Jump to content

"Bit Measuring Contest" - Odell


Recommended Posts

(I thought I would segregate this wildly off-topic, yet informative, debate from the discussion Current State of Sprint. Thanks to Odell for the inspiration of the thread title. - Duffman)

 

Actually, to further clarify, bits are the basic unit of measurement for data. They are the 1s and 0s that build the data we see. A byte, on the other hand, refers to the amount of bits required to encode a single character of text. Currently, this is generally accepted as 8 bits, but it has varied in the past.

This is why bytes are more suitable for referring to storage capacity, where bits are more suitable for referring to digital information transfer. Your internet connection will only deliver so many 1s and 0s, no matter how your computer uses them to put together the transmitted data.

 

Sent from my Galaxy Nexus using Tapatalk 2

  • Like 1
Link to comment
Share on other sites

Actually, to further clarify, bits are the basic unit of measurement for data. They are the 1s and 0s that build the data we see. A byte, on the other hand, refers to the amount of bits required to encode a single character of text. Currently, this is generally accepted as 8 bits, but it has varied in the past.

This is why bytes are more suitable for referring to storage capacity, where bits are more suitable for referring to digital information transfer. Your internet connection will only deliver so many 1s and 0s, no matter how your computer uses them to put together the transmitted data.

 

Sent from my Galaxy Nexus using Tapatalk 2

 

Warning: Nerd content ahead!

 

8-bits do make up a byte but that does not mean it is the storage space for a single character. Unicode separated the concept of "code points" (the visual symbols we think of as letters, numbers, etc) from the encoding (the format used to store them). These days a lot of stuff is UTF-8 which uses 1 byte for US English characters and from 2-4 bytes for others. UTF-16 uses 2 bytes for most code points in the most common languages with the benefit that it typically doesn't vary so you can jump forward in a string without having to examine all the bytes in between.

 

 

Not to mention for network transfers, you have overhead that makes your effective transfer speeds lower because not all the bits are used for your content.

 

 

Anyway for historical reasons network transfer speeds are often quoted in bits per second, so 1Mbps is 1024*1024 bits per second. Divide by 8 to get the MB/s. Oh yeah I forgot to mention that bits/bytes are usually measured in multiples of 1024 because in binary 1024 is a direct power of 2 (equivalent to 1000 being a direct power of 10, which is why we insider it to be a "round" number.... Similarly computer systems often stick to "round" numbers only in base 2).

Link to comment
Share on other sites

a6fa6572-c2a2-7049.jpg

 

Megabits per second

 

Sent from my Galaxy Nexus using Tapatalk 2

You on 3G, 4G, or WiFi

Link to comment
Share on other sites

Warning: Nerd content ahead!

 

 

Not to mention for network transfers, you have overhead that makes your effective transfer speeds lower because not all the bits are used for your content.

 

 

Anyway for historical reasons network transfer speeds are often quoted in bits per second, so 1Mbps is 1024*1024 bits per second. Divide by 8 to get the MB/s. Oh yeah I forgot to mention that bits/bytes are usually measured in multiples of 1024 because in binary 1024 is a direct power of 2 (equivalent to 1000 being a direct power of 10, which is why we insider it to be a "round" number.... Similarly computer systems often stick to "round" numbers only in base 2).

 

1Mbps is not equal to 2^20 bits/sec. They use powers of two in reference to memory and cpu addressing. Network speeds and storage (i.e. Hard drives and such) use powers of 10. See table on top right here http://en.wikipedia.org/wiki/Binary_prefix.

Link to comment
Share on other sites

 

 

1Mbps is not equal to 2^20 bits/sec. They use powers of two in reference to memory and cpu addressing. Network speeds and storage (i.e. Hard drives and such) use powers of 10. See table on top right here http://en.wikipedia.org/wiki/Binary_prefix.

 

No, only hard drives use powers of 10. Network transfer speeds are definitely powers of 2.

Link to comment
Share on other sites

Hard drives used to use powers of 2 years ago. Then they changed and most operating systems didn't follow suit until Mac OSX Show Leopard in 2009. It was weird when I discovered that was how Apple saved hard drive space.

