New SI prefixes clear the way for quettabytes of storage The range of prefixes used within the International System of Units (SI) has been expanded with new names covering very large and very small numbers, driven in part by the ballooning requirements of data storage in some sectors such as data science. According to the UK's National Physical Laboratory (NPL), these changes were …
This post has been deleted by its author
This strikes me as really getting a bit silly now, and is pretty much ending up replicating the confusion of the imperial system, with a ludicrous number of prefixes/units to have to try to remember.
Almost nobody is going to be able to remember all of these additional prefixes off the top of their head (and I definitely include exa and beyond in this, along with their equivalent tiny friends), and even if you can somehow remember the names, can you remember what they actually mean, otherwise it's surely just jargon gibberish?
They should have pretty much called it a day once they got to roughly the range between pico and tera (10^-12 – 10^12) or thereabouts, which, future storage devices aside, covers pretty much everything that most people normally might refer to in everyday life, and for these 'exceptional' use cases of very very big and very very small it would be much simpler and less confusing all round just to write them as the respective powers of 10 instead: 3.14×10^27 m (for example) is probably just as, if not more, clear to anyone who might need to refer to it than 3.14 Rm will ever be…
Or yoy could allow prefixing the extremal prefix, so exa- could have been kilo-peta-. That would have taken us easily up to peta-peta (quetta) and left an obvious stop-gap solution for how to go beyond that.
It is, after all, what people did when they coined million-million, or even ten-thousand now that I think of it. (I think Chinese has a distinct name for that, but I don't think Indo-European languages do.)
Except we eventually needed numbers large enough that that wasn't a great solution. A quadrillion is not a number we need very often, but with the frequency with which we use trillion, we are likely to see it cropping up and it does happen now and again. It would get a bit old to call it a thousand million million every time.
Your English trillion are not real trillion. A trillion should be a million x million x million.
It's all started when using 'billion' for something that isn't a million x million.
Those English people... not even able to speak French correctly ^^
== Bring us Dabbsy back! ==
Excuse me, but to proper English-type people, as the great Neddy Seagoon would put it, billion == bi-million == million * million, trillion == million * milion * million and do forth.
The weird ideas that billion == thousand * million came from the Yanks. For unfortunate reasons (i.e. managers and politicians like to use big words) there has been a creeping rot from the 1970s to the 2000s to abuse the language and cause bloody mayhem as the school text books started disagreeing with each other!
To say we do not know what a billion is is a vile canard!
"The weird ideas that billion == thousand * million came from the Yanks."
Nope. The short scale is French. The Americans started using it in the 1700s. The British Government formally adopted it in 1974, making the long scale obsolete there despite cries to the contrary. Strangely, the French changed back to the long scale in the middle of the last century.
Wiki "Long and short scales" for more than you ever wanted to know on the subject.
At one point, we didn't need tera as a prefix either. Who has a terabyte of storage after all when disks were 5 MB slabs, and when you get to the point of terameters different units start being used instead. Still, although any one of us could have done it, nobody talks about their 2*10^12 B drives. The shortened forms are useful for brevity if the measurement becomes common, whereas scientific papers can use alternate notations as they have done and still do. You also don't have to remember the units; by the time you need them very often, you'll come to know them. Few average people in the 1980s would have used giga or tera, hence the famous mispronunciation of giga in Back to the Future which nobody would make today, but now that they're more common, people understand what they mean. If we get to the point where we need quettabytes, we'll learn the prefix. Until then, nothing will require you to know it.
"hence the famous mispronunciation of giga in Back to the Future which nobody would make today"
The root is the same as that of "giant" (from the greek by way of Latin and French), so I'd say the film got it right, the rest of us not so much. The soft G goes way back, cf. "In þat tyme wer here non hauntes Of no men bot of geauntes." —Wace's Chronicle, ca. 1330
See this comment from 2013 for a further discussion of the histories of “gigantic”, “giant”, and “giga-”, both of etymology and pronunciation. Greek didn’t (and doesn’t) have the soft G sounds of either English or French; in modern Greek, γ (g) before ε (e) or ι (i) is pronounced /ʝ/, which is found in some pronunciations of the “y” in English “yeast”, so its closest standard English approximation is /j/, the sound of modern English Y. The modern English Y sound was also one of the sounds of Old English G; this was the first sound in Old English gigant.
