Linked by Thom Holwerda on Wed 30th Jan 2013 00:38 UTC
Hardware, Embedded Systems Marco Arment: "Everyone should play by the same rules. A proposal: storage capacities referenced or implied in the names or advertisements for personal computers, tablets, and smartphones should not exceed the amount of space available for end-user installation of third-party applications and data, after enough software has been installed to enable all commonly advertised functionality. With today's OSes, iPads could advertise capacities no larger than 12, 28, 60, and 124 GB and the Surface Pros could be named 23 and 83 GB." Wholly agreed. When I buy a box of 100 staples, I expect it to contain ~100 staples - not 50 because the other 50 are holding the box together.
Thread beginning with comment 550900
To view parent comment, click here.
To read all comments associated with this story, please click here.
ndrw
Member since:
2009-06-30

The whole situation is unfortunate but not that important as we (geeks) paint it. I've been using binary prefixes in the past but nowadays I just stick to SI system. IMHO over time the use of "kilo=1024x" will become sort of a slang in niche markets.

There are many reasons for that:

"Kilo" has always meant "1000x". That's the concept 99% of users are familiar with. Yes, 20 years ago most users were geeks and had no problem with binary numbers but now we are a minority.

As you mentioned, "kilo" means "1000x" in all non data storage applications (e.g. networking). That's because all performance figures are derived e.g. from frequency, which has always been decimal (at least above 1 Hz).

Even in data storage applications, memory size is no longer tied to the size of a 2D array (overhead of error correction, framing added by interfaces etc.). So binary units are no longer better suited to measuring memory size.

Finally, most users don't care about it and the difference is small enough even at "tera" scale. We can just stick to a familiar decimal system and forget binary units altogether (which is already happening).

Myself, I have adopted following practices:
- I always use decimal prefixes when providing data.
- I check what the author has assumed when getting such data (unfortunate but unavoidable step anyway).
- If I absolutely have to use binary units (very rare now), I tend to write X*2^N rather than using "kibi" prefixes, which frankly speaking look even more geeky to me than use of kilo=1024x. Since such data are consumed by technical users it works just fine.

Reply Parent Score: 2