If you don't see yourself using continuous shutter bursts, you don't need the extra speed, but if you shoot little kids (Eliane) where you might want to just hold the button down to catch fleeting facial expressions, the speedier cards might help you get the shot, whereas the slower card might cause your camera to stop shooting.
John :
Natural selection is responsible for every living thing that exists.
D3s, D500, D5300, and way more glass than the wife knows about.
There is only about an 8% difference in write speeds between the fastest tested card, a Sandisk Extreme IV 2 Gig card, and the Sandisk Ultra II 2 Gig card.
You are much better served in purchasing more Ultra II cards and not trying to completely fill the cards. The last 15 percent of memory cards tend to write much more slowly than the first 85 percent.
The write speeds (in camera) are fairly the same. However as you move up to bigger cards, like the 4GB, the read speeds are much better. I even bought one of the SanDisk readers and it dumps a 2GB Ultra II in under 2 minutes to my Mac.
"Don't ask me what I think of you, I might not give the answer that you want me to. Oh well."
-Fleetwood Mac
There is a difference it when comes to reading the cards. If you don't use dng-conversion or something similar while importing the photos, and the size of card you like to use is 2 GB or more, I'd say go for the faster. Unless you don't mind reason for a coffee break
And if memory serves, the E III cards are advertised as more reliable as well. Can't say on that meself though, both my UII and EIII cards are perfectly fine.
I just came back from a trip to the Antarctic, and although the conditions were no where near "extreme," many of the folks with cheaper cards (no idea which brand or model) were left dangling in the cold, unable to write pictures because the cards "seized" or would write very, very slowly (for those with bigger, rapid-fire cameras). A Canon shooter also cracked in half an el cheapo UV filter on one of her L lenses (lesson learned there!)
And for sure my "faster" cards dump quicker on to my laptop through a card reader than the slower card I first had. YMMV
VI
dgrin.com - making my best shots even better since 2006.
I have a Canon 40D and have been using SanDisk Ultra II CF cards. Would there be any significant advantage to using extreme III cards?
In camera speed improvement is 10% (about .7MB/s faster)
The Card->PC speed will jump from 9MB/s to 20MB/s with
a fast cardreader (such as the SanDisk Extrememate CF).
“To consult the rules of composition before making a picture is a little like consulting the law of gravitation before going for a walk.”
― Edward Weston
I use an Ultra II and an Extreme III in a 30D. In the camera there is no noticable difference, the camera writes at ~6.4 Mb/sec with both cards. I do however notice a difference downloading with a card reader, the Extreme III is considerably faster. They claim top speed of 10Mb/s with the U-II and 20Mb/s with the E-III.
The speed test for the 30D is at Rob Galbraith, there isn't a test published for the 40D yet.
"The point in life isn't to arrive at our final destination well preserved and in pristine condition, but rather to slide in sideways yelling.....Holy cow, what a ride."
The last 15 percent of memory cards tend to write much more slowly than the first 85 percent.
Ziggy,
I'd be interested in knowing more about this. I could see that happening on a card that's been fragmented due to having images deleted. When the card is nearing full, there might be some linear searching in a directory table for free blocks. However, I don't see how that could happen on a freshly formatted card. The time to allocate a new image should be fixed right up until the card is full because all the free blocks are contiguous. Any clues here?
I'd be interested in knowing more about this. I could see that happening on a card that's been fragmented due to having images deleted. When the card is nearing full, there might be some linear searching in a directory table for free blocks. However, I don't see how that could happen on a freshly formatted card. The time to allocate a new image should be fixed right up until the card is full because all the free blocks are contiguous. Any clues here?
Thanks,
-joel
This slowing down happens only when all files are written to the same
directory. The (fat) filesystem holds list of files for each directory and if
it is growing overly long the time increases to append a new file name
to that list (using linear seach). Hence the slowing down.
“To consult the rules of composition before making a picture is a little like consulting the law of gravitation before going for a walk.”
― Edward Weston
This slowing down happens only when all files are written to the same
directory. The (fat) filesystem holds list of files for each directory and if
it is growing overly long the time increases to append a new file name
to that list (using linear seach). Hence the slowing down.
That could be a large part of it as I first noticed the slowing shooting football with a Canon 1D MKII, which does use a single file directory.
