Posted by: Kai_LeRai | August 6, 2008

Dreamcast Myth: GD-ROM Storage Capacity

Whenever new mediums hit the market, there is a strange confusion about their storage capacities. Usually, the first numbers come from the producers of these media and they are sometimes hyped up to make the medium sound more appealing and fantastic than it actually is. I remember reading a Panasonic brochure about Video-DVDs in 1997. It listed all the amazing features, including a storage capacity of 17GB. Nowadays we know that they used a double sided medium to crank up the numbers. Of course, that is rather deceiving, because it doesn’t matter if you have to flip the same disc or insert just a second one to access the other half of the content, the data that can be read in one session is still just half of that what Panasonic tried to make people think back then.

Today, people are smarter in that regard, due to excessive use (=copying) of the medium and thanks to a massive amount of informations being available by now. The GD-ROM however doesn’t share this level of transparency and therefor you can find differing numbers when it comes to the storage space of Yamaha’s proprietary medium.

The English section of Wikipedia says a GD-ROM holds up to 1.2GB of data and therefor offers almost twice as much space as a normal CD-ROM. Marcus Comstedt, the bright mind, who was one of the first to actually disassemble and run code on the console and who also created the awesome DreamSNES port, writes on his page that it only holds 1GB of data. So who is right on this subject?

The answer is easy: Marcus is right. Contrary to what a lot of private wikis, the official wikipedia and many other highly regarded sources claim, a GD-ROM holds not more than 1GB of data. GD-ROM stands for Gigabyte Disc and that’s a pretty accurate description of the medium. It consists of a low density area that can be read with normal CD-ROM drives and which can hold 18.000 sectors (approx. 35MB) of data. The interesting part of the disc is the high-density part that holds the actual game data. It can store 984MB max. That’s it. More isn’t possible. Combine the space of the low and high-density area and you end up with about 1019MB of storage space, which is 5MB shy of actually being a gigabyte, but who wants to split hairs.

How could this false information of 1.2GB develop, spread and even oust the correct value of 1GB? Let’s check out the official GD-ROM specification that Sega released for developers. It says:

“A GD-ROM has 1.2GB capacity and a conventional CD has 650MB capacity.”

That sounds strikingly similar to what you can read at wikipedia, but how does Sega get these numbers and why does it spread false informations?

The Mathematical Mystery

Whenever data is stored on an optical medium it is being stored in sectors on the disc. These sectors have a distinctive size. Each sector on a GD-ROM (and same goes for most CD-ROMs and DVD-ROMs) has a size of 2352 Bytes, but only 2048 Bytes of data are being stored in it. The rest is taken up by checksums that allow restoring the sector in case of read errors. This is a highly needed feature and is basically the life line of optical media. Without it data security could not be guaranteed and the lightest scratches and even sunlight could be enough to render sectors unreadable.

But not all disc formats make use of the recovery data. Audio-CDs and Video-CDs sacrifice this security data to gain an extra of 15% space. Let’s calculate the maximum sector storage capacity of a GD-ROM:

1019MB : 2048 * 2352 = 1227104256 Bytes = 1170MB

For everybody who doesn’t understand this: if you store 1019MB of data on a GD-ROM AND throw in the recovery data, then you end up with 1170MB. You could also store 1170MB of raw audio data as audio tracks on a GD-ROM. Already confused? Good, there are games in my forum for download. Go grab some. The rest may read along.

If you didn’t start using a PC yesterday, then you might have realized that many mediums don’t hold what they promised. Thanks to some nifty marketing strategies of media producers, many discs, hard drives and sticks don’t contain as much data as they claim. The reason for that is, that they use the total amount of the medium’s storage space in Bytes and use a conversion factor of only 1000 instead of 1024. This results in a higher storage capacity than the medium actually offers. The bigger the storage space of a medium, the bigger are the resulting differences. This started way back with floppy discs in the early 90’s and made people wonder why they only could store 1.37MB on their 1.44MB floppies. While CD-ROMs received a proper capacity description when they hit the market, hard drives suddenly experienced a massive mislabeling of their actual capacities starting in the mid 90’s up to today. Since 1997 all my hard drives boasted bigger capacities than they actually offered. I was surprised when I purchased a 300GB Samsung model a while back, which actually offered 299.7GB of space. Soon after a that, I got my hands on a new 500GB HDD, which turned out to have only 460GB.

