Actual hard drive sizes....

[brad]

howdy all.....

What's the deal with hard drive sizes ???

Yesterday, I set up 2 x SEAGATE 80gig SATA hard disks on RAID with my INTEL
865 PERLL board....
After setup, I`m noticing that I`m missing 11gig out of my supposedly 160gig
raid setup.....
Total space available on the freshly NTFS formatted drives was 149gig.....

Originally , I was running only one drive for a few weeks which was missing
about 6gig. of space....
I just bought this other one and set them up together on raid and now it`s
missing 11gig...

Seems to me I`m getting shafted on what I`m actually paying for....

I did use a Western Digital 80gig. drive a while ago on my previous machine
and that was no where near as bad as these.....

also , the Seagate`s are quite noisy compared to the western digital....
So much for improvements / upgrades..........

Thanks guys....
Brad.

 

[Bastet]

howdy all.....

What's the deal with hard drive sizes ???

Yesterday, I set up 2 x SEAGATE 80gig SATA hard disks on RAID with
my INTEL 865 PERLL board....
After setup, I`m noticing that I`m missing 11gig out of my supposedly
160gig raid setup.....
Total space available on the freshly NTFS formatted drives was
149gig.....

Originally , I was running only one drive for a few weeks which was
missing about 6gig. of space....

That would be correct.

I just bought this other one and set them up together on raid and now
it`s missing 11gig...

That's 11GB over *TWO* drives not one.

Seems to me I`m getting shafted on what I`m actually paying for....

Suggest you learn the difference between binary and decimal - you paid for
160 *decimal* GBs and you've got 160 *decimal* GBs. You're not 'missing'
anything.

I did use a Western Digital 80gig. drive a while ago on my previous
machine and that was no where near as bad as these.....

And you didn't notice that was 'missing' 6GB?

In what way - 'missing' space?! Well it's proportional - you 'lose' 11GB of
a 160GB, ergo you 'lose' 5.5GB on an 80GB.

 

[MrToad]

howdy all.....

What's the deal with hard drive sizes ???

Yesterday, I set up 2 x SEAGATE 80gig SATA hard disks on RAID with my
INTEL 865 PERLL board....
After setup, I`m noticing that I`m missing 11gig out of my supposedly
160gig raid setup.....
Total space available on the freshly NTFS formatted drives was
149gig.....

Originally , I was running only one drive for a few weeks which was
missing about 6gig. of space....
I just bought this other one and set them up together on raid and now
it`s missing 11gig...

Seems to me I`m getting shafted on what I`m actually paying for....

I did use a Western Digital 80gig. drive a while ago on my previous
machine and that was no where near as bad as these.....

also , the Seagate`s are quite noisy compared to the western
digital.... So much for improvements / upgrades..........

Thanks guys....
Brad.

Hard-drive manufacturers usually determine the size of the drive based on
the old metric standard, where kilo = 1000. So the notations are powers
of ten (kilo, mega, giga). But some people seem to believe that kilo =
1024 (or 210) and that PC makers are using deceptive notations to promote
drive sizes. Not so. A kilobyte (KB) normally means 1000, but a kibibyte
(KiB) = 1024, according to the International Electrotechnical Commission
standard. Manufacturers are using the correct notation, as they are
calculating the hard-drive size (unformatted) using the old standard
(metric).

Cluster size also has an effect on the reported *formatted* size.
But that's another whole can 'O worms....

 

[Alien Zord]

What's the deal with hard drive sizes ???

snipped <

Right-click a drive in My Computer and click Properties. In the centre you
will see the decimal values and on the right their binary equivalents.

 

[S.Boardman]

snip!

Cluster size also has an effect on the reported *formatted* size.
But that's another whole can 'O worms....

