“Newly formed hard disk drive company Hitachi Global Storage Technologies on Monday unveiled plans for a tiny drive that will be used in consumer electronic devices and said it will turn a profit in 2004. Hitachi Global Storage said it plans to sell next fall a 1-inch microdrive that has 4 gigabytes, or 4 billion bytes, of storage space. That compares to the current top of the line microdrive with 1 gigabyte of storage that IBM had made.” Read the article at ZDNews.
Hitachi Global Unveils 4GB microdrive
2003-01-06 Hardware 15 Comments
There are 1024 megabytes per gigabyte, but for years the hard drive makers have been pretending it is 1000 and shaving several percent of the advertised amount of storage space. If this hard drive contains 4 billion bytes of storage space, then it is infact about 3.75 gigabytes.
My ’30’ gb drive is actually under 28 gigabytes.
What do you mean.
K = 1000 not 1024.
Break out the dictionary and you’ll find that this is correct. Only in computers do we say a K = 1024; this is because 1024 is a power of 2. The hard disk people are using the dictionary meaning and not the computer version.
I do agree with you though; they are in the computer bisnesss and should be using the computer version of K and not the dictionary.
>>Newly formed hard disk drive company Hitachi Global Storage Technologies <<
I think they mean newly *PURCHASED* company. Hitachi just bought (part of?) IBM’s HDD division.
Just like news websites (ZDNews and OSnews) can get away with misleading their readers.
The title said “Hitachi unveils 4Gb microdrive” and when you read fine print it’s actually plans to produce were unveiled. Not the microdrive itself.
After reading OSNews snippet you may also fall under impression that Hitachi is competing with IBM in manufacturing microdrives. If you read original article then you can find that IBM is out of harddisk manufacturing business now and Hitachi Global is using their technology and facilities to develop further.
The news about IBM downfall in storage market is more interestinig to me than promise of 4Gb microdrive in next year. After all , Linux, BeOS and *BSD fit existing microdrive OK. Which still makes little sense as reguar BIOS doesn’t boot from pcmcia and usb adapter.
The hard disk people are using the dictionary meaning and not the computer version.
In other words, the hard disk companies are deceiving people, as it only becomes customary to use 1000 instead of 1024 when it works in your favor financially.
or firewire or.. Luckily I heard rumor of some openBIOS projects. Maybe they’ll patch up the missing functionality for my hardware. See how closed source computing hurts us?
Oh and “deception” is a subdivision of the marketting department in all commercial corporations.
“”What do you mean.
K = 1000 not 1024.
Break out the dictionary and you’ll find that this is correct. Only in computers do we say a K = 1024; this is because 1024 is a power of 2. The hard disk people are using the dictionary meaning and not the computer version. “”
Nope, it’s completely cynical and business based.
Here’s an example why.
1.44mb 3.5 inch floppy has 2880 * 512 byte sectors
In disk manufacturer terms
Total bytes = 1474560
Total kilobytes (1024 bytes/kb) = 1440 kb
Total megabytes (1000 mb/kb) = 1.44 mb
You can’t get away with the excuse of using the dictionary definition for mega whilst not using the dictionary definition of kilo. Either you use one, the other, or you’re deliberately selecting definitions to massage your figures. The disk manufacturers are guilty of the latter.
There are computing scientists working on non-binary computing theories (i.e. base-3 ternary logic). Does this mean that in say 30 years, the retail box for a hard drive should contain a base-10 capacity, base-2 capacity, base-3 capacity, base-4/5/6/7/8/9/10… capacity chart.
If your smallest unit is tristate logic then you can no longer describe it in terms of bits. A bit has only 2 states. Bytes are a collection of 8 bits and so on up the line. You can’t describe 3 and only 3 possible states using bits. 2 bits doesn’t cut it, you can represent 00,01,10,11 ie 4 states.
Once you have 3 possible states you have to come up with a different description (A trit?). At that point kilobytes/megabytes/gigabytes become irrelevant and we’ll still be discussing the hardware manufacturers massaging figures.
Let’s imagine for a second. Perhaps the smallest addressable unit of memory becomes 27 trits or a tryte (3^2). One kilotrye becomes 59049 trytes(3^10). One megatryte becomes 3486784401 trytes (3^20). One gigatryte becomes 205891132094649 (3^30).
Now let’s say the manufacturers continue in their current ways. So they use the programmer’s definition of a kilotryte and the dictionary definition of a gigatryte.
The manufacturer sells me a 30gb drive.
The manufacturers view:
This drive contains 30*1000*1000*59049 trytes of space.
Manufacturer total: 1771470000000 trytes
This drive contains 30*205891132094649 trytes of space
Programmers total: 6176733962839470 trytes
Difference: 6174962492839470 trytes
Or to put it another way I’ve received 0.03% of the space I was expecting.