Even some optical media is being mislabeled nowadays. Some CD-Rs with sizes of 90min and 99min boast wrong sizes of 800MB and 900MB instead of their actual storage capacities of 790MB and 870MB. The differences are even more apparent with DVD-Rs. A dual-layer DVD-R holds 8152MB (=7.95GB), but is being advertised as a 8.5GB medium thanks to using the false conversion factor of 1000. This also brings me back to my Panasonic example from the beginning. They just multiplied that already wrong number by 2 to get a maximum storage space of 17GB…adding 1.1GB of not existing space to the medium…more than a GD-ROM could even hold.

Speaking of GD-ROMs, let’s go back on topic. As we already calculated, a GD-ROM can hold about 1227104256 Bytes. Sega just used a decimal conversion like everyone else in the industry does and this resulted in 1.23GB of disc space. Subtract the basically useless 35MB from the single density area and you end up with 1.2GB of space for game data…or at least that is what Sega tries to make you believe.

Still, there’s another possibility how Sega could have “calculated” a capcity of 1.2GB. A GD-ROM holds in any case 122:30min. This equals a true storage capacity of 1077MB or 1236MB of pure audio data. Take the latter as base for further calculations and you end up with true 1.2GB. There’s just a little problem with that calculation. Each GD-ROM has a security session between the single and high density area. This security stripe is basically a blank and can’t be used for data storage. So theoretically a GD-ROM could store 1.05GB of data or 1.2GB of audio tracks if that security stripe wasn’t there, but since this is a technological requirement for this format, counting the security session’s theoretical storage amount into the total capacity is like counting all theoritically possible layers into a BluRay-Disc’s  capacity, even though they can’t be even achieved technically right now.

Now let’s have fun with numbers again and let’s show how Sega’s 1.2GB is a total exaggeration. Your head is smoking already? Remember the games in the forum. Convert those 1.2GB properly the binary way and you’ll end up with roughly 1230MB, but we already calculated that even if you abuse the checksum data as storage space, a GD-ROM can’t hold more than 1170MB, so Sega exaggerated the capacity even more. Now consider this: You have 984MB for your game data, but Sega implies a storage space that is 250MB bigger…They just add a whopping 25% of space that you can’t use. Well, that isn’t correct either. Theoretically you can stack a GD-ROM with audio tracks and store over 1.1GB of audio data on it, but when we speak about data storage, we talk about normal 2048 Byte sector storage to ensure data integrity. Also, raw audio data can also be stored as files on the GD instead of audio tracks. Quite a few games rather use this method. So you would be limited to 984MB again, and so this argument doesn’t hold up to anything. Even though there are very few games that are stacked with audio tracks, resulting in combined game data of 1.1GB, sacrificing data integrity for storage capacity is a bad trade and can’t be compared to a storage medium that doesn’t use this “cheap trick”:

Sega compared apples with oranges in their statement. The 650MB of a CD-ROM are first of all correctly calculated by the binary system and also being stored in the data integrity ensuring 2048 sector format, while Sega uses a hyped up value for the GD-ROM that isn’t suitable for data storage at all. If Sega had meant pure data storage without caring for checksum data, then they would have had to use a value of about 750MB for the CD-ROM, because that’s the amount of audio data you can store on a 74min disc. If you go even further and use the exaggerated decimal conversion in conjunction with the wrong sector size, then you would end up with 800MB. Now you have a basis on which to compare these sizes and they already show that a GD-ROM can nowhere near store twice as much as a CD-ROM. Let’s go back to the correct numbers again. That’s the additional storage a GD-ROM offers compared to the following media:

51% compared to a 650MB CD-R (74min)

40% compared to a 700MB CD-R (80min)

25% compared to a 790MB CD-R (90min)

13% compared to a 870MB CD-R (99min)

12% compared to a 878MB CD-R (99:59min)

I used 984MB as a basis for the calculations, because it’s the space that you can use for your games that counts. The 35MB on the single density area of the disc can’t be utilized for the game and are therefor lost and only good for bonus material. You can see that Sega exaggerated a lot. They claimed that a GD-ROM could hold twice the data of a CD-ROM, while it only offers half of that additional space.

Reasons for the lie

id-game developer John Carmack explained in an interview during QuakeCon 2008 that the only advantage of the the PS3 over the XBOX360 is its data storage thanks to Bluray. This offers developers the possibility to ship a game on only one disc instead of multiple ones as on 360, given that the game is even that big. This is important, because the developers have to pay license fees for every disc they release…not just one per game. I don’t know if that was the same back then with Sega, but this seems pretty standard practice, so it’s very likely. The option to utilize the amount of 2 full CD-ROMs while only paying half of the license fees could have been, what Sega wanted to catch developers with. Another positive effect, which Sega might have aimed at, could have been to discourage anyone, who would try to copy GD-ROMs. This might have worked for those, who couldn’t access the discs anyways. but hackers always look for a way, regardless of what their options are. They did and they succeeded, so much for that.

