# Geeky Computer Question - MB vs Mb

#### AtHomeBoy_2000

##### macrumors 6502a
Original poster
I have no idea how this came into my head this evening, but I have a question. I know in computer terminology there is a differance between MB and Mb. MB meaning megabyte and Mb meaning megabit. Where my question comes in is about my DSL line. According to AT&T my downstream speed is 1.5 to 3.0 Mbps. So, if my understand is correct, to the common user, my max speed is really .375 MBps (since 8 bits make up a byte). So to download a 1MB file, at peak speed, it would take me 2.67 seconds to download the file.

Is this correct? If so, isn't saying to the average customer you have 3Mb download speeds misleading since MOST people associate Mb to MB sicne they would never think there is a differance.

#### ChrisBrightwell

##### macrumors 68020
AtHomeBoy_2000 said:
Is this correct?
More or less. There are a lot of things that go into calculating a "real world" throughput value, but you got the basics.

#### Brize

##### macrumors 6502a
I don't necessarily think it's misleading, but it certainly causes confusion. I regularly read technology articles in national newspapers that refer to '1 megabyte per second' connections, for example.

#### TMA

##### macrumors 6502a
You might find this MacGuides page useful: http://guides.macrumors.com/Bits_And_Bytes

As for your DSL speed. Well that's say it's operating at full 3Mbps, this should give around 384 MBps sync speed (the absolutely maximum download speed) Because 3mb = 3 x 1024 = 3072 then divide by 8.

However because we're talking about DSL, there are other things to take into account such as overheads and acks. Usually a good rule of thumb is divide by 10 instead of 8. So 307.2 k/sec would be a more reasonable achievable download speed.

I'd say with your connection you should expect anything between 153.6 k/sec and 307.2 k/sec download speeds.

#### sushi

##### Moderator emeritus
TMA said:
However because we're talking about DSL, there are other things to take into account such as overheads and acks.
And that is why when I test a connection I always DL the Mac OS 9.2.1 update (82 Megabytes) to see how long it really takes to determine the throughput of the connection. Granted it is not perfect by a long shot, but it is way more accurate than those web based speed meter tests.

BTW, I think that the same holds true for FOIS and FTTH.

#### TMA

##### macrumors 6502a
sushi said:
BTW, I think that the same holds true for FOIS and FTTH.
I always got the impression that the speed of cable was 'cleaner' and more predictable as it's less prone to noise/cross talk. But yes I guess any network will have its overheads.

I also find timing a large download is the best way to calculate real speeds. It's best to find a fairly close and reliable server though, so Apple probably aren't the best choice if you're not in the UK (even though i'm sure they use caching and other technologies) My ISP offer as a 'speed test' several 100mb files that are just full of 1's or 0's http://fuller.zen.co.uk/test/

#### sushi

##### Moderator emeritus
TMA said:
I always got the impression that the speed of cable was 'cleaner' and more predictable as it's less prone to noise/cross talk. But yes I guess any network will have its overheads.

I also find timing a large download is the best way to calculate real speeds. It's best to find a fairly close and reliable server though, so Apple probably aren't the best choice if you're not in the UK (even though i'm sure they use caching and other technologies) My ISP offer as a 'speed test' several 100mb files that are just full of 1's or 0's http://fuller.zen.co.uk/test/
Cable systems generally are loop based so there is overhead associated with moving the info around the loop. Plus you are sharing bandwidth with your neighbor. One of my buddies can be on his connection when the HS let's out and you can see his connection slowing down as they log onto his loop.

Cool that your ISP has a large file to test. The reason that I use Apple is that everyone can access it. Their servers have decent speed. After all, my friend could DL Mac OS 9.2.1 with an average time of 9 seconds.

#### evoluzione

##### macrumors 68020
sushi said:
After all, my friend could DL Mac OS 9.2.1 with an average time of 9 seconds.

damn.

to the OP, yeah i've often tried to clarify it in my head also, wondering if any of the ads have MB/s when they do in fact mean Mb/s. i'm sure 99% of the broadband customers have no idea there's even a difference.

#### CanadaRAM

##### macrumors G5
Modems have always been stated in megabits - a 56K modem is 56,000 megabits.

#### sushi

##### Moderator emeritus
evoluzione said:
i'm sure 99% of the broadband customers have no idea there's even a difference.
I would tend to agree with this.

A few years ago over lunch bunch of us got into a relatively heated discussion about providers and their DL speed. The confusion in part was the B vice b issue.

I settle things by asking how long it would take everyone to DL a 10 Megabyte file. As everyone around the room gave a rought estimation it became apparent to the others that there was a difference.

The other thing that I find interesting is the amount of variance in the ADSL DL speeds. Just because an ADSL line is advertised as a 24Mbps line does not mean customer will see this type of speed. Sometimes I've seen a 3Mbps line provide the same throughput as a 12 or 24 line.

Lot's of marketing hype is out there for sure. And of course variance due to installation and distance from the switch.

#### TMA

##### macrumors 6502a
Its getting particularly confusing in the UK at the moment I believe.

Some ISP's are offering 512k/1mb/2mb as a guaranteed connection speed, depending on how far away from the exchange you are and line noise. Other ISP's are offering 'up to' 8mb (it adapts to the best your line can handle and stays at that rate) and in most cities/highly populated areas there are the LLU providers that offer 'up to' 24mb (again, highly dependent on being reasonably close to your exchange and on a good line).

They are the choices for SYNC speed that decides what your modem will connect at.

Actual speeds are then governed by contention - If there are many people on your exchange all trying to use their connections your speed can drop significantly. As far as I know this applies to all ADSL connections in the country, with some people being lucky and on a quiet exchange (never suffers from contention) and others regularly experiencing slow downs at peak times.

As if that wasn't enough, there is then a question of prioritisation. 'Home' DSL products, at roughly half the monthly line rental, are put second behind 'office' products.

FINALLY you've got to try and factor in your ISP's own core network. Many budget ISP's become overloaded and cannot keep up with their customers demands for more bandwidth. This is happening more and more as faster sync speeds become available. It is now at the point that it is rare to find a completely unlimited service for ADSL above 2mb. Most of us on the faster connections are sadly limited to 20gb/50gb monthly transfer limits.

Then there is cable...

I have a hard enough time keeping up with it all.

#### MarkCollette

##### macrumors 68000
When I took a Communications course in University for my Computer Science degree, I asked the prof why everything was measured in bits for data transmission, since people only really care about bytes. There's several reasons, some of it being that it's just historically how things are done. Also, there's the fact that physical transmission on a wire or over the air always comes down to sine wave properties involved in sending a single bit. But probably the biggest reason for why it's more correct to talk about bits, is that many low level protocols add bits in to a data stream, for error correction/detection, frame separation, or for special codes that change modes in the protocol. So, if a medium can transfer at 512 bits/second, you won't necessarily get 64 bytes/second, because some protocols use up some of the bits for themselves. Think of telephone modems, and that setting N81, which means No parity bit, 8 data bits, 1 stop bit. So, you have to send 9 bits to get your byte across.

#### mkrishnan

##### Moderator emeritus
ChrisBrightwell said:
So CanadaRAM made his first typo ever.

Welcome to the club, Trevor. We'll be sending your 2 THz iMac Core Duo shortly.