Sdio

[rabidpoobear]

Remember that a conversion from megabits to megabytes by dividing by 8 only works if you're talking about network speeds, storage wise you'd probably want to talk about mebibytes instead. 54 megabits = 54,000,000 bits per second.

1) Sorry, but dividing bits by 8 to get bytes works EVERYWHERE. At least, since all non-embedded systems standardised on the 8-bit byte many moons ago. The whole SI unit debate *only* applies from byte level upwards.

I didn't say bytes weren't 8 bits, I said kilobytes are 1000 bytes in some situations and 1024 bytes in other situations. Same with mega, giga, tera. It starts to make a significant difference at a certain point.

 

[pelrun]

Rabidpoobear, you got a whole bunch of unit conversions and conventions mixed up in the one sentence, which is an impressive feat... and Tor, you just repeated my second point in different words (network/serial protocol overhead). Well done!

...heh, I think I see what's going on. Interesting that the wifi manufacturers are pulling the same trick the hard driver makers are; I'd never really considered it before.

 

[Kramy]

...heh, I think I see what's going on. Interesting that the wifi manufacturers are pulling the same trick the hard driver makers are; I'd never really considered it before.

What trick? Using an analog measurement for an analog medium? Yeah, some trick. :lol:

Go after the SSD manufacturers. At least they're lying rather than choosing the favourable (yet correct) measurement. Nomatter how I look at it a 32GB SSD does not turn into 27.9GB.

P.S. Most of your space is lost to filesystem overhead.

 

[pelrun]

Pfft to 'correct'. It's only convention. Computer science had a iron-clad convention of power-of-2 units, which was used universally and never caused any confusion. That it conflicted with the SI definition doesn't make it incorrect, because it was clear to everyone what the convention was and where it applied and where it didn't.

The attempt to drag the SI usage in was deliberately self serving and has brought only confusion and deception. That can't possibly be a worthwhile definition of 'correct'.

 

[Kramy]

It's correct because that's how they've done it from the start. (HDD manufacturers, I mean - and by "start", I mean recent decades, where sizes were approaching and exceeding 1GB - if you go back farther I have no clue what anyone was doing)

Computer science had a iron-clad convention of power-of-2 units, which was used universally and never caused any confusion.

Proof? From what I remember, everyone was doing it differently.

Bits or bytes - rounding to hundreds or thousands. We buy 65k colour LCDs and 16m colour LCDs. Those are 16bit and 24bit. 4mbit cartidges and 64K of ram are power-of-2 (but K of what? bits or bytes?) My point - everyone was doing it different. If you look back, there really was no standard, except that we eventually dropped into using SI prefixes for non-SI amounts. We shouldn't have, but we did.

Hey, I have a question - if everything is power-of-2 units, why are CPU speeds measured in decimal? Surely they should be measured the same way? They're just as digital as HDDs. I think AMD and Intel must be trying to screw us! Conniving thugs! I want my 258 extra mhz!


Near as I can tell, the only thing ever measured in power-of-2 units was RAM, and technologies similar to RAM. Everything else has always been decimal - mislabelled or not. When people slam HDD manufacturers, they also never consider the vast amounts of disk space you get for free that go to ECC. We're certainly not being maliciously short changed.

What's really sad though is this could all be mitigated by getting Microsoft and RAM manufacturers to replace MB, GB, TB, with MiB, GiB, and TiB. It's the easiest solution available.

never caused any confusion.

Lawsuits beg otherwise.


What's so silly about this is, Microsoft is at the heart of all these problems - but they're not getting sued. They took SI prefixes and used them incorrectly, resulting in HDD manufacturers getting sued - but not them. They've artificially limited the available RAM for their operating systems, but nobody complains. A pre-servicepack copy of XP with one of the boards it supported can make use of a full 8GB of RAM in 32bit Windows - but now most people only get ~3.25GB out of 4GB. Despite having the option of re-enabling this with Vista, they decided not to - and then they changed the usable RAM reported by System Properties to just your total RAM, adding even more confusion. Many people bought 32bit systems reporting 8GB of RAM. <_< Utterly diabolical.

Microsoft likes deception. HDD manufacturers are just going with the more favourable SI prefix. (which is technically correct - after all, HDDs are an analog medium)


Edit: Oh, and Intel! I've met people that were fooled into thinking they had a quad-core because of 4 "cores" showing up in the Task Manager. That's a far bigger scam than a little bit of HDD space, don't you think?

What I'm trying to get at is... there's a lot of worse companies out there. HDD manufacturers don't deserve the hate - and I believe they're in the right.