With a slower Kingston 2 Gig card, I could "feel" the slowdown also on the Canon XT/350D. With the faster SanDisk cards I don't notice at all on the XT and it's much less with the faster card on the MKII, but still there.
I am not the only one to notice but I haven't found a formal study or review that documents the phenomenon.
This slowing down happens only when all files are written to the same
directory. The (fat) filesystem holds list of files for each directory and if
it is growing overly long the time increases to append a new file name
to that list (using linear seach). Hence the slowing down.
If that's true, then the slowing-down isn't due to the percentage of the card being full, but purely a function of the number of files in the directory. So if you see slowing at 75% of a 2G card, you're going to see it at 25% of an 8G card. Is that what you're saying?
I also find it hard to believe that the camera does a linear search through the entire directory for every image, when all it has to do is keep a pointer to the end of the directory in its memory.
If that's true, then the slowing-down isn't due to the percentage of the card being full, but purely a function of the number of files in the directory. So if you see slowing at 75% of a 2G card, you're going to see it at 25% of an 8G card. Is that what you're saying?
I also find it hard to believe that the camera does a linear search through the entire directory for every image, when all it has to do is keep a pointer to the end of the directory in its memory.
Good points. I'm not sure why it happens and I don't have any cards larger than 2 Gig, so I can't test that hypothesis.
I first noticed the effect when I would fill up the buffer on the MKII. The buffer is not that smart and makes you wait until it clears before you can shoot again. (More modern cameras have a smart buffer that allows you to shoot again when the buffer empties even fractionally.)
I can't tell you how nerve wracking it is waiting on the buffer when there is still action on the field.
When the card would get too slow, I'd swap it out with a new card and the speed improvement with the empty card was very noticeable.
Oh, I know all about the full-buffer, miss-the-action routine.
So, I scanned the FAT and CF specifications, and have a theory why the card might slow down when it's getting full. I'll be as brief as I can to describe this.
CF cards are allowed to have bad blocks, and part of formatting is to map out these bad blocks, and to create fixed-sized clusters out the good blocks. Clusters are smaller than images, so images span multiple clusters. I'm assuming the format routine sets up the file allocation table (FAT) with clusters arranged from contiguous memory first, and then non-contiguous clusters caused by fragmentation due to bad blocks last in the FAT. I'm further assuming the camera hardware optimizes writes to continguous memory. The slow-down happens when the camera gets down to the end of the FAT where the fragmented clusters live, and has to do more work to write the image into the scattered memory.
So according to the above theory, if you had a perfect CF with no bad blocks, it wouldn't slow down. There are ways to test for bad blocks, so an experiment could be conducted between a known perfect card and a non-perfect card to see if there's a difference in speed when they're almost full. Then we'd know if my theory is full of crap, or what.
Well all I know is my experience: with a standard Sandisk 1GB CF, I can cause my 350D to pause when writing to CF when doing rapid,multi-frame (RAW) sports shoots. But with the Sandisk Ultra II CF, I notice no delay when shooting multi-frames (RAW). I can basically shoot unlimited frames. I notice no difference in JPEG. Otherwise, I dont think an Extreme III will improve matters much.
Comments
Eric
It's better to be hated for who you are than to be loved for who you're not.
http://photosbyeric.smugmug.com
Thanks, that's what I thought, too.
Comments and constructive critique always welcome!
Elaine Heasley Photography
Natural selection is responsible for every living thing that exists.
D3s, D500, D5300, and way more glass than the wife knows about.
Elaine,
According to this page (Canon 30D, the 40D has not yet been tested):
http://www.robgalbraith.com/bins/multi_page.asp?cid=6007-8478
There is only about an 8% difference in write speeds between the fastest tested card, a Sandisk Extreme IV 2 Gig card, and the Sandisk Ultra II 2 Gig card.
You are much better served in purchasing more Ultra II cards and not trying to completely fill the cards. The last 15 percent of memory cards tend to write much more slowly than the first 85 percent.
