The First Terabyte Hard Drive Reviewed

mikemuch writes "ExtremeTech has a review and benchmarks of the Hitachi Deskstar 7K1000 1TB Hard Drive, which ushers in the terabyte age. It performs well on HDTach and PCMark benchmarks, though not as speedily as professional-grade drives. It could be just the ticket for digital media junkies. 'One of the first issues to note is that you may not see an actual one terabyte capacity on your system. First, the formatted capacity is always less than the raw space available on the drive. Directory information and formatting data always take up some space. Second, the hard drive industry's definition of a megabyte differs from the rest of the PC business. One megabyte of hard drive space is 1,000,000 bytes: 10^6 bytes. Operating systems calculate one megabyte as 2^20 bytes, or 1,048,576 bytes. Once installed and set up, Hitachi's 1TB hard drive offers up an actual formatted capacity of about 935GB, as measured by the OS. That's still a lot of space, by anyone's definition.'" Update: 05/17 21:52 GMT by Z : Adding '^s' missing from article.

Re:106 bytes and 220 bytes, ???

[Roger W Moore]

They mean that superscript tags don't work when submitting stories to slashdot. If should read 10^6 and 2^20.

Re:Zonk

[ReverendLoki]

I see where they screwed up.. in this sentence:

One megabyte of hard drive space is 1,000,000 bytes: 106 bytes. Operating systems calculate one megabyte as 220 bytes, or 1,048,576 bytes.

"106" bytes should have been written as "10^6 bytes", and "220 bytes" should be "2^20 bytes". Either that or actually put the powers in superscripts.

Re:Zonk

[nine-times]

It was probably originally typed as:

One megabyte of hard drive space is 1,000,000 bytes: 10⁶ bytes. Operating systems calculate one megabyte as 2²⁰ bytes, or 1,048,576 bytes.

And then the tags got stripped somehow.

First review? ummm... Anandtech, March 19th....

[Fallen Kell]

Initial review March 19th:

http://www.anandtech.com/storage/showdoc.aspx?i=29 49 [anandtech.com]

Follow-up RAID performance April 19th:

http://www.anandtech.com/storage/showdoc.aspx?i=29 49 [anandtech.com]

Follow-up to the follow-up April 23rd:

http://www.anandtech.com/storage/showdoc.aspx?i=29 74 [anandtech.com]

Math

[KrayzieKyd]

1,000,000,000,000 bytes / 1024^4 = 931.23 GB formatted. Math is our friend.

Megabyte/Terabyte

[Bobb Sledd]

We can only guess what Zonk meant to say. But I'll try to make some sense.

First, hard drive manufacturers have always calculated drive space differently than the rest of the entire computing world. It allows them to say that a drive is bigger than it really truly is. They've been able to do it for years, and lawsuits have been lost and won on this very issue. But essentially, their use of the metric words "kilo," "mega," and "giga" are the literal meanings of "1000," "1,000,000" and "1,000,000,000" instead of the computing world's 1024 multiplier.

Therefore, a "kilobyte" to them is 1,000 bytes (as opposed to 1,024 bytes in real life), and a "megabyte" is "1,000,000" bytes (as opposed to 1,048,576 bytes [1024 x 1024]), and a "gigabyte" is 1,000,000,000 bytes (instead of 1,073,741,824 [1024 x 1024 x 1024] bytes in real life).

The real difference in a terabyte? Divide 1,000,000,000,000 by 1024/1024/1024 and you get 931.32 gigabytes. That's a theoretical limit, mind you, and there is overhead for cluster size, partition info, FAT tables, etc., so you really don't even get that.

Doesn't that byte?

Re:why explain prefixes?

[imsabbel]

No, it isnt.
From a user perspective, it doesnt matter shit if a byte is 8 bit. That solves the whole "base 10" crisis you seem to have.

Next, gather your brain for some thought: you have a CPU that runs at some Ghz, and memory busses/network cards that run at megabyte/s.

Now guess what kind of "mega" those aspects used from the beginning of time? Yes, SI.
Just for some strage reason, for memory and disks people thought that 1024 is close enought to 1000 as not to matter.
Too bad now we are at 10^9 vs 2^30, where the relative differences arent ignorable anymore (8.5% is quite abit..)

Now the next problem with the binary idiociy in storage space is the plethora of bastardisations: People doing megabytes as 1000 kiB, or Gbyte as 1000 MiB, or 1000000 kiB, which all gives different results, of course...

Re:eh?

[Anonymous Coward]

Yeah, say "kibibyte", "mebibyte", or "gibibyte" in any IT shop, and you WILL be called "faggot".

I have one - this is what ext3 defaults too.

[Anonymous Coward]

df .
Filesystem           1K-blocks      Used Available Use% Mounted on
/dev/sdb1            961432072    221096 912372976   1% /data

Re:Damn...

[matt21811]

My website on historical hard disk pricing shows that 10GB HDs were only sale roughly between 1998 and 2001. given the maximum extremes this puts the poster current age range between 15 to 20.

http://www.mattscomputertrends.com/harddiskdata.ht ml [mattscomputertrends.com]

This page is great for when you want to date a hard disk or when a certain size disk first became available.

Re:Lots of space?

