jrdave: hey guys… a quick question for u… I have an old PC with 512MB RAM… debian (free -m) is reporting 504MB only! Any explanations? (the video is NOT shared)
mechtech: jrdave: rounding…I have 1024, but Ubuntu says only 1001
jrdave: mechtech: so if I convert the amount manually it show be correct (from bytes)?
mechtech: jrdave: makes sense to me
jrdave: mechtech: lemme try
jrdave: mechtech: NO… its in KB and when divided by 1024 it shows 504MB!!!
mechtech: jrdave: why would you divide by 1024?
jrdave: mechtech: coz the default output is in KB
mechtech: jrdave: you need only move the decimal place three places to the left to convert from KB to MB
jrdave: mechtech: 1MB=1024KB and not 1000
stovicek: jrdave: marketing in the storage world isn’t entirely accurate as many find with more often with hard drives. same holds true for memory. you probably have somewhere around 512000 bytes in marketing terms, that’s 512MB, not 504MB. It all depends on how you divide the bytes, either by 1024 or 1000.
jrdave: stovicek: BIOS reports 512MB
mechtech: jrdave: as I said before and stoveicek has more verbosely described…rounding
stovicek: jrdave: because bios divided it by 1000. Either way, you have a half gig.
mechtech: jrdave: therefore, to convert KB into MB, simply move the decimal point three places to the stovicek: jrdave: I’ve got 4Gb. Bios reports it as 4GB. But I see 3830MB. It’s still the same 4GB, just divided by a different factor.
jrdave: stovicek: I see

The explanation I gave wasn’t entirely accurate, but it was still partly true. Anyone buying a new hard drive can attest to the feeling of an immediate loss of hard drive space when they’re 1TB hard drive shows as having only 931GiB. Both numbers are correct and the manufacturer took advantage of the two possible measurements by selling a 1TB drive to the consumers. So what gives?

If you noticed, I used the notation of GiB, instead of GB. The reason for this is that byte prefixes such as kilo, mega, and giga are defined by the International System of Units to mean multiples of 1000. A kilobyte (kB) is then equal to 1000 bytes, a megabyte (MB) equal to 1,000,000 bytes, and a gigabyte (GB) meaning 1,000,000,000 bytes. This is the decimal system.

Traditionally, memory and storage measurements were made using a factor of 1024. Your end result would be a measurement in “binary”. A kilobyte would equal 1024 bytes, a megabyte would equal 1,048,576 bytes (1024 x 1024), and a gigabyte would equal 1,073,741,824 bytes (1024 x 1024 x 1024). But that would mean that the same terms are equal to two different sets of measurements.

The IEC, who is a non-profit international standards organization, has proposed using byte prefixes such as kibi, mebi, and gibi (among others) in place of kilo, mega, and giga for the “binary” measurements. Therefore, 1 kilobyte (KB) equals 0.976 kibibytes (KiB). Getting back to the consumer’s new 1TB drive, 1 terabyte (TB) equals 931 (GiB) gibibytes.

This also relates to the IRC conversation above. When the user issued free -m at the command line, he was actually seeing the “binary” measurement. The man page for free incorrectly states that the different display switches will show kilobytes, megabytes and gigabytes. But as I described above, those would be the decimal factors of 1000, not the “binary” factors of 1024 that it uses. For instance, the following two examples are from my own laptop (slightly cropped to better fit the blog’s layout):

adam@obsidian:~$ free
             total       used       free
Mem:    4016529408 3787829248  228700160
-/+ buffers/cache: 1780539392 2235990016
Swap:            0          0          0

First, show in straight bytes and below showing “megabytes”:

adam@obsidian:~$ free -m
             total       used       free
Mem:          3830       3611        218
-/+ buffers/cache:       1697       2132
Swap:            0          0          0

My laptop has 4GB installed. From the first display, you can clearly see over 4,000,000,000 bytes. So, technically, there is actually 4.02GB. But in the next display, we see something dramatically different. Now there’s only “3830MB”. What’s incorrect is the expected outcome and the implied notation. If the man page was correct, it should show 4017MB, but it doesn’t. What we truly see is 3830MiB, mebibytes and not megabytes. Nothing was lost, it’s just a consequence of using different factoring.

It’s still confusing for me even after knowing it as well as I do. The point I’m trying to make is that it can be unsettling especially in today’s computer age with such massive storage devices, but users shouldn’t panic. You can start by reading the entries at Wikipedia for a more in depth, and possibly clearer, explanation than what I provided here.

The OLPC XO was designed for children in remote areas of the world. The screen had to be visible in bright sunlight, the case and the keyboard had to withstand the extremes of nature, it needed to be powered in areas where an electrical source is rare, and it needed to be cheap.

Prices for the G1G1 program are the same as they were last year. A $199 USD donation allows one to be given, while a $399 USD donation allows one to be gotten as well. Amazon was chosen as the sponsor for this year’s promotion in order to alleviate the distribution issues that occurred last year.