 

Edit: Correct me if I'm wrong, but I don't even thing Windows changed to power of 10 for hard drive sizing in the operating system yet.

Link to comment
Share on other sites

Network transfer speeds are definitely powers of 2.

 

Nope, they are in powers of 10.

 

http://searchstorage.techtarget.com/definition/Kilo-mega-giga-tera-peta-and-all-that

 

Power-of-10 multipliers are also used to define binary data speeds. Thus, for example, 1 kbps (one kilobit per second) is equal to 103, or 1,000, bps (bits per second); 1 Mbps (one megabit per second) is equal to 106, or 1,000,000, bps. (The lowercase k is the technically correct symbol for kilo- when it represents 103, although the uppercase K is often used instead.)

 

Original link I gave says the same thing.

Link to comment
Share on other sites

Nope, they are in powers of 10.

 

http://searchstorage...ta-and-all-that

 

 

 

Original link I gave says the same thing.

 

Don't trust Wikipedia with everything. Network links are quoted in binary. If in doubt on Speedtest and other apps, change their reporting from kbps to mbps. You'll see that all of them figure 1024 kbps = 1 mbps. The only place of significance that uses the 1000 bits / kilobit (power of 10) is hard drives so they can sell you an inflated size.

Link to comment
Share on other sites

Don't trust Wikipedia with everything. Network links are quoted in binary. If in doubt on Speedtest and other apps, change their reporting from kbps to mbps. You'll see that all of them figure 1024 kbps = 1 mbps. The only place of significance that uses the 1000 bits / kilobit (power of 10) is hard drives so they can sell you an inflated size.

 

I like my inflated size hard drives, lol. At least my format size shows like what the box says. :lol:

Link to comment
Share on other sites

No, only hard drives use powers of 10. Network transfer speeds are definitely powers of 2.

 

Would it make you feel better if both of you are quasi correct?

 

Binary bit metrics follow base 2 logarithms. The base is 2, and the exponent is a multiple of 10 (e.g. kilo = 2^10, mega = 2^20, giga = 2^30, tera = 2^40).

 

AJ

Link to comment
Share on other sites

Don't trust Wikipedia with everything. Network links are quoted in binary. If in doubt on Speedtest and other apps, change their reporting from kbps to mbps. You'll see that all of them figure 1024 kbps = 1 mbps. The only place of significance that uses the 1000 bits / kilobit (power of 10) is hard drives so they can sell you an inflated size.

 

I posted 2 links saying the exact same thing. It is a standard.

 

Just did a speed test at Speedtest.net and it says in kbits down 30408. I have a 30Mbps line 30408/1000=30.408 Mbps. Changed it back to Mbps on their site and it says 30.41 (rounded off). So obviously they are dividing by 1000.

 

 

Would it make you feel better if both of you are quasi correct?

 

Binary bit metrics follow base 2 logarithms. The base is 2, and the exponent is a multiple of 10 (e.g. kilo = 2^10, mega = 2^20, giga = 2^30, tera = 2^40).

 

AJ

 

Not all bit metrics are exponents of 10: nybbles, words, double words, and even paragraphs. Bits and bytes are the more commonly referred to measurements.

Link to comment
Share on other sites

I posted 2 links saying the exact same thing. It is a standard.

 

Just did a speed test at Speedtest.net and it says in kbits down 30408. I have a 30Mbps line 30408/1000=30.408 Mbps. Changed it back to Mbps on their site and it says 30.41 (rounded off). So obviously they are dividing by 1000.

 

 

Check the weblinked png drawing-- you'll find it doesn't match what the screen shows-- app bug in some versions. Network rates are always given in 2^10 increments rather than the 1000 between.

Link to comment
Share on other sites

1911487907.png

 

 

 

 

 

What bug? Matches up just fine. I checked it in my history.

 

30989/1000=30.989. They just rounded to 2 decimal places for Mbps.