Moderator of the Cameras and Accessories forums
-Fleetwood Mac
And if memory serves, the E III cards are advertised as more reliable as well. Can't say on that meself though, both my UII and EIII cards are perfectly fine.
http://pyryekholm.kuvat.fi/
Wedding Photographer Glasgow | Scotland
SWPP Pet Photographer of the Year 2010
Follow us on Facebook - Click Here
I had no idea - thanks for the info Ziggy!
http://www.jonathanswinton.com
http://www.swintoncounseling.com
And for sure my "faster" cards dump quicker on to my laptop through a card reader than the slower card I first had. YMMV
VI
In camera speed improvement is 10% (about .7MB/s faster)
The Card->PC speed will jump from 9MB/s to 20MB/s with
a fast cardreader (such as the SanDisk Extrememate CF).
― Edward Weston
www.ivarborst.nl & smugmug
The speed test for the 30D is at Rob Galbraith, there isn't a test published for the 40D yet.
www.achambersphoto.com
"The point in life isn't to arrive at our final destination well preserved and in pristine condition, but rather to slide in sideways yelling.....Holy cow, what a ride."
I'd be interested in knowing more about this. I could see that happening on a card that's been fragmented due to having images deleted. When the card is nearing full, there might be some linear searching in a directory table for free blocks. However, I don't see how that could happen on a freshly formatted card. The time to allocate a new image should be fixed right up until the card is full because all the free blocks are contiguous. Any clues here?
Thanks,
-joel
Link to my Smugmug site
This slowing down happens only when all files are written to the same
directory. The (fat) filesystem holds list of files for each directory and if
it is growing overly long the time increases to append a new file name
to that list (using linear seach). Hence the slowing down.
― Edward Weston
That could be a large part of it as I first noticed the slowing shooting football with a Canon 1D MKII, which does use a single file directory.
With a slower Kingston 2 Gig card, I could "feel" the slowdown also on the Canon XT/350D. With the faster SanDisk cards I don't notice at all on the XT and it's much less with the faster card on the MKII, but still there.
I am not the only one to notice but I haven't found a formal study or review that documents the phenomenon.
http://64.233.167.104/search?q=cache:uE4tkSb3oeYJ:www.tomshardware.com/forum/28494-5-write-rate-slows-large-card-fills+%2B%22compact+flash%22+%2B%22as+it+fills%22&hl=en&ct=clnk&cd=6&gl=us
Moderator of the Cameras and Accessories forums
If that's true, then the slowing-down isn't due to the percentage of the card being full, but purely a function of the number of files in the directory. So if you see slowing at 75% of a 2G card, you're going to see it at 25% of an 8G card. Is that what you're saying?
I also find it hard to believe that the camera does a linear search through the entire directory for every image, when all it has to do is keep a pointer to the end of the directory in its memory.
Link to my Smugmug site
Good points. I'm not sure why it happens and I don't have any cards larger than 2 Gig, so I can't test that hypothesis.
I first noticed the effect when I would fill up the buffer on the MKII. The buffer is not that smart and makes you wait until it clears before you can shoot again. (More modern cameras have a smart buffer that allows you to shoot again when the buffer empties even fractionally.)
I can't tell you how nerve wracking it is waiting on the buffer when there is still action on the field.
When the card would get too slow, I'd swap it out with a new card and the speed improvement with the empty card was very noticeable.
Moderator of the Cameras and Accessories forums
So, I scanned the FAT and CF specifications, and have a theory why the card might slow down when it's getting full. I'll be as brief as I can to describe this.
CF cards are allowed to have bad blocks, and part of formatting is to map out these bad blocks, and to create fixed-sized clusters out the good blocks. Clusters are smaller than images, so images span multiple clusters. I'm assuming the format routine sets up the file allocation table (FAT) with clusters arranged from contiguous memory first, and then non-contiguous clusters caused by fragmentation due to bad blocks last in the FAT. I'm further assuming the camera hardware optimizes writes to continguous memory. The slow-down happens when the camera gets down to the end of the FAT where the fragmented clusters live, and has to do more work to write the image into the scattered memory.
So according to the above theory, if you had a perfect CF with no bad blocks, it wouldn't slow down. There are ways to test for bad blocks, so an experiment could be conducted between a known perfect card and a non-perfect card to see if there's a difference in speed when they're almost full. Then we'd know if my theory is full of crap, or what.
-joel
Link to my Smugmug site