"Almost nobody is going to be able to remember all of these additional prefixes off the top of their head"
Almost nobody will ever actually need them (outside school), not in RealLife anyway, so who cares?
Those of us who do need them will remember the bits we need, and know how to look up the bits we only use occasionally on an as-needed basis.
As others have said, if these are not things/scales you regularly talk about then these prefixes simply aren't for you. There is nothing wrong with that. SI prefixes are NOTHING like the mess that is imperial units where nothing is a simple 10^x of something else. These prefixes are mostly intended for hard-core scientist discussions, likely to be used more conversationally where writing things down or saying 3.14 times 10 to the power of 27 is both confusing and time consuming. It's fine if you don't have a use for these prefixes, you can forget all about them, but other people do have a use and will use them. You probably don't ever use Hounsfield Units either, but you'll be glad your radiologist knows about it when you ever need an X-ray or CT scan.
Well, unless there are now 10 bits in a byte, this is not true. According to some quick searching I just did :
A yottabyte is 2 to the 80th power bytes.
The prefix yotta is based on the Greek letter iota. In decimal format, a yottabyte is written as 1,208,925,819,614,629,174,706,176
It would cost $100 trillion to buy a yottabyte of storage and would take approximately 86 trillion years to download a 1 YB file. *
* You might be able to get a yottabyte for about $40 trillion if you bought the cheap stuff on eBay, and the download speed is unspecified, but if you have a good UK fibre broadband connection, it might take only 43 trillion years, although you'd have to reboot your router 8600 trillion times before the download completed, which might get a bit tiring if your work space is upstairs and the router is downstairs, like mine.
It has nothing to do with how many bits are in a byte. You can use decimal or binary multiplication but each will use 8-bit bytes. Meanwhile, if you did have 10-bit bytes, a kilobyte would still be 1000 or 1024 bytes depending on whether you made a disk or RAM. It would store more stuff, and it would have grown from 8 to 10 kilobits, and it would require a lot of hardware and software changes, but those two ratios are independent.
The modification of the prefixes to use 2^10 instead of 10^3 as a factor leads to some interesting names if you use the insertion of "bi" into the second syllable. If you can find a way to pronounce "quebibyte, robibyte, zebibyte, exbibyte, pebibyte, tebibyte, gibibyte, mebibyte, and kibibyte" without sounding crazy, you get a prize.
There were better options.
I'd kinda hoped that they'd extrapolate from the last two, zetta and yotta, and go with (say) whitta, vonta, urga, etc. in reverse alphabetical order. (I suppose you could have, say, xena between yotta and whitta, but I'd be inclined to skip it.) I don't always remember the order of the "higher" prefixes. I may have to come up with some mnemonic, the way I remember the taxonomic divisions, kingdom/phylum/class/order/family/genus/species, with "Kelly, please come over for good sex".
On the bright side, these do maintain a run of descending alphabetical order.
In order to fit the exapartition that would make, you're going to need a new kind of partition table. If we scale it in the same way that GPT does, that will take us from needing 16 KiB for GPT's current 128-partition limit to needing 128 EiB for the partition table assuming we don't want any more. That is unless you're comfortable nesting partitions and getting the OS to accept and perform well with partitions that contain partitions that contain partitions that contain ... eight levels deep.
According to my quick internet search, a few quettabytes of storage would be non-trivial even if you had molecular encoding. We'd hit the AI singularity first, and our new overlords probably won't need the units.
Regardless, "quettascale" should be appearing in job postings and resumes any time now.
Even a quectoGoogol is not within cooee of a quetta :)
How much porn is there? Abstractly probably not that much once you take into account the limitations of anatomy,physiology and laws of physics. Even allowing for the unlimited capacity of human beings for perversion I can imagine a recent bit of AI(ML) kit, a couple of thousand rules and some state of the art computer graphics could give the industry a run for its money. How much of the output would land you in quad is another matter. "Max Headroom does...."
Hate to think what in this line awaits braver souls than mine in the metaverse.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