[Anonymous Coward]

My first hard disk was on my Grog 1. It consisted of a large rock that was spun on a stone platter using a Dino-Drive(tm) and an irritable gorilla with a hammer and chisel for a read-write head. It held 40 bytes (because no-one would ever need more than 640 bits), was kind of slow and weighed a ton (sic) but hasn't lost a bit to this day. You had to take cover when it was writing to avoid the flying shards and, well, the dino and gorilla doo-doo kind of got on your shoes but was good for the garden.

Re:"...by anyones definition"

[david_thornley]

Note to future self: remember when 1 terabyte was considered a lot of storage? those were the days....

Then there are those of us who remember when 1 gigabyte was considered a lot of storage.

Re:Megabyte/Terabyte

[init100]

Therefore, a "kilobyte" to them is 1,000 bytes (as opposed to 1,024 bytes in real life)

Actually, it is more like the "kilo = 1000" is the real life meaning, and the "kilo = 1024" is something dreamed up by some hacker in his own little world. I mean, one kilogram is 1000 grams, one kilohertz is 1000 hertz, one kilometer is 1000 meters, etc.

Re:Ready for my RAID5.

[bnenning]

I don't know about in 16 years but I bet within 10 we're talking about the first petabyte drives.

Moore's Law has been pretty accurate for drive capacities, so factor of 1000=10 doublings=15 years. I'd expect "only" 100TB drives in 10 years.

Re:why explain prefixes?

[SEMW]

For an example of the difference, people don't say that they have a megadollar or a gigaduck (not normal people)

Maybe not, but they do say "centidollar" -- at least in its abbreviated form, "cent".

I'd like to see the pedantic community start discussing things in mebimeters and gibiwatts before they start trying to change the computer community's standard meanings.

Huh? There would be no reason for people to use mebimeters and gibiWatts, because metres and Watts are Metric units and thus are base 10. It would not make sense to use base 2 prefixes (mebi, gibi) in a base 10 system. It does, on the other hand, make sense to use base 10 units in a base 10 system such as the metric system, and likewise base 2 units in a binary system.

Re:Why is this still a discussion?

[clem.dickey]

Ever try to design a processor that uses 1000 byte pages? Good freaking luck.

No, but these guys [wikipedia.org] did.

Re:Zonk

[DaveWick79]

They still don't have it right because the gigabyte is 10^9 bytes, not 10^6. A gibibyte is 2^30 bytes, not 2^20 bytes.

Pick your poison

[SIGBUS]

My worst experience has been with Western Digital. I have a large stack of dead WD drives where I work; one of them went blooey just after its one-year warranty expired.

More recently at home, one of four Samsung 120GB SATA drives in a Linux software RAID-5 array bit the dust. Hmm... just after the three-year warranty expired. What a coincidence! Fortunately, the array kept on chugging along in degraded mode without skipping a beat, and I quickly took the opportunity to back it up - restoring the contents onto a 3x500GB RAID-5 array of Seagate drives.

For huge storage, I'll stick with three or four disks in a RAID-5 rather than buy one giant drive. A small number of drives in RAID-5 give enough redundancy to provide time to replace a drive or migrate the data.

Re:Because giga is CLEAR FROM CONTEXT

[dgatwood]

Ah, but you don't know the size in megs even with the metric prefixes. Well, maybe to the nearest meg, but not to the nearest kilobyte. The actual amount of storage used by that file is 123, 457, 024 bytes (assuming 512-byte allocation units). You can't realistically store a file using a fraction of a disk block, and thus, it is easy to be off far enough to round to a different kilobyte value, but people working in base 10 can't see that because they're playing fast and loose with the allocation units.

In more typical filesystems with 4k allocation units, it's 123,457,536 bytes. Your metric prefixes end up off by an entire k, and can be off by as much as 3k. 1000 byte units are an arbitrary division that inherently fails to line up with the physical organization of data, and thus, make no sense. As the size of allocation units grow, the disparity between these artificial base-10 quantities and the real quantities will grow ever larger just as the disparity between the stated hard drive sizes and their base-2 size grows large now. It doesn't make sense to continue to perpetuate these silly base-10 units. They are inherently imprecise because they are not a unit into which storage can actually be divided.

Using base-10 prefixes for storage is like choosing the base unit of mass to be 98% of the mass of a proton. Only an idiot would something like that. So why, then, do we continue to try to force the inherently imprecise use of base-10 quantities in storage? :-)

Re:Zonk

[Hal_Porter]

It should read: Hard disk manufacturing company marketing departments define one megabyte of hard drive space as 1,000,000 bytes: 10^6 bytes. Fucking reality calculates one megabyte as 2^20 bytes, or 1,048,576 bytes.

Actually, megabyte [wikipedia.org] has always meant 10^6 bytes. The IEC have defined new prefixes for binary, e.g. Mebibyte [wikipedia.org] for 2^20.

Re:kibibytes & mibimeters

[Alioth]

AC> Pilots are just nuts, with their knots and feet.

While the use of feet might be dubious, there's a good rationale for using knots. One minute of latitude is one nautical mile long. This makes it very easy to do quick measurements on a chart while in-flight - since however you have a chart folded, you'll be able to see a longitude line (which has latitude tick marks as it goes up the page).