Can you expand/ help here? I have Win98 SE, and want to change my 2 x 20Gb
RAID 0 (striped) to RAID 1 (mirrored) using 2 x 80Gb.
Currently I have 2 partitions. After formatting with (the patched) FDISK or
PatitionMagic 6, will I lose to much space if only keep two partitions? Or
should do it into thirds?

 

[DaveW]

When you FORMAT harddrives you lose space to the formatting. Losing 11 GB
out of 160 GB sounds about right.

 

[MrToad]

snip!
Can you expand/ help here? I have Win98 SE, and want to change my 2 x
20Gb RAID 0 (striped) to RAID 1 (mirrored) using 2 x 80Gb.
Currently I have 2 partitions. After formatting with (the patched)
FDISK or PatitionMagic 6, will I lose to much space if only keep two
partitions? Or should do it into thirds?

IIRC, Partition Magic will let you specify the cluster sizes(?), [correct
me if I wrong], there for, you can keep your two large partitions *if*
you keep your cluster sizes down to 4K, that way you'll lose the least
amount of drive space.

With FAT32, after 8GB, the cluster size increases to 8KB, 16GB is 16KB
and anything larger than 32GB the cluster size is 32KB!!!

IF PMagic 6 doesn't allow you to keep the 4KB cluster size for your
drives, I would suggest upgrading to Windows XP Pro if possible, the
cluster sizes stay at 4KB up tp 2TB, [tb=terabytes].

 

[kony]

IIRC, Partition Magic will let you specify the cluster sizes(?), [correct
me if I wrong], there for, you can keep your two large partitions *if*
you keep your cluster sizes down to 4K, that way you'll lose the least
amount of drive space.

With FAT32, after 8GB, the cluster size increases to 8KB, 16GB is 16KB
and anything larger than 32GB the cluster size is 32KB!!!

IF PMagic 6 doesn't allow you to keep the 4KB cluster size for your
drives, I would suggest upgrading to Windows XP Pro if possible, the
cluster sizes stay at 4KB up tp 2TB, [tb=terabytes].

The overhead from cluster size differences isn't all that significant
these days with drives having dozens of GB capacity.

I would advise the opposite, to use PM to INCREASE the cluster size to
32K, which slightly improves performance but also decreases
fragmentation, another performance increase.


Dave

 

[xyzzy]

Hard-drive manufacturers usually determine the size of the drive based on
the old metric standard, where kilo = 1000. So the notations are powers
of ten (kilo, mega, giga). But some people

Yes, the rest of computer industry, when talking about storage. How
many bytes are there, for instance, in 128 MB of RAM?

seem to believe that kilo =
1024 (or 210) and that PC makers are using deceptive notations to promote
drive sizes. Not so.

Not true. Rather, it probably started out as cheating. But since every
HD manufacturer is now using the metric Kilo, it's not unfair cheating
anymore, just an overall inflation compared to, say, RAM. Of course
manufacturing constraints are different between RAMs and HDs.

A kilobyte (KB) normally means 1000

No, it doesn't. Common usage has been, for several decades, that
KiloByte = 1024 bytes. Granted, it's not SI, but that doesn't mean
it's wrong and more than imperial units are wrong. The two kilos are
homonyms, one isn't more or less correct.

but a kibibyte
(KiB) = 1024, according to the International Electrotechnical Commission
standard. Manufacturers are using the correct notation, as they are
calculating the hard-drive size (unformatted) using the old standard
(metric).

All fine and dandy but it remains to be seen if that nomenclature will
be widely adopted. Does MS report file sizes in KB or KiB? Like it or
not, that's the deciding factor.

 

[MrToad]

IIRC, Partition Magic will let you specify the cluster sizes(?),
[correct me if I wrong], there for, you can keep your two large
partitions *if* you keep your cluster sizes down to 4K, that way
you'll lose the least amount of drive space.

With FAT32, after 8GB, the cluster size increases to 8KB, 16GB is 16KB
and anything larger than 32GB the cluster size is 32KB!!!