 

And per their site

 

By default, Speedtest.net measures your connection speed in mbps, meaning Megabits Per Second. Mbps is the ISP industry-standard, and we use it on Speedtest.net so you can easily compare your result to your broadband plan's speed.

 

However, we offer four different options on your settings page:

  • kbps or Kilobits Per Second - One kilobit is 1000 bits, and bits are the smallest possible unit of information (a little on/off switch). This was typically used by mobile connections, but as mobile carriers get faster they're switching over to megabits.
  • kBps or KiloBytes Per Second - Bytes are made up of eight bits, so one kilobyte equals eight kilobits. File-sizes on your computer are typically measured in bytes, so you'll usually see kilobytes used by download utilities. Bytes are capitalized when used in acronyms to distinguish them from bits, since both start with the letter B.
  • mbps or Megabits Per Second - The default, as we've already discussed. It takes 1000 kilobits to make a megabit.
  • mBps or MegaBytes Per Second - It takes eight megabits to make one megabyte. Most of the files on your computer are measured in megabytes, and if you have a fast connection you'll see this used in download utilities.

 

So where you are getting this 2^10 increment stuff is beyond me.

Edited by Sandman
Link to comment
Share on other sites

 

 

So where you are getting this 2^10 increment stuff is beyond me.

 

The 2^10 increment "stuff" is how the kernel network monitors actually count the bits at the software level. If they are now reporting wrong to artificially inflate the figures like the hard disk manufacturers started doing, "buyer beware". In the code, if you open a terminal emulator in android (or Linux) to look at the network interface status, it WILL count network transfers with kilobits as 1024 bits, etc. That's how computers work internally.

Link to comment
Share on other sites

For reference, Sprint tracks data usage in properly calculated kilobytes, megabytes, and gigabytes. See the Ts and Cs:

 

Data: Services are not available with all Sprint phones. The amount of data transmitted over our network is measured in kilobytes (KB), megabytes (MB) or gigabytes (GB). Unless specified otherwise. 1024KB equal 1MB. 1024MB equal 1GB.

 

AJ

Link to comment
Share on other sites

Yeah I write software and buffer size calculations are powers of 2. This makes sense when you realize there is usually a huge performance hit for "unaligned" memory access (hitting odd memory addresses not even multiples of the appropriate power of 2, eg: 4 bytes or 32 bits on a 32 bit system). So a 16K memory buffer should be 16 * 1024, not 16000.

 

The powers of 10 thing was invented by HDD manufacturers. I'm sorry to see others copying it. There has been a push to differentiate by using Xib suffixes to denote powers of two, eg 2MiB but so far hasn't caught on.

Link to comment
Share on other sites

The 2^10 increment "stuff" is how the kernel network monitors actually count the bits at the software level...In the code, if you open a terminal emulator in android (or Linux) to look at the network interface status, it WILL count network transfers with kilobits as 1024 bits, etc.

 

As they should since they have demodulated the signal received. Two different animals: modulated signal vs. demodulated signal.

 

 

That's how computers work internally.

 

I know how they work internally, usually better than some people. I use to dabble in x86 assembler back in the day, mainly 16bit. :)

Link to comment
Share on other sites

As they should since they have demodulated the signal received. Two different animals: modulated signal vs. demodulated signal.

 

 

 

 

I know how they work internally, usually better than some people. I use to dabble in x86 assembler back in the day, mainly 16bit. :)

 

Regardless of what you want to call it modulated or not, a kilobit (or kilobyte) as measured by Sprint or any ISP, is 1024 bits (or bytes) as noted above. And that is the proper unit of measure.... oh, and I did assembly and ML programming on 8-bit, 16-bit, and 32-bit processors.

Link to comment
Share on other sites

Wow the last thing I thought I would see on this site is guys getting onto a bit measuring contest.

 

Lets just all agree that regardless of how you measure your bits, there still not big enough.

  • Like 2
Link to comment
Share on other sites

Regardless of what you want to call it modulated or not, a kilobit (or kilobyte) as measured by Sprint or any ISP, is 1024 bits (or bytes) as noted above. And that is the proper unit of measure.... oh, and I did assembly and ML programming on 8-bit, 16-bit, and 32-bit processors.

 

What Wiwavelength pointed out was how Sprint tracks data usage not how it measures data transmission.

Link to comment
Share on other sites

What Wiwavelength pointed out was how Sprint tracks data usage not how it measures data transmission.

 

Actually, the data transmission is modulated with CDMA, which means it is spread over 1.2288 mcps (far greater than the data rate). What they track is just as they state: 1KB = 1024 bytes, 1MB = 1024 KB, 1GB = 1024 MB. This is the data transmitted over the network. The same definition is used by AT&T U-verse, Verizon on their data plan usage widget, and Cox on their usage widget. As numerous people on this thread have tried to tell you, this is the proper (and billed) measurement for multiples of computer data. You're wrong.

Link to comment
Share on other sites

What Wiwavelength pointed out was how Sprint tracks data usage not how it measures data transmission.

 

Sure, for transmission, we have QPSK, which carries two bits per symbol, 8-PSK, which carries three bits per symbol, and 16-QAM, which carries four bits per symbol. Heck, with LTE now, we have 64-QAM, which carries six bits per symbol. But I fail to see how RF modulation is at all relevant to this discussion.

 

AJ

Link to comment
Share on other sites

Sure, for transmission, we have QPSK, which carries two bits per symbol, 8-PSK, which carries three bits per symbol, and 16-QAM, which carries four bits per symbol. Heck, with LTE now, we have 64-QAM, which carries six bits per symbol. But I fail to see how RF modulation is at all relevant to this discussion.

 

AJ

 

I agree-- and for the reverse link on EVDO Rev. A, we're limited to 8-PSK and 16-QAM on the forward link. The transmitted data far, far exceeds the amount actually processed and used in the device. I was merely pointing out that nitpicking the transmission rate versus the transmitted total is pointless. The ISP (Sprint in this case) like every other ISP I can think of uses the proper binary-based definition for network data transceived. I'm well aware of the IEC's introduction of the Mebi- and Kibi- SI prefixed in 2000 to try to "simplify" measurement of computer data; however, aside from hard drive manufacturers as pointed out in this thread, nobody uses this new "standard". The guy that is arguing is short-changing himself from Sprint's 300 megabyte per month data roaming limit anyway-- by my calculations, he would short-change himself by 14,572,800 bytes using his rather than Sprint's (and everyone else's) definition of KB, MB.

Link to comment
Share on other sites

I'm not trying to get this started again, I just happened upon this in another forum. From Sprint T&C:

 

The amount of data transmitted over our network is measured in kilobytes (KB), megabytes (MB) or gigabytes (GB). Unless specified otherwise 1024KB equals 1MB. 1024MB equal 1GB.[/Quote]

https://shop2.sprint.com/en/legal/os_general_terms_conditions_popup.shtml

 

Hope this helps clarify things.

 