 

[pelrun]

The power-of-2 convention isn't for 'digital' values; I don't know where you got that from. It was used specifically for bits/bytes, and was due to how data buses are physically implemented. That's it. Every other unit is power-of-10.

Microsoft didn't make the convention, btw. And the rest of your post just doesn't apply to the subject.

 

[Kramy]

The power-of-2 convention isn't for 'digital' values; I don't know where you got that from. It was used specifically for bits/bytes, and was due to how data buses are physically implemented. That's it. Every other unit is power-of-10.

Data busses? You mean like SATA and PCIe?

 

[paulguy]

It is kinda confusing to have 1024 bytes in a kilobyte of RAM, but 1000 bytes in a kilobyte of hard disk space. Kinda silly to make it different just for magnetic storage media, when every other storage device seems to use 1024 bytes/kb, like RAM, ROM and I think even a 700MB blank CD is literally ~700MB...generally a little more, even. (1024 bytes/kb).

Can't say there isn't foul play when one group of people decides to do it differently than everyone else when it obviously suits them.

 

[Kramy]

It is kinda confusing to have 1024 bytes in a kilobyte of RAM, but 1000 bytes in a kilobyte of hard disk space.

Oh, I agree. That's where this entire debate came from. We could just call it 1GiB of RAM though. The kernel could work with 4KiB pages. Rather than having two+ meanings for each prefix, we could knock it down to one. And since the SI prefixes are unlikely to disappear, why not do it now?

Kinda silly to make it different just for magnetic storage media, when every other storage device seems to use 1024 bytes/kb, like RAM, ROM and I think even a 700MB blank CD is literally ~700MB...generally a little more, even. (1024 bytes/kb).

That could be reversed into saying it's silly doing it when it's just NAND flash or ROM using 1024 bytes/KB. RAM excluded, since that's not intended for long term storage - a CD has the same predicament as HDDs, but their solution was to sell you less space. You can burn 2048 byte blocks, or ~2350 byte blocks. Burning the latter literally increases the amount of data you can store.

http://www.osta.org/...ology/cdqa7.htm

A "700MB" CD probably has ~780,000,000 bytes available - 780MB, or ~744MiB.

You need to change burning modes, and possibly enable overburn. Then you can stuff more data on the discs.

Can't say there isn't foul play when one group of people decides to do it differently than everyone else when it obviously suits them.

Totally agree. That's one area where I'm annoyed with Intel and AMD. You'd think they'd be able to standardize max TDP and power consumption measurements.

 

[bzar]

I think the power-of-two units should be used in address space sizes (where it is logical) and anything related to them via sharing those units. Memory sizes, storage capacities and by extension transfer rates should all work this way, so a kilobyte is always a kilobyte (or kibibyte as it is sometimes called) or 1024 bytes. Processor clock speed on the other hand has no relation to address space size, and thus a kilohertz is 1000 Hz. Just my 10 cents.

 

[urjaman]

I think the power-of-two units should be used in address space sizes (where it is logical) and anything related to them via sharing those units. Memory sizes, storage capacities and by extension transfer rates should all work this way, so a kilobyte is always a kilobyte (or kibibyte as it is sometimes called) or 1024 bytes. Processor clock speed on the other hand has no relation to address space size, and thus a kilohertz is 1000 Hz. Just my 10 cents.

Exactly.

Note: Clock signal's hz (switching speed) is in general an analog value. That is a "1600Mhz" processor will propably not be running at exactly 1600.000000Mhz, but maybe something like 1600.223Mhz (or 1599.83455... Mhz) and can vary a little.

 

[Tor]

It is kinda confusing to have 1024 bytes in a kilobyte of RAM, but 1000 bytes in a kilobyte of hard disk space. Kinda silly to make it different just for magnetic storage media[..]

There was a time when hard disk space was indeed measured in "proper" 2^n sizes, i.e. 1GB=1024MB=1024*1024KB=1024*1024*1024B. But then one day a vendor (I don't remember which one anymore, but it was one of the major ones) changed to decimal, simply to make their disks look larger compared to the competitors. At that point the competitors were forced to follow suit.

 

[Kramy]

It is kinda confusing to have 1024 bytes in a kilobyte of RAM, but 1000 bytes in a kilobyte of hard disk space. Kinda silly to make it different just for magnetic storage media[..]

There was a time when hard disk space was indeed measured in "proper" 2^n sizes, i.e. 1GB=1024MB=1024*1024KB=1024*1024*1024B. But then one day a vendor (I don't remember which one anymore, but it was one of the major ones) changed to decimal, simply to make their disks look larger compared to the competitors. At that point the competitors were forced to follow suit.

Was it IBM?

I looked into it a while back. The major vendors have been using SI capacities since the 80's.