Ok, so I’m just bored and messing around, but don’t think they won’t do it :>
I guess the real reason this bugs me so much is that filesizes are currently shown in one scheme whilst disk drive sizes are sold in another. This is bound to lead to confusion, and the drive manufacturers are the ones benefiting from it because it gives the impression the drives have more capacity than they actually do.
Should of course be a 30 Gigatryte drive. That’s the problem with fantasy technology, you can never keep up :>
I run an msi kt4-ultra, and i can boot off of my usbmass storage devices just fine, as i can off firewire hds. I booted off of a creative MuVo 64 mb for the fun of it once .
I just thought you all should know.
giga means one billionth (10^9). Meaning that if a gigametre, it is one billion metres. However, due to the fact that computers use binary numbers, giga for storage is actually 1,024,000,000. So gigabit can either mean 1,000,000,000 bits or 1,024,000,000 bytes. Neither is right, neither is wrong.
Deceptive, maybe. I’ll bet hard disk companies wouldn’t use the original dictionary defination of giga is the number would be higher using computer terms.
Most of the whiners here obviously are ungrateful to the HD industry for its amazing achievements over the last 20yrs. When I 1st bought a Fat Mac 512K with no possible Apple HD upgrade, a 3rd party 20M HD cost me $1400+ and it produced more return on investment than all the following HDs purchased afterwards. My recent 60G HD was $70, thats 60,000 * cheaper per byte.
I consider modern HDs to be disposable since they fail after a few yrs unless treated like royalty. Every disk drive is an analog device with extremely high error rates at the surface level. Typically about 2 bits are written on the surface using some very fancy ECC codes to give just one useable data bit. So what if K is 1000 for that industry, its been historical long before consumer drives were even on the horizon. But like the CRT diagonal fiasco, they probably should have switched to 2^10 useage since a 1G v “1G” is far bigger than 20M v “20M” percentage wise.
I’d agree that the HD industry has made some very big improvements in size of their products, but access time hasn’t improved by anything near as much.
I think that this particular form of media is reaching the end of its lifespan. There are very real physical limits on just how low you can take seek and read times.
With modern processors capable of handling data rates well in excess of those even a fast HD setup can supply I think the search is on for something entirely different.
What it will be I don’t know, first thought I have would be some kind of large RAM array with builtin battery power supply. Such drives would have a natural lifespan of course (When the battery runs out). Memory state would be saved across shutdowns in the same way the clock is saved on a PC. With RAM so cheap I figure you could stuff at least a few gigabytes into the same space as a hardrive takes up atm.
Now that would solve the access time problem, along with the error checking problem (The equivalent of a bad sector would just become unadressable memory which you patch around with clever circuitry). Then the real bottleneck becomes the bus.
Of course if you went the persistent OS route you could throw away the entire conventional RAM, hook your state saving RAM up to the address lines instead and have the entire system essentially run in memory. That solves the bus problem as well.
Now consider that in this system all memory accesses that used go through the HD ports are being done at the same speed as conventional memory. It would be very very fast indeed. Unfortunately programmers would once again have to start taking app size into account when they are coding, aww poor babies, because you’ll never be able to get these drives to the same size as conventional HDs.
Some good points, I esp aggree about lousy HD speeds rel to amazing sizes. The Micro drives only make the speed worse.
One likely successor is EEROM in small quantities, I’d be interested to see how the Keychain SS HDs perform once they use USB2 controller. Some mobos can now boot off USB2.0. EEROMs suffer only a few problems, 1) not cheap enough for vast storage but cheap enough to put OS critical file on. 2) wear out if written too often in same addresses. If the OS was aware of these limits, could be a good boot drive for mostly rd only access with usual HD for bulk store.
Longer term is the Tunneling Electron array stuff that IBM has been working on in Zurich, not solid state, but based on MEMs devices.
You would think that DRAM drives would be obvious, ie near 0 latency, & much higher bandwidth etc. I met a guy from Quantum, told me about their SS ram drives of a few yrs ago, ultimately most people were dissapointed with SS ram drives because the OS using it had no idea what it was and pretended it was still and IDE device.
Once the Keychain devices get fast & bootable on USB2, it becomes a simpler matter to add DRAM to bulk it up as a cache for regular HD. But then why not let the OS use main mem as ram drive?
As for end of the road for magnetic storage, nope, there is plenty more to come, probably another 1000x in the next decade, the speed will hardly improve unless you want very hot high RPMs.
If I can get BeOS to run off a USB2 bootable mobo with EEROM and a few Gig of Ram for ram drive, I’d be pretty pleased, but there are a few problems.