Compact Flash card speeds in Nikon Digital Cameras
I've put up this page to answer the "is it worth it getting a faster compact flash card for my Nikon digital camera" question.
Note that the times shown on this page are just measuring the effect of different CF card speeds on the total time needed to take a picture; from when you press the release until the picture is completely stored on the card. This total time includes time to read out the CCD, compress the file, ect. that cannot be improved by CF speed; this fixed overhead is probably about 3 seconds for the CP990. Only a portion or the total picture time is spent saving your picture on the card. However, for the slowest cards listed (10 seconds total), this time can be 75% of the total time, so it is worth making comparisions between different cards. The fastest cards listed take about 4 seconds to take and store a Fine mode shot on the CP990, the slowest about 10 seconds.
Does the speed REALLY matter?
Before you spend too much time looking at the speed tests, you might want to ask yourself if the speed really matters in real life. Keep in mind that the CP950 and CP990 have an internal RAM buffer that will carry you through several shots. The only time that the speed really matters is when using TIFF (and I never do any more), and when waiting for several shots to save out after you take 3 or more in a row (which does not happen that often to me either). Sure, I like to have the fastest card on the Planet as much as the next person, but I'm not sure that it makes my actual shooting any faster.
Last update: August 4, 2002
Jump to:
e-mail me some CF card timings
Don't have a stopwatch? Use this web based one (click here)
How to time your card in the CP990
To do the timing, start with an empty card. Reformat it in the camera is the best way. Half press to pre-focus (the camera must be in single autofocus mode). Time from shutter press until the hourglass goes away. If possible, use a stopwatch. The Fine mode times are short, and the time varies with the subject matter a little, so take at least three readings and average them for the Fine mode. A single shot for TIFF is OK. Don't use a lens cap on or white wall shot because these will compress a lot, use a shot out the window, for example.
Also, the camera should not be using the flash, and the exposure should be fast enough that it will not significantly affect the results (point it at something bright).
Note! The camera is showing different timings with firmware 1.1. Please make a note of your firmware version if you submit a timing. Also, with firmware 1.1 you need to time until the new "card" icon goes away, not until the hourglass is gone.
CP990 flash card timings (Firmware 1.0)
(See table below this for firmware 1.1 timings)
Following the same procedure as for the CP950 below. Keep in mind that a TIFF shot is 50% bigger for the 990 vs. the 950. Shutter starts half pressed and pre-focused. Times until the hourglass is gone with the card empty. Shown smallest to largest in the table. In some cases, I don't know what the precision of the times that were given to me, (submitted as "4 seconds" or to "nearest second" so I've included them with no decimal place, it does not mean that they are 4.0 seconds, for example)
Card Time for TIFF shot Time for Fine shot
Lexar USB 16MB "8X" 28.4 4.20 (av. of 3, Bryan's card)
Lexar USB 16MB "8X" 28.2 4.1 (Thanks Corey P)
Lexar USB 16MB "8X" 21 4 (Thanks Ed A. "to nearest second")
Lexar USB 16MB "8X" 27.9 4.1 (Thanks John T.)
Lexar USB 16MB "8X" 27 4.1 (Thanks Mark L.)
Lexar USB 16MB "8X" 29 5 (Thanks Don C. "to nearest second")
Lexar USB 16MB "8X" 27.2 4.1 (Thanks Wade S.)
Lexar USB 16MB "8X" 28.2 4.3 (Thanks Greg S.)
Nikon EC-16CF 50 46.5.3 (Thanks Jan L..)
Lexar USB 24MB "4X" 30 4 (Thanks Don C. "to nearest second")
Lexar 32MB "4X" 30.8 4.79 (av. of 3, Bryan's card)
Simple Tech 48MB 50.2 6.55 (av. of 3, Bryan's card)
Lexar 64MB 8X USB 26 4.07 (av. of 5, Bryan's card)
Kingston 64MB 43 6.0 (Thanks Wade S.)
Kodak 64MB 55 8.7 (Thanks Jan L.)
Lexar 80MB "8X" * 32 8 (Thanks Don C. "to nearest second")
Lexar USB 80MB "8X" 30.3 6.6 (Thanks Corey P)
Lexar USB 80MB "8X"**** 28.2 4.1 (Thanks Corey P)
Lexar 80MB USB "8X"*** 28.3 4.1 (Neil M. Replacement card)
Viking 80MB 37.9 14.0 (av. of 4, Thanks Ruedi S.)
Sandisk 96MB 53.2 10.0 (Thanks John T.)
Sandisk 96MB 45.5 5.9 (Thanks Vincent B.)
Sandisk 96MB 46 4.0 (Thanks Tom B.)
SCM CFA096M-13C-R2 96MB 32 4.6 (Thanks Jan L.)
Simple Tech 128MB 54 - (Thanks Matt H.)
Lexar 128MB "8X" 26.6 4.1 (Thanks John T.)
Lexar 128MB "10X" 26.2 3.8 (thanks Vincent B.)
Sandisk 128MB 50.8 8.0 (Thanks John T.)
Kingston 128MB 32 5.0 (Thanks Jan L.)
Sandisk 160MB 43 6 (Thanks Ed A.)
Lexar USB 160MB "8X" 27 4.1 (Thanks Mark L.)
Lexar 160MB "10X" ** 26.82 4.09 (av of 5, Thanks Ernie S.)
* Don (and others) report some difficulties with this card in the CP990. He has the non-"USB" version of the card. First of all, the card is 2 seconds slower than the other "8X" Lexar cards. While this makes no difference for a TIFF shot, it is about 50% longer than the 16MB "8X" Lexar card for Fine mode shots. Secondly, the card is interfering with the review and delete function on the camera. Don can't use the garbage can to delete shots after review and can't pause the picture using this card only. What is wrong remains to be seen, but Lexar is apparently aware of the problem and may allow exchange of your card. Some reports say that only the non-USB version of the cards seem to have the problem. Reports are that the Lexar 32MB "8X" cards may have this problem also, but obviously the 16MB card provided with the camera does not.
*** Neil sent his original "non-USB" Lexar 80MB card to Lexar for replacement, and received this "USB" 80MB card with updated firmware from them. He reports that the latest firmware revision on the Lexar cards is rev "G". I don't know how you can check the firmware yourself, sorry.
**** Corey sent his USB Lexar 80MB in to Lexar for replacement. This is the replacement card, and it is performing up to specifications now.
** I have good confidence in Ernie's timings, note that the 10X lexar card does not seem any faster than the 8X card in the 990. It will be interesting to see more 10X card timings.
Firmware 1.1 timings
Card Time for TIFF shot Time for Fine shot
Lexar USB 16MB "8X" 29.7 4.6 (av. of 5, Bryan's card)
Lexar USB 16MB "8X" 30.1 4.3 (av. of 5, Scott B.)
Lexar 32MB "4X" 33.4 5.4 (av. of 5, Bryan's card)
Sandisk 32MB 33.3 -(Vincent S.)
Simple Tech 48MB 53.5 6.7 (av. of 5, Bryan's card)
Lexar 48MB "4X" 28 - (Adolfo C)
Lexar 64MB 8X USB 28.5 3.6 (av. of 5, Bryan's card
Lexar 128MB 8X USB 29 -(Vincent S.)
Kingston 64MB "high speed" 44.8 6.56 (av. of 5, Bryan's card)
Delkin 64MB 51.9 7.0 (av. of 5, Scott B.)
Simple 64MB 47 4.2 (av. of 5, Eduard S.)
Unity Digital 64MB 29.5 4.3 (av. of 5, Bill M.)
Simple 96MB #1 47.0 6.9 (av. of 5, Scott B.)
Simple 96MB #2 39.4 6.0 (av. of 5, Scott B.)
Microtech 96MB 32.5 4.7 (av. of 5, Ola S.)
Lexar 128MB 8X USB 29 - (Vincent S.)
PQI 128MB 54 - ( Adolfo C.)