But my magic hourglass fails to go further back in time, or account for all the folded companies.

 

[kingoddball]

OSX 10.6 is now using base 10 for hard drive space. Which is also what hard drive manufacturers use to calculate the space on their storage devices.


Windows/Linux still use base 16. Pretty sure thats what the numbering is.

Edit: Windows uses base-2. Not 16!!

 

[Kramy]

Base 16 is hex. (0-9, A-F)

Both measurement systems are decimal. One is just SI, and the other... isn't. I don't know a definitively correct term to distinguish between them - but as long as everyone understands, it doesn't really matter.

 

[kingoddball]

Sorry. I was wrong. Windows uses base-2.
Sorry! Can't remember random numbers like that!

OSX Base-10

Windows - Base-2

 

[peca]

Sorry. I was wrong. Windows uses base-2.
Sorry! Can't remember random numbers like that!

There is nothing about random numbers. It is about 10 groups of people: those who count in binary and those who not.

 

[rabidpoobear]

Rabidpoobear, you got a whole bunch of unit conversions and conventions mixed up in the one sentence, which is an impressive feat... and Tor, you just repeated my second point in different words (network/serial protocol overhead). Well done!

Hey congratulations on being offensive.

My point was that if you have
54 megabits
and you convert it to megabytes
you will have
54 / 8 megabytes

but you still have to convert it between SI megabytes and powers-of-2 megabytes if you want to compare it to a power-of-2 megabyte measurement.
That is,
54 mb in SI is 54,000,000 bits -> 6.75 MB

but if you want to compare it to a measurement in powers-of-2, then
54 mb in power-of-2 is 56,623,104 bits -> 6.75 MiB or -> 7.08 MB.

If that's wrong then tell me why it's wrong, don't just say I messed up somewhere and I'm an idiot. I obviously don't think I'm incorrect and you haven't specified exactly where my logic is flawed; if you do so I'm happy to concede the point. You don't have to be a jerk about it, just be clear what you mean and let me be the one to say "oops I'm an idiot." Then you get the satisfaction of being right, and on top of that, the satisfaction that I know you're right, without being so abrasive.

 

[Kramy]

but if you want to compare it to a measurement in powers-of-2, then
54 mb in power-of-2 is 56,623,104 bits -> 6.75 MiB or -> 7.08 MB.

If that's wrong then tell me why it's wrong, don't just say I messed up somewhere and I'm an idiot. I obviously don't think I'm incorrect and you haven't specified exactly where my logic is flawed; if you do so I'm happy to concede the point. You don't have to be a jerk about it, just be clear what you mean and let me be the one to say "oops I'm an idiot." Then you get the satisfaction of being right, and on top of that, the satisfaction that I know you're right, without being so abrasive.

54,000,000 bits would become 54 SI megabits, or 6.75 SI megabytes. (8 bits in a byte)

To get MiB you need to divide by 1024 a few times.

54000000/(1024*1024*8) is only 6.437 MiB.

But that doesn't factor in protocol overhead. Apparently 300mbit Wireless N is almost as fast as 100mbit wired. Well, unless you have crazy cord lengths. (would that be full duplex, though? Hmm... )

 

[darionapole]

http://en.wikipedia....or_abbreviation This explanation is extracted from here

From Wikipedia
Unit multiples

There has been considerable confusion about the meanings of SI (or metric) prefixes used with the unit byte, especially concerning prefixes such as kilo (k or K) and mega (M) as shown in the chart Prefixes for bit and byte. Since computer memory is designed with binary logic, multiples are expressed in powers of 2, rather than 10. The software and computer industries often use binary estimates of the SI-prefixed quantities, while producers of computer storage devices prefer the SI values. This is the reason for specifying computer hard drive capacities of, say, 100 GB, when it contains 93 GiB of storage space.

While the numerical difference between the decimal and binary interpretations is small for the prefixes kilo and mega, it grows to over 20% for prefix yotta, illustrated in the linear-log graph (see right) of difference versus storage size.

54Mb ( look at the lowercased b ) means: 54 (fifty four) Millon Bits, thats 54.000.000 Bits (01010101010.......)

divide that by 8 (the value in bits) and you got it in Bytes (eight bits == 1 byte)

54000000 / 8 = 6750000 Bytes

6750000 / 1024 (for kibi) / 1024 (for mibi) = 6,437301636 MiB (this reads: "Six and a Half Millon Binary Bytes(mibi)" )

MB (a Millon Bytes) is wrong

54 / 8 = 6,75 WRONG

(54*1000*1000)/1024/1024 = 6,437.... CORRECT

to summarize
1000000 is one millon > M > this for bits
1048576 is one binary millon > Mi > this for bytes