree with you that consistency is good, it's not always imperative to have it. And this is a perfect example of it.

I agree that having KiB mean 1024 byest and KB mean 1000 bytes removes confusion, but should there be any confusion? Where do you need to use decimal prefixes, where would you use them? Lets say we adopt the binary prefixes completely, where would you use the decimal (SI) prefixes now? In that world, binary is all you have, all you need and all you use. It is perfectly acceptable to reuse nomenclature, if there isn't a possibility of overlap and confusion. Maths, chemistry, physics do this all the time. And it can be done here, there is no confusion except the one that the marketing people created. It would all be fixed, fine and dandy if you forbid them from misrepresenting capacities as they do, and all confusion will be eliminated.

[mendhak]

But why is it there on that page? Anyone can calculate a ratio and figure this gap out. It's basic maths.

Wikipedia is making me upset.

[JuggaloBrotha]

I don't agree with this. The terms mega, giga, kilo, etc, are used in nearly all SI measurement units, and they all mean exactly the same thing. Whether you are talking about a kilogram or a kilometer, it means the same thing: 1000 grams or 1000 meters.

Why change this 'in the context of memory'? Why should a kilometer be 1000 meter, and a kilogram be 1000 grams, and a kilowatt be 1000 watt, but a kilobyte be 1024 bytes? That is simply inconsistent, and it has nothing to do with being smart enough to realize the difference or not. Things like this should be consistent, and having exceptions is only likely to cause more trouble, not less.

It is simply inconvenient that one kilobyte is 1024 bytes, and it is even more inconvenient that this has grown into common usage. Inventing new prefixes for 1024 instead of 1000 seems the logical thing to do to me. The problem is that it is difficult to change something this common, so we should have done it much earlier.

Yes, I realize it is also inconvenient to use the usual system for memory sizes, because memory is stored as binary numbers instead of decimal numbers. But that doesn't mean we need to use the same prefixes and then simply make them mean something else...


At least now, when you read KiB, you know it means 1024 bytes, and there is no possible way it could mean 1000 bytes. If you see 1 kB (or even KB which is technically wrong), there might be confusion because in SI units it would mean 1000 bytes, while in 'common usage' it would mean 1024 bytes. In most cases, common usage will win, but in some cases it won't, and it's this ambiguity that should be avoided. If only we adopted this convention (KiB) earlier, we would also know for sure that kB stood for 1000 bytes and never for 1024 bytes. Unfortunately, due to common usage, you always have to ask yourself whether kB means 1000 bytes or 1024 bytes...



Since we are talking about powers of 2 vs powers of 10, the ratio between them is not constant. The ratio between 1 KiB and 1 kB (1000 bytes) is nearly 1 (so they are nearly equal), while the ratio between 1 YiB and 1 YB is only about 0.8, which means there is a 20% difference between them. This 20% is "only" about 2E²³ bytes... I can't imagine how many harddisks you could fill with that amount of data.

But a Kilobit (kb) = 1000 bits and a kilobyte (KB) is 1024 bytes, it has to do with there being 8 bits in a byte which is mathematically logical, it just doesn't fit the metric scale. Now we've got a kilibit being 1020 bytes and kilobyte being 1000 bytes, I dont see a need for kilobit when we've already got kilobit.

I remember people being confused between memory/hdd and network speeds (dialup days) because of the B vs b. Networking uses bits per second, usually in kilobits (kb) and memory/hard drive sizes are in bytes. Or so I was taught in college.

Basically it comes down to hard drive manufacturers wanting to sell you an hdd that's capacity is less than what the customer thinks it is, really.

[FireXtol]

While I agree with you that consistency is good, it's not always imperative to have it. And this is a perfect example of it.

I agree that having KiB mean 1024 byest and KB mean 1000 bytes removes confusion, but should there be any confusion? Where do you need to use decimal prefixes, where would you use them? Lets say we adopt the binary prefixes completely, where would you use the decimal (SI) prefixes now? In that world, binary is all you have, all you need and all you use. It is perfectly acceptable to reuse nomenclature, if there isn't a possibility of overlap and confusion. Maths, chemistry, physics do this all the time. And it can be done here, there is no confusion except the one that the marketing people created. It would all be fixed, fine and dandy if you forbid them from misrepresenting capacities as they do, and all confusion will be eliminated.

You've been programming for too long. Relax.

The difference is simple.

2^10 (binary) versus 10^2 (decimal). Each is aligned to a 4 byte boundary, so lets just all be happy.

They're marketed to the average person. The average person likely being more aware with standard SI values(decimal) rather than binary. Most people don't even know what binary is.

Where the average person is likely going to find fault not with the marketing but with the OS which uses 'some non standard(binary) way of using SI prefixes'.

I don't think it really qualifies as misrepresentation. That'd be like going to a grocery store and complaining that because you can express a quantity in binary SI that you're entitled to the 'extra units'. 1 kilogram of flour?! "Excuse me, but that's 10000000000 in binary, you're shorting me 11000 bits of flour!" Good luck with that.

[baja_yu]

That'd be like going to a grocery store and complaining that because you can express a quantity in binary SI that you're entitled to the 'extra units'. 1 kilogram of flour?! "Excuse me, but that's 10000000000 in binary, you're shorting me 11000 bits of flour!" Good luck with that.

Someone doing that would be an idiot. It's well known that the prefix in that case implies that the the quantifier is 1000.
But these are all conventions. Nothing here is written in stone or God given to us. People made all of this up, the letters, what they mean. The world does not exist in euclidan space, we just chose to represent it like that (among other ways). Distance between points (0,0) and (1,1) does not have to be 1, depends on what metric space you use. So I don't know why people feel so strongly about it. Oooh, don't mess with the SI system or you'll go to hell or whatever.

To me it's perfectly understandable to use it in binary context when it's used with memory capacity, as the there is no actual need or use for the decimal context.

[penagate]

New?

[baja_yu]

Well, time is relative (depends on what side of the bathroom door your standing ). New(er) than the old/current ones.

[minitech]

Weren't GiB, MiB, and KiB originally French prefixes that mean the same thing as GB, MB, and KB? Please consider the French, people.

That would explain the Wikipedia article, too...