IF PMagic 6 doesn't allow you to keep the 4KB cluster size for your
drives, I would suggest upgrading to Windows XP Pro if possible, the
cluster sizes stay at 4KB up tp 2TB, [tb=terabytes].

The overhead from cluster size differences isn't all that significant
these days with drives having dozens of GB capacity.

I would advise the opposite, to use PM to INCREASE the cluster size to
32K, which slightly improves performance but also decreases
fragmentation, another performance increase.


Dave

Actually it does make a big difference.
Lets take a 1KB shortcut for example, with FAT32 on a drive bigger than
32GB, it will consume 1KB of disk space with 31KB wasted.
Sure, you may have tons of disk space, but the OP wanted to know how to
minimize the loss. Keeping cluster sizes down is an economical way to do
just that.
Plus, with using the NT filing system, [NTFS], fragmentatioin doesn't
make as big as an impact as it does with the FAT-based filing structures.
Have a look at this:
http://www.ntfs.com/ntfs_vs_fat.htm

 

[kony]

The overhead from cluster size differences isn't all that significant
these days with drives having dozens of GB capacity.

I would advise the opposite, to use PM to INCREASE the cluster size to
32K, which slightly improves performance but also decreases
fragmentation, another performance increase.


Dave

Actually it does make a big difference.
Lets take a 1KB shortcut for example, with FAT32 on a drive bigger than
32GB, it will consume 1KB of disk space with 31KB wasted.
Sure, you may have tons of disk space, but the OP wanted to know how to
minimize the loss. Keeping cluster sizes down is an economical way to do
just that.
Plus, with using the NT filing system, [NTFS], fragmentatioin doesn't
make as big as an impact as it does with the FAT-based filing structures.
Have a look at this:
http://www.ntfs.com/ntfs_vs_fat.htm

Well, big difference and significant difference can be two different
things. I'd expect that most people don't have several thousands of
tiny files, that once they're using a few dozen GB of space they have
a lot of larger files... after all it'd take >32,000 shortcuts to use
up 1 GB of HDD space using 32K clusters.



Dave.

 

[MrToad]

Yes, the rest of computer industry, when talking about storage. How
many bytes are there, for instance, in 128 MB of RAM?

Ram manufacturers use Kibibytes.

Not true. Rather, it probably started out as cheating. But since every
HD manufacturer is now using the metric Kilo, it's not unfair cheating
anymore, just an overall inflation compared to, say, RAM. Of course
manufacturing constraints are different between RAMs and HDs.

See below.

No, it doesn't. Common usage has been, for several decades, that
KiloByte = 1024 bytes. Granted, it's not SI, but that doesn't mean
it's wrong and more than imperial units are wrong. The two kilos are
homonyms, one isn't more or less correct.

The metric system is still based on 10.

All fine and dandy but it remains to be seen if that nomenclature will
be widely adopted. Does MS report file sizes in KB or KiB? Like it or
not, that's the deciding factor.

I believe MS reports both, KiB in Windows and KB in [old] DOS.
The KiB is already in "common usage", people just get confused when the
HDD manufacturer's use the metric Kilobyte and Windows report KiB.

All in all, it makes no difference, HDD manufacturer's aren't goimg to
change the way they report disk sizes because they believe they are
correct, [and they are, in metric terms], what needs to change, IMO, is
the way Windows reports disk space, abandon the KiB, embrace the KB!
;o]

 

[MrToad]

The overhead from cluster size differences isn't all that
significant these days with drives having dozens of GB capacity.

I would advise the opposite, to use PM to INCREASE the cluster size
to 32K, which slightly improves performance but also decreases
fragmentation, another performance increase.


Dave

Actually it does make a big difference.
Lets take a 1KB shortcut for example, with FAT32 on a drive bigger
than 32GB, it will consume 1KB of disk space with 31KB wasted.
Sure, you may have tons of disk space, but the OP wanted to know how
to minimize the loss. Keeping cluster sizes down is an economical way
to do just that.
Plus, with using the NT filing system, [NTFS], fragmentatioin doesn't
make as big as an impact as it does with the FAT-based filing
structures. Have a look at this:
http://www.ntfs.com/ntfs_vs_fat.htm

Well, big difference and significant difference can be two different
things. I'd expect that most people don't have several thousands of
tiny files, that once they're using a few dozen GB of space they have
a lot of larger files... after all it'd take >32,000 shortcuts to use
up 1 GB of HDD space using 32K clusters.



Dave.

You'd be surprised at how many <32K files are present on your system,
shortcuts are just an example.

Did you look at the other benefits from using NTFS on the link I gave,
wasted disk space is a minor problem compared to the other benefits
gained by changing to WinXP Pro...

 

[kony]

The overhead from cluster size differences isn't all that
significant these days with drives having dozens of GB capacity.

I would advise the opposite, to use PM to INCREASE the cluster size
to 32K, which slightly improves performance but also decreases
fragmentation, another performance increase.


Dave

Actually it does make a big difference.
Lets take a 1KB shortcut for example, with FAT32 on a drive bigger
than 32GB, it will consume 1KB of disk space with 31KB wasted.
Sure, you may have tons of disk space, but the OP wanted to know how
to minimize the loss. Keeping cluster sizes down is an economical way
to do just that.
Plus, with using the NT filing system, [NTFS], fragmentatioin doesn't
make as big as an impact as it does with the FAT-based filing
structures. Have a look at this:
http://www.ntfs.com/ntfs_vs_fat.htm

Well, big difference and significant difference can be two different
things. I'd expect that most people don't have several thousands of
tiny files, that once they're using a few dozen GB of space they have
a lot of larger files... after all it'd take >32,000 shortcuts to use
up 1 GB of HDD space using 32K clusters.



Dave.

You'd be surprised at how many <32K files are present on your system,
shortcuts are just an example.

Did you look at the other benefits from using NTFS on the link I gave,
wasted disk space is a minor problem compared to the other benefits
gained by changing to WinXP Pro...

I have WinXP Pro, don't care for it at all except for it's much
greater stability... unfortunately stability is pretty important,
else I wouldn't have WinXP at all. On the other hand IF you know
how to set up Win9x and use decent drivers it's perfect for many
dedicated-purpose systems, an emphasis on "many" since I'm practically
tripping over Win9x licenses.


Dave

 

[V W Wall]

The overhead from cluster size differences isn't all that
significant these days with drives having dozens of GB capacity.

I would advise the opposite, to use PM to INCREASE the cluster size
to 32K, which slightly improves performance but also decreases
fragmentation, another performance increase.


Dave

Actually it does make a big difference.
Lets take a 1KB shortcut for example, with FAT32 on a drive bigger
than 32GB, it will consume 1KB of disk space with 31KB wasted.
Sure, you may have tons of disk space, but the OP wanted to know how
to minimize the loss. Keeping cluster sizes down is an economical way
to do just that.
Plus, with using the NT filing system, [NTFS], fragmentatioin doesn't
make as big as an impact as it does with the FAT-based filing
structures. Have a look at this:
http://www.ntfs.com/ntfs_vs_fat.htm

Well, big difference and significant difference can be two different
things. I'd expect that most people don't have several thousands of
tiny files, that once they're using a few dozen GB of space they have
a lot of larger files... after all it'd take >32,000 shortcuts to use
up 1 GB of HDD space using 32K clusters.



Dave.

You'd be surprised at how many <32K files are present on your system,
shortcuts are just an example.

Did you look at the other benefits from using NTFS on the link I gave,
wasted disk space is a minor problem compared to the other benefits
gained by changing to WinXP Pro...

I just looked at a ~30GB partition on my 80GB HD using FAT32
(Chosen as the maximum FAT32 partition with cluster size at 32K).

File sizes: 0-16K 100 2.5% 16K-32K 80 2.0%

Most of the ~100 files are shortcuts, organized into 10 folders whose shortcuts
appear on the desktop together with 3 .pif shortcuts. I don't have any
MP3 files or many .jpg files, so storage is probably biased towards smaller
files. This partition still has ~25GB free, so files <32K will be a still
smaller percent in the future.

I can't comment about the virtues of NTSF, but I use FAT to be able to access
files from DOS. Yeah, I know I read NTSF files from Linux, but modifying them
from there, (or DOS), is still risky.

With the bloatware now common to Windows, it's hard to find files <32K! ;-(

Virg Wall

 

[CBFalconer]

.... snip about wastage versus allocation quantum ...

You'd be surprised at how many <32K files are present on your
system, shortcuts are just an example.

Did you look at the other benefits from using NTFS on the link
I gave, wasted disk space is a minor problem compared to the
other benefits gained by changing to WinXP Pro...

The count of small files doesn't matter. The average waste is 1/2
an allocation unit per file. All you have to do is count the
files, ignoring size.

 

[V W Wall]

... snip about wastage versus allocation quantum ...

The count of small files doesn't matter. The average waste is 1/2
an allocation unit per file. All you have to do is count the
files, ignoring size.

That's quite true, but it's like saying almost half of your friends have
below average IQ! It depends on who your friends are, and what "average"
your using. It also depends on the mix of files as to how much,
(percentage wise), the waste is.

I have 38761 files smaller than the 32K cluster size, but they take up
only about 4% of the ~5GB now used in my 30GB FAT32 partition.

A larger cluster makes defrag go faster.

Virg Wall

 

[xyzzy]

xyzzy <[email protected]> Ran in the back door and shouted

Ram manufacturers use Kibibytes.

Not really. See:

http://www.micron.com/products/dram/

They use KB, MB, GB.. etc as in the good old days with K=1024. Very
few people outside of academia write KiB, MiB, GiB.

I know what you mean and you are correct in your interpretation but it
still sounds misleading and I don't think you can say Micron is wrong.

My position is that Micron means MB and writes MB. You are saying that
it means MiB but writes MB and hence is wrong.

Note that I don't have a beef with IEC, it's just that the K/k=1024 de
facto standard has been in place for a long long time before 1999.

The metric system is still based on 10.

True but doesn't matter so long as nobody erroneously uses the metric
system interpretation. It's an "accident" that both words have the
same spelling. You tell what it means from the context - they are
homonyms!

I believe MS reports both, KiB in Windows and KB in [old] DOS.

See first paragraphs. DOS was bit of a mess though.

 

[xyzzy]

The count of small files doesn't matter. The average waste is 1/2
an allocation unit per file. All you have to do is count the
files, ignoring size.

Close but not quite. The average wasted space will likely be somewhat
larger than

1/2 * (allocation unit size - 1)

for a natural distribution of filesizes.

 

[S.Boardman]

IIRC, Partition Magic will let you specify the cluster sizes(?), [correct
me if I wrong], there for, you can keep your two large partitions *if*
you keep your cluster sizes down to 4K, that way you'll lose the least
amount of drive space.

With FAT32, after 8GB, the cluster size increases to 8KB, 16GB is 16KB
and anything larger than 32GB the cluster size is 32KB!!!

IF PMagic 6 doesn't allow you to keep the 4KB cluster size for your
drives, I would suggest upgrading to Windows XP Pro if possible, the
cluster sizes stay at 4KB up tp 2TB, [tb=terabytes].

The overhead from cluster size differences isn't all that significant
these days with drives having dozens of GB capacity.

I would advise the opposite, to use PM to INCREASE the cluster size to
32K, which slightly improves performance but also decreases
fragmentation, another performance increase.

OK, I take the 80 Gb, tell it to do 4 Gb for Primary partition, and the rest
as one extended, with the cluster size as 32KB?