Sent via Forum Runner on my redsn0w iOS 5.1 iPhone 4 

Edited by gopher_otis
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

  • large.unreadcontent.png.6ef00db54e758d06

  • gallery_1_23_9202.png

  • Posts

    • So, in summary, here are the options I tested: T-Mobile intl roaming - LTE on SoftBank, routes back to the US (~220ms to 4.2.2.4) IIJ physical SIM - LTE on NTT, local routing Airalo - LTE on SoftBank and KDDI (seems to prefer SoftBank), routed through Singapore (SingTel) Ubigi - 5G on NTT, routed through Singapore (Transatel) US Mobile East Asia roaming - 5G on SoftBank, routed through Singapore (Club SIM) Saily - 5G on NTT, routed through Hong Kong (Truphone)...seems to be poorer routing my1010 - LTE on SoftBank and KDDI (seems to prefer KDDI), routed through Taiwan (Chunghwa Telecom) I wouldn't buy up on the T-Mobile international roaming, but it's a solid fallback. If you have the US Mobile roaming eSIM that's a great option. Otherwise Ubigi, Airalo, or my1010 are all solid options, so get whatever's cheapest. I wouldn't bother trying to find a physical SIM from IIJ...the Japanese IP is nice but there's enough WiFi that you can get a Japanese IP enough for whatever you need, and eSIM flexibility is great (IIJ as eSIM but seems a bit more involved to get it to work).
    • So, the rural part of the journey still has cell service for nearly all the way, usually on B18/19/8 (depending on whether we're talking about KDDI/NTT/SoftBank). I think I saw a bit of B28 and even n28 early on in the trip, though that faded out after a bit. Once we got to where we were going though, KDDI had enough B41 to pull 150+ Mbps, while NTT and SoftBank had B1/B3 IIRC. Cell service was likewise generally fine from Kawaguchiko Station to Tokyo on the express bus to Shinjuku Station, though there were some cases where only low-band LTE was available and capacity seemed to struggle. I also figured out what I was seeing with SoftBank on 40 MHz vs. 100 MHz n77: the 40 MHz blocks are actually inside the n78 band class, but SoftBank advertises them as n77, probably to facilitate NR CA. My phone likely preferred the 40 MHz slices as they're *much* lower-frequency, ~3.4 GHz rather than ~3.9, though of course I did see the 100 MHz slice being used rather often. By contrast, when I got NR on NTT it was either n28 10x10 or, more often, 100 MHz n78. As usual, EMEA bands on my S24 don't CA, so any data speeds I saw were the result of either one LTE carrier or one LTE carrier plus one NR carrier...except for B41 LTE. KDDI seems to have more B41 bandwidth live at this point, so my1010 or Airalo works well for this, and honestly while SoftBank and NTT 5G (in descending order of availability) have 5G that's readily available it may be diminishing returns, particularly given that I still don't know how to, as someone not from Hong Kong, get an eSIM that runs on SoftBank 5G that isn't the USM "comes for free with the unlimited premium package" roaming eSIM (NTT is easy enough thanks to Ubigi). In other news, I was able to borrow someone's Rakuten eSIM and...got LTE with it. 40 Mbps down, 20 Mbps up, 40ms latency to Tokyo while in Tokyo...which isn't any worse than the Japan-based physical SIMs I had used earlier. But not getting n77 or n257 was disappointing, though I had to test the eSIM from one spot rather than bouncing around the city to find somewhere with better reception. It's currently impossible to get a SIM as a foreigner that runs on Rakuten, so that was the best I could do. Also, I know my phone doesn't have all the LTE and 5G bands needed to take full advantage of Japanese networks. My S24 is missing: B21 (1500 MHz) - NTT B11 (1500 MHz) - KDDI, SoftBank B42 (3500 MHz) - NTT, KDDI, SoftBank n79 (4900 MHz) - NTT Of the above, B42/n79 are available on the latest iPhones, though you lose n257, and I'm guessing you're not going to find B11/B21 on a phone sold outside Japan.
    • T-Mobile acquiring SoniqWave's 2.5 GHz spectrum  Another spectrum speculator down! T-Mobile is acquiring all of their BRS/EBS licenses and their leases. Details are lacking but it looks like T-Mobile might be giving them 3.45GHz in exchange in some of the markets where they're acquiring BRS/EBS to sweeten the deal and stay below the spectrum screen. Hopefully NextWave is at the negotiating table with T-Mobile so NYC can finally get access to the full BRS/EBS band as well.  — — — — — Edit: Turns out this is a spectrum swap where T-Mobile is basically giving them DoD spectrum in a bunch of markets in exchange for all of SoniqWave's BRS/EBS. SoniqWave will likely turn around and sell the DoD spectrum to AT&T whenever the FCC removes the 40MHz cap.
    • Maybe. The taller buildings on one side of the street all have Fios access and the NYCHA buildings are surrounded by Verizon macros that have mmWave. I don’t think this site will add much coverage. It’d be better off inside the complex itself.
  • Recently Browsing

    • No registered users viewing this page.
×
×
  • Create New...