[springtide]

Quote:

Originally Posted by smuh Actually that could be very funny (or not ) I recently compressed pictures send them to my email at work and after decompressed them I noticed that everybody on the pics were much slimmer... blah

That's funny..... I get exactly the opposite.....

I compress (lossless obvious) pictures on my PC upstairs and when I view them downstairs on my Widescreen TV everyone looks fatter?

Could this be due to the lossless compression or maybe the difference in height between the two rooms? (..time is running slower downstairs right? .... Or is that upstairs.....)

Strange, but true...

 

[Distroyed]

Quote:

Originally Posted by AdamWill I don't think you'd get an 'anomaly' on a disk which produced anything other than a read error. In other words you'd get a trashed file, not something that worked but sounded different.

Im talking about a physical anomaly that would alter a few bits here and there, without totally destructing the data. It's possible. But I concede - it's still far too unlikely to be taken seriously (unless probability theory is questioned)

 

[devwild]

Quote:

Originally Posted by Distroyed Im talking about a physical anomaly that would alter a few bits here and there, without totally destructing the data. It's possible. But I concede - it's still far too unlikely to be taken seriously (unless probability theory is questioned)

These errors are detected by parity checks. If they fail on a data drive (cd, hd, network, etc) the bits are re-read to see if there was a simple error, and if the bits can't be read properly after a few tries, you get a read error (and probably go buy a new disk).

On an audio cd this is done as well, but if the parity fails more than once the standard allows for guessing. Honestly I think redbook should have been updated at some point so it didn't, or at least the cd players reported these errors. That would reduce a lot of stupid arguements in the audiophile world, and have stopped the insanity of parity-error-inducing copy protection methods.

While technically possible for an error to occur that passes parity check, the odds are small, and typically when there is one error, there are more, and you have a detected error anyway, making the odds almost infinitesimally small. In addition, many data files and file systems these days use checksums, which pretty well pick up the slack.

In any case, it is impossible for errors to occur repeatedly in a file in such a way that you don't notice them, and hear a tonality difference in the entire file or files. And insanely small for even one error with current technology.

 

[JahJahBinks]

A long time ago a professor told us in class that 30-40% space on a CD is used for parity checks and ECC, to ensure data integrity.

 

[springtide]

Quote:

Originally Posted by JahJahBinks A long time ago a professor told us in class that 30-40% space on a CD is used for parity checks and ECC, to ensure data integrity.

Unfortunatly (I believe) they decided to use some of this space for additional music!

Yes, I remember that (before CD's).............................

The was a technology TV program in the UK called "Tomorrows World"... They took an Audio CD, drilled a 5mm hole in it and played it back without any problems (and described about the ECC).

I've never understood how they got from there (maybe they weren't playing the track with the hole through it?) to a small scratch causing the CD to skip.

 

[Dave_M]

Quote:

Originally Posted by devwild These errors are detected by parity checks. If they fail on a data drive (cd, hd, network, etc) the bits are re-read to see if there was a simple error, and if the bits can't be read properly after a few tries, you get a read error (and probably go buy a new disk)...

When i copy a large amount of data from 1 hard disk to another (15GB for example), i will usually get 1 or 2 checksum errors if i use MD5 cheacksums to check the data copied ok. Coruption may be more common than you think. Probably most people will never notice it (or maybe its just my PC). But 1 or 2 bits in 15 GB is so small its probably not worth worrieing about. Especaly in music, you will never notice.

 

[Regus]

Quote:

Originally Posted by Dave_M When i copy a large amount of data from 1 hard disk to another (15GB for example), i will usually get 1 or 2 checksum errors if i use MD5 cheacksums to check the data copied ok. Coruption may be more common than you think. Probably most people will never notice it (or maybe its just my PC). But 1 or 2 bits in 15 GB is so small its probably not worth worrieing about. Especaly in music, you will never notice.

if data corruption was that common your programs would crash constantly...

It may not be a problem with media content but program code cannot survive data errors - you would have pointers pointing all over the place and loops running 1 mio rather than 10 loops and so on and so forth - and large corporate databases would be going haywire every other minute oops that data page doesn't really start at block 1549403 but at 96832...

Computer software is absolutely dependant on 100% data retrieval accuracy.

 

[devwild]

Quote:

Originally Posted by Dave_M When i copy a large amount of data from 1 hard disk to another (15GB for example), i will usually get 1 or 2 checksum errors if i use MD5 cheacksums to check the data copied ok. Coruption may be more common than you think. Probably most people will never notice it (or maybe its just my PC). But 1 or 2 bits in 15 GB is so small its probably not worth worrieing about. Especaly in music, you will never notice.

This should not be happening, at least not commonly... either there is something wrong with your setup, or you're picking up on files that actually change around the time of the copy, like temp files, system files, or file headers modified by the os for some reason. If it is really happening on random data files on regular basis, you likely have a bad disk or disk controller. I would suggest checking your event viewer/syslog to see if there are any reports of block read errors. Frequent "undetectable" errors like that are usually accompanied by detectable ones, unless the mismatched checksums are because of some other reason. You would also be seeing other problems such as unexplained crashes if this was happening under normal operation.

Errors definitely happen, but not on a frequent basis - except on file sharing networks, irc, nntp, ftp, etc, where the transport (the internet) is unreliable and the checking is minimal (especially on UDP based protocols). Bittorrent uses MD5 checksum for this reason, which has actually made it more feasible for legitimate distribution, not just a p2p replacement. On the filesystem side, you should definitely be using journalling file systems these days - NTFS/HFS+/EXT3, rather than FAT or the like, which also helps.

 

[Tachikoma]

Quote:

Whatever you do, though, don't use DDR (Double Data Rate) memory on your computer: It leads to a weird echoing resonance in the sound.

Say no to DDR! Now where did I leave my 5 year old mobo...

 

[Guust-Fi]

A little off topic again, but I was wondering... say, I've got a music cd and I rip it in EAC (Secure with NO C2, accurate stream, disable cache).

This way if only 1 bit would be unreadable the cd wouldn't be ripped correctly.
What if this would have been a data cd and I was trying to copy the contents to my hd?
Would the copying procedure fail or would it just copy the data including errors?

...or is the computer able to recover this lost data in some kind of way?

 

[Regus]

Quote:

Originally Posted by Guust-Fi What if this would have been a data cd and I was trying to copy the contents to my hd? Would the copying procedure fail or would it just copy the data including errors? ...or is the computer able to recover this lost data in some kind of way?

Data CDs have a lot more error correcting information on them than audio cd's do. For some reason they slacked on the requirements for audio cd's (something about Bethovens 9th or something being able to fit onto a single cd according to the rumor anyways)

And so a Data CD has to be a lot worse off for you not to be able to read it with 100% accuracy, and if you manage to copy a file off of it the odds of it being 100% identical to what was put on is nearly 100% - it is probably possible for the data to be altered in such a way that it can be read off the CD containing errors but the probability of that happening is infinitessimal.

If Audio CDs were made to the same spec there would be no need for programs like EAC.