Last but not least, 1.2GB and twice as much storage capacity as a CD-ROM sounds also much better and much more appealing than less than a Gigabyte capacity and only 50% more storage space of a CD-ROM. In the end Sega had to justify the higher production costs for the mediums. Tons of factories out there were able to produce ready to sell commercial CD-ROMs for prices that Sega couldn’t have possibly been able to compete with, but only Sega and its production facilities were able to produce GD-ROM games (apart from Yamaha I guess). Just like Nintendo, they may have seen that as a source of additional income. Demanding a lot more for producing a GD-ROM for the small benefit of 50% more storage space would have sounded little convincing.

Advertisements

Responses

  1. Very well explained :-)

  2. Xiaopang, I was wondering what brand of 99min cdrs work good for the dreamcast

    • all i tried worked nicely (skc + mediarange)

      it’s more a question of how your burner handles upper storage parts of 98+min. that can vary from brand to brand, even batch to batch

  3. Actually the factor 1000 for KiloByte (or 1000*1000 for MegaByte) is mathematically (and SI-based) correct and not a false labeling.
    Kilo means 1000 and mega means 1000 kilo. period.
    The SI-Unit for the factor 1024 would be kibi. (e.g. KibiByte)

    • uhm no… mathematically it is correct, sure. but we’re talking about informatics here. you can’t just use definitions of one science and throw it in with another to make a point. when the prefixes were first used in the computer genre they were not si-based any more and therefore any comparison with the mathematical side is moot. and it is even more so by considering that storage mediums are being handled in a binary way, which makes decimal a false advertisement.

      btw, the whole kibi/mebi/gibi-mess was only invented after that marketing scheme of willingly false conversions was used and after users like you started to adapt to it. when i studied computer technics, no one actually used those values, because as a computer programmer you know that 1kb=1024bytes, regardless of what it means on other scientific fields or what some hdd-ads try to tell you. period.

      according to your logic you would also have to argue that a byte is not 8 bits but 10, but you can’t. the amount for the size of this unit was established just as the size for the prefixes.

      whether or not this kind of labeling is mathematically correct, it is false advertising in any case. this goes especially for a computer company that targets not just mere consumers, but developers. do you think developers suddenly started to expect that sega uses decimal based prefixes for size measurement? CD-sizes were binary based back then (and still are) and devs who stuck with sega through the ages actually could expect binary based calculations, because those were also used for sega’s cd-based console predecessor. same goes for even earlier consoles. buying a 16MBit cartridge gave you the full 2097152 bytes and not just 2000000, so there was no reason to expect that sega suddenly switched to the non-standard decimal-calculation. and it was non-standard back then. cds used correct binary sizes, so did RAM (and still does), proms and even hdds before ibm came along in the mid 90s and changed that…

      i think it’s pretty sad that normal users defend a marketing ploy from which they have nothing to gain, but only to lose. this is the lowest it can get and the best a marketing prank can achieve. the binary based prefixes have been around for decades and there is no good reason to change that, other than to save some money by selling people less for more, or can you show me one happy guy who thinks it’s a great idea that his 2TB hdd only holds 1860GB. decimal-binary-skew is not constant, but grows with every new conversion. it’s already so big, that it can’t be dismissed as being minor and unless we see some form of decimal data handling in the future this can be considered false advertising. and actually the companies agree… there was a lawsuit against the main manufacturers of computers in 2003 and it was settled out of court with the plaintiffs being the benefactors, e.g. western digital paid half a million dollars just to shut the consumers up. apparently they didn’t even believe a judge would agree on this asinine practice, or can you name any other example, where you pay for more than you get, just because a non-compatible conversion is being used? or just think about what would happen if your paycheck was suddenly reduced thanks to such a conversion. in the end, there is no good reason to drag a decimal system into a binary environment…that’s why the binary system was invented in the first place, because computers don’t work with it.

      in the end, you totally missed the point of the article by saying using a decimal conversion is just valid. the article deals with the misconception of gd-roms holding up to 1.2gb of data and apparently you over read the citation of sega’s own manual: “A GD-ROM has 1.2GB capacity and a conventional CD has 650MB capacity.”

      even if they actually used decimal conversion to get the value of 1.2gb, then that doesn’t excuse the fact that they compare a value of a decimal conversion with one of a binary conversion. that is false advertising in its purest form and there is no justification that could explain it.


Categories

%d bloggers like this: