CF larger than 2gb

madman

New member
Messages
5
Reaction score
0
Location
CA, US
Has anyone used CF's larger than 2gb on the 1D? ANy problems?
Can someone confirm if the 1D supports CF's greater than 2gb?

thanks.
 
1D only supports 2gb. It will reformat a 4gb card to be able to only read 2gb.
Has anyone used CF's larger than 2gb on the 1D? ANy problems?
Can someone confirm if the 1D supports CF's greater than 2gb?

thanks.
 
CF card > 2G requires FAT32 support which is only present in 1Ds or newer dSLRs (10D, D300) and is not supported in D30, D60 or 1D.
Has anyone used CF's larger than 2gb on the 1D? ANy problems?
Can someone confirm if the 1D supports CF's greater than 2gb?

thanks.
 
Can this be a software enhancement or would it have to be a
hardware enhancement?
There's no hardware difference at all. The file system (weather FAT16, FAT32, MacOS, NTFS, or Linux) is purely a function of the software.

Kind of like Nikon. The original D100 formware (version 1.0 and 1.01) could use a FAT32 CF or microdrive, if you formatted it in a PC first. Field reports say the 2.0 formware will automatically FAT32 format any CF or MD over 2 gig.

There are hardware limits, but they're up arounf 8 terrabytes. With storage capacity doubling every 19 months, you don't ahve to worry about that till 2019.

--
Ciao!

Joe

http://www.swissarmyfork.com
 
The write performance of the 4gb CF's is notably worse than the comparable

Nill
~~
http://www.toulme.net
Can this be a software enhancement or would it have to be a
hardware enhancement?
There's no hardware difference at all. The file system (weather
FAT16, FAT32, MacOS, NTFS, or Linux) is purely a function of the
software.

Kind of like Nikon. The original D100 formware (version 1.0 and
1.01) could use a FAT32 CF or microdrive, if you formatted it in a
PC first. Field reports say the 2.0 formware will automatically
FAT32 format any CF or MD over 2 gig.

There are hardware limits, but they're up arounf 8 terrabytes. With
storage capacity doubling every 19 months, you don't ahve to worry
about that till 2019.

--
Ciao!

Joe

http://www.swissarmyfork.com
 
CF card > 2G requires FAT32 support which is only present in 1Ds or
newer dSLRs (10D, D300) and is not supported in D30, D60 or 1D.
If you format the CF card using A:64K (assuming WinNT/2000/XP/2003), then FAT16 will be able to support 4GB.

Now... The real question is: Is the 1D capable of handling 64K cluster size? (The D60 almost supported it G )

--
Rune, http://runesbike.com/
 
Keep in mind also that, based on several ads I saw over the weekend, that 1 gig cards are currently the best priced ones on a cost per meg basis. With a 1D that will give you almost 200 RAW images or 400 highest quality large JPGs, so maybe that is not so horrible, especially compared to 36 of years past.
Nill
~~
http://www.toulme.net
Can this be a software enhancement or would it have to be a
hardware enhancement?
There's no hardware difference at all. The file system (weather
FAT16, FAT32, MacOS, NTFS, or Linux) is purely a function of the
software.

Kind of like Nikon. The original D100 formware (version 1.0 and
1.01) could use a FAT32 CF or microdrive, if you formatted it in a
PC first. Field reports say the 2.0 formware will automatically
FAT32 format any CF or MD over 2 gig.

There are hardware limits, but they're up arounf 8 terrabytes. With
storage capacity doubling every 19 months, you don't ahve to worry
about that till 2019.

--
Ciao!

Joe

http://www.swissarmyfork.com
 
The write performance of the 4gb CF's is notably worse than the
comparable
performance is slower than FAT16's.
I'd bet it's not particularly noticible. These are CF cards, not hard drives (well, except for microdrives, which I won't normally touch, anyway).

A 2 gig FAT16 drive has 32k clusters. A 2 gig to 8 gig FAT32 drive has 4k clusters.

So, if you're writing a 10 meg file (for example) you're going to have to hit the FAT table 312 times on the FAT16 system, and 2,500 times on the FAT32 system.

Now, if the software in the camera were as bad as it's possible to be, the FAT would not be cached, so that would mean 5000 actual seeks and writes of small FAT entries, interleaved with writing the data (which should be mostly contiguous, assuming the drive isn't fragmented).

So, for a CF card, we're simply writing 80,000 more bytes (not k or megs or anything) of FAT enteries, which is 0.8% of the size of the data, so we should see a 0.8% performance hit.

Now, a microdrive, if the camera OS was really screwed up, would eat up about 50 seconds doing all those seeks. I've never tried FAT32 on a Canon with a microdrive, but a 1 gig microdrive on a Nikon D100 with FAT32 does run just about the same speed as a FAT16, so we'll assume the camera OS is smart enough to either cache the FAT, or preallocat the FAT and use a much smaller "extent table".

Now, if we had NTFS or Linux on the camera, both of which use full extent file systems, instead of FAT....

--
Ciao!

Joe

http://www.swissarmyfork.com
 
CF card > 2G requires FAT32 support which is only present in 1Ds or
newer dSLRs (10D, D300) and is not supported in D30, D60 or 1D.
If you format the CF card using A:64K (assuming
WinNT/2000/XP/2003), then FAT16 will be able to support 4GB.


Now... The real question is: Is the 1D capable of handling 64K
cluster size? (The D60 almost supported it G )
Ah, "almost support". That's kind of like Windows 95, 98, and ME (I don't know about FAT16 under MacOS). Supposedly something weird with the use of signed and unsigned integers in those systems, so FAT16 sizes above 2 gig are negative numbers.

--
Ciao!

Joe

http://www.swissarmyfork.com
 
You're right... it's still faster than an MD, it's just about a 10% performance hit compared to FAT16.

Nill
~~
http://www.toulme.net
The write performance of the 4gb CF's is notably worse than the
comparable
performance is slower than FAT16's.
I'd bet it's not particularly noticible. These are CF cards, not
hard drives (well, except for microdrives, which I won't normally
touch, anyway).

A 2 gig FAT16 drive has 32k clusters. A 2 gig to 8 gig FAT32 drive
has 4k clusters.

So, if you're writing a 10 meg file (for example) you're going to
have to hit the FAT table 312 times on the FAT16 system, and 2,500
times on the FAT32 system.

Now, if the software in the camera were as bad as it's possible to
be, the FAT would not be cached, so that would mean 5000 actual
seeks and writes of small FAT entries, interleaved with writing the
data (which should be mostly contiguous, assuming the drive isn't
fragmented).

So, for a CF card, we're simply writing 80,000 more bytes (not k or
megs or anything) of FAT enteries, which is 0.8% of the size of the
data, so we should see a 0.8% performance hit.

Now, a microdrive, if the camera OS was really screwed up, would
eat up about 50 seconds doing all those seeks. I've never tried
FAT32 on a Canon with a microdrive, but a 1 gig microdrive on a
Nikon D100 with FAT32 does run just about the same speed as a
FAT16, so we'll assume the camera OS is smart enough to either
cache the FAT, or preallocat the FAT and use a much smaller "extent
table".

Now, if we had NTFS or Linux on the camera, both of which use full
extent file systems, instead of FAT....

--
Ciao!

Joe

http://www.swissarmyfork.com
 
Facinating:

I've formatted a 1GB Lexar 32x card with both filesystyems and had a look if it makes a difference to the write times. I've taken 10 shots with the 1Ds in succession in RAW + large JPG mode (compression 8), manual focus and burst mode, to generate some large amount of data.

In FAT32 it took the camera 01:11 (one minute and eleven seconds) to write the entire amount down, generating 117MB of data.

In FAT16 it took the camera only 01:05 - and the amount of data occupied 116MB on the card.

Is it save to say that FAT16 is indeed slightly quicker than FAT32?
 
Hey Rune--

I've just tried that, but the answer by Windows XP was:
"The specified cluster size is too big for FAT32". Hmmmm....
Is it save to say that FAT16 is indeed slightly quicker than FAT32?
To me it looks like you're benchmarking the difference in cluster
sizes...

Try:
FORMAT x: FS:FAT32 A:32K
where x: is your CF card

Compare that result with your FAT16 result...

--
Rune, http://runesbike.com/
 
You're right... it's still faster than an MD, it's just about a 10%
performance hit compared to FAT16.
That's pretty strange. I wouldn't expect a FAT32 system to be 10% slower. If I had to guess why, I'd say one of two things. Either the CF cards really do have some sort of seek time (i.e. they have to dump an internal cache when moving from sector to sector), or the driver in the camera is doing something similar, caching entire 4k clusters, so it has to read and dump an entire 4k cluster when it updates a 32 byte FAT table entry.

In either case, it's something "broken" that should get fixed.

--
Ciao!

Joe

http://www.swissarmyfork.com
 
Facinating:

I've formatted a 1GB Lexar 32x card with both filesystyems and had
a look if it makes a difference to the write times. I've taken 10
shots with the 1Ds in succession in RAW + large JPG mode
(compression 8), manual focus and burst mode, to generate some
large amount of data.

In FAT32 it took the camera 01:11 (one minute and eleven seconds)
to write the entire amount down, generating 117MB of data.

In FAT16 it took the camera only 01:05 - and the amount of data
occupied 116MB on the card.

Is it save to say that FAT16 is indeed slightly quicker than FAT32?
That agrees exactly with Nill's 10% figure. I just can't figure out where that huge difference is coming from. It shouldn't take 10% longer to write 1% more data.

I'm guessing it has something to do with caching sectore (at the card hardware level) or clusters (at the camera OS level) so caches are being dumped when moving from the actual data area to the FAT table or directory entries.

--
Ciao!

Joe

http://www.swissarmyfork.com
 
That's weird. In FAT32, the maximun cluster size is tied to the sector size. It will use up to 128 sectors for a cluster. So the only way that 32k clusters would be too big is if you only had 128 byte sectors (128 sectors of 128 bytes is 16k).

CF cards typically have at least 256 byte sectors, so you should be good for clusters to 64k.
I've just tried that, but the answer by Windows XP was:
"The specified cluster size is too big for FAT32". Hmmmm....
Is it save to say that FAT16 is indeed slightly quicker than FAT32?
To me it looks like you're benchmarking the difference in cluster
sizes...

Try:
FORMAT x: FS:FAT32 A:32K
where x: is your CF card

Compare that result with your FAT16 result...

--
Rune, http://runesbike.com/
--
Ciao!

Joe

http://www.swissarmyfork.com
 
See this article re the Lexar 4gb card:

http://www.infoworld.com/article/03/08/05/HNlexar_1.html

The reference to the 1D presumably should be to the 1Ds. The interesting part though is the point about "writing with FAT32 takes longer," which would seem to explain the poorer results in Rob's test ( http://www.robgalbraith.com/bins/multi_page.asp?cid=6007-6010 ) as compared to the comparable 1gb and 2gb Lexars.

There's also some discussion of these cards, FAT32 etc. here:

http://www.robgalbraith.com/ubbthreads/showflat.php?Cat=&Board=UBB18&Number=156574&fpart=&PHPSESSID=

...including this statement from Chuck Westfall:

"EOS-1Ds and 10D are FAT32, but all other EOS digitals are FAT16. That means you could use a 4GB card in a 1D, D60 or D30 but those cameras would only "see" 2GB of space. There is no indication that this status will ever be changed via firmware update for models released before the 1Ds."

Nill
~~
http://www.toulme.net
That's pretty strange. I wouldn't expect a FAT32 system to be 10%
slower. If I had to guess why, I'd say one of two things. Either
the CF cards really do have some sort of seek time (i.e. they have
to dump an internal cache when moving from sector to sector), or
the driver in the camera is doing something similar, caching entire
4k clusters, so it has to read and dump an entire 4k cluster when
it updates a 32 byte FAT table entry.

In either case, it's something "broken" that should get fixed.
 

Keyboard shortcuts

Back
Top