Sandisk 128MB 57.3 8.2 (av. of 5?, Cristian C.)
Toshib 128MB 30.0 -(Ranier W.)
Mr. Flash 128MB 43.3 6.8 (av. of 5, Bryan's card)
CP990 conclusion:
Unlike the CP900 and CP950, the CP990 does seem to benefit from faster flash cards. Note that with Lexar, "8x" does not mean "8X" faster than other cards, it means "8X" faster than a "1X CDROM". (see below). The Lexar cards are not necessarily faster than others. There may be diminishing returns with the speed, note that the "10X' and "8X" Lexar cards appear to be identical. I would say that it might be worth getting a faster card for the 990, especially if you plan to shoot TIFF. The fact that cards show increased speed in the 990 might partly be due to the larger file size, as files get larger, the camera spends more time writing to the card, and flash card speed will become more important. Note also how newer cards from the same manufacturer (Simple tech) are faster than the older cards.
There has been talk about the "fact" that larger cards are slower for some reason, but the tests here with empty cards do not seem to bear that out. The Lexar "8X" 160MB card appears to be just as fast as the Lexar 16MB card, for example. Larger cards do get slower as they fill up due to overhead in the directory probably. My pet theory about this is that it has to do with the sequential file numbering feature. In the original CP950 1.0 firmware, the camera would overwrite files that already existed on the card with files in the sequential numbering sequence. After some complaints, this was fixed in more recent firmware releases. I suspect that what has to happen is that the camera has to check for duplicate file names on the card, and this gets slower and slower as you add more files. Plus add in the overhead of the larger directory, plus possible file fragmentation, and the operation gets slower as you add more files.
One funny thing, most cards that have been tested thus far are slower with firmware 1.1, with the exception of my Lexar 64MB 8X card which was faster.
CP950 flash card timings
(May, 1999) I have decided to expand my flash card timings to include the CP950 camera. I was not able to use the same testing procedure that I used for the CP900 (10 shots in continuous mode) because the memory buffer on the CP950 skews the results by buffering the shots (and will not allow 10 shots in Fine mode). Instead, I reasoned that if there is a timing difference that it would show up best on the recording and display of the giant TIFF images that the camera is able to take. As you can see, unlike the CP900 where there was no significant difference between cards, the CP950 did seem to benefit a little from a "faster" flash card.
Time to take store a TIFF shot. From shutter press until the hourglass is gone. The camera was pre-focused with a 1/2 press on the release, and the cards were empty for the test. (two times taken
for each card to show that the timings are consistant)
Delkin 32MB card 28 sec (Dougal Austin)
Microtech 32MB card 28 sec, 29 sec (David Wei)
Microtech 32 MB as delivered 36, 37 sec (Scott Aronian)
Microtech after in camera formatting 32, 33 (Scott Aronian)
Simple Tech 48 MB card 35.48 sec, 35.09 sec (Bryan's card)
Lexar 32 MB "4X" card 30.15 sec, 30.09 sec (Bryan's card)
Lexar 80MB "8X" card 30 sec (Thanks Leo Evershed!)
Lexar 16MB "8X" 30.02, 29.92 (Bryan's card)
Nikon 8MB card 38.10 sec, 38.50 sec (Bryan's card)
Nikon 8MB 31 sec, 31 sec (Scott Aronian)
Sandisk 96MB 34.3 sec, 34.9 sec (Thanks Walter Kimmel!)
Sandisk 30MB 38.1 sec (Thanks Indra Laksono!)
Sandisk 40MB 35sec, 34 sec (Sent in by "G.")
Sandisk 15MB 35 sec, 35 sec ("G".)
Nikon 8MB 32 sec ("G.")
Time taken to display a TIFF shot, from switch to Play, until the hourglass is gone.
Delkin 32MB 18 sec (Dougal Austin) (buffer flushed)
Microtech 32MB card 23sec, 24sec (David Wei)
Microtech 32MB 23, 23 sec (Scott Aornian)
Microtech 32