The ati and radeon drivers are fine, but the fglrx makes the compiz effects a lot smoother. I usually install compizconfig-settings-manager to tweak the effects like activating Desktop Cube and Rotate Cube. I’ve gotten a few colorful metaphors from friends after watching me switch to another desktop by rotating the cube around.

Here’s the complete message:

Reports have been coming in that the e1000e ethernet driver for Intel GigE
chipsets, as included upstream in Linux 2.6.27, may under certain conditions
irreparably damage your ethernet hardware by corrupting the on-board
firmware.

While we expect alphas to include a fair number of bugs, and rely on your
continued support and testing to help resolve these bugs for Ubuntu
releases, bugs that damage hardware are quite another matter. As a result
of this bug we must recommend that users do *not* use Intrepid alphas,
including the LiveCDs, on machines with Intel GigE ethernet.

Efforts to safeguard the hardware of affected users are ongoing, though
initially this will consist of a workaround in the form of disabling the
e1000e driver. Alpha 6 CD images will not be re-issued for this fix, so
affected users are advised to either use daily CD images once a resolution
is in place, or to wait for the release of the 8.10 beta, due to be released
on October 2.

Further discussion of this issue can be found on the ubuntu-devel mailing
list at
<https://lists.ubuntu.com/archives/ubuntu-devel/2008-September/026559.html>,
and progress on resolving this issue can be tracked at
<https://bugs.launchpad.net/bugs/263555>.

May  5 17:42:02 nyx kernel: [ 4265.867147] UDF-fs: No partition found (1)
May  5 17:42:02 nyx kernel: [ 4265.867234] isofs_fill_super: bread failed, dev=sr0, iso_blknum=16, block=16
May  5 17:46:31 nyx kernel: [ 4394.145010] UDF-fs: No partition found (1)
May  5 17:46:31 nyx kernel: [ 4394.145119] isofs_fill_super: bread failed, dev=sr0, iso_blknum=16, block=16
May  5 17:46:58 nyx kernel: [ 4405.904726] Buffer I/O error on device sr0, logical block 0
May  5 17:46:58 nyx kernel: [ 4405.904732] Buffer I/O error on device sr0, logical block 1
May  5 17:46:58 nyx kernel: [ 4405.904734] Buffer I/O error on device sr0, logical block 2
May  5 17:46:58 nyx kernel: [ 4405.904737] Buffer I/O error on device sr0, logical block 3
May  5 17:46:58 nyx kernel: [ 4405.904739] Buffer I/O error on device sr0, logical block 4
May  5 17:46:58 nyx kernel: [ 4405.904741] Buffer I/O error on device sr0, logical block 5
May  5 17:46:58 nyx kernel: [ 4405.904743] Buffer I/O error on device sr0, logical block 6
May  5 17:46:58 nyx kernel: [ 4405.904745] Buffer I/O error on device sr0, logical block 7
May  5 17:46:58 nyx kernel: [ 4405.905542] Buffer I/O error on device sr0, logical block 0
May  5 17:46:58 nyx kernel: [ 4405.905549] Buffer I/O error on device sr0, logical block 1
May  5 17:47:10 nyx kernel: [ 4411.068188] printk: 10 messages suppressed.
May  5 17:47:10 nyx kernel: [ 4411.068192] Buffer I/O error on device sr0, logical block 0
May  5 17:47:10 nyx kernel: [ 4411.068196] Buffer I/O error on device sr0, logical block 1
May  5 17:50:03 nyx kernel: [ 4488.505841] isofs_fill_super: bread failed, dev=sr0, iso_blknum=16, block=16
May  5 17:51:10 nyx kernel: [ 4517.864765] isofs_fill_super: bread failed, dev=sr0, iso_blknum=16, block=16
May  5 17:55:45 nyx kernel: [ 4644.173955] isofs_fill_super: bread failed, dev=sr0, iso_blknum=16, block=16
May  5 17:59:51 nyx kernel: [ 4799.395757] UDF-fs: No partition found (1)
May  5 17:59:51 nyx kernel: [ 4799.396032] isofs_fill_super: bread failed, dev=sr0, iso_blknum=16, block=16
May  5 18:06:01 nyx kernel: [ 5073.300780] UDF-fs: No partition found (1)
May  5 18:06:01 nyx kernel: [ 5073.302026] isofs_fill_super: bread failed, dev=sr0, iso_blknum=16, block=16

After googling numerous terms and reading many pages of possible solutions, I finally gave up tinkering with config files and decided to swap the drive out for another. Considering that the CDs and DVDs that I’m trying to mount work quite well on the laptop and the server, it must be a problem unique to this machine. With the new drive at hand, I began the removal process just to find that the DVD drive’s connector fell apart. There were just enough connected pins to identify its presence, but nothing more. After swapping out the cable and rebooting the machine, we’re back reading everything five by five.

I had been working on that problem for days. I can only assume it broke while I was relocating drives between the desktop and the server. If I hadn’t had the spare DVD drive, I probably would not have checked the cable. Pictures of the broken cable compared to a not-broken cable appear to the right. The broken one was so subversive, I don’t even know what happen to the clip. I remember it in my hand and setting it down, but simply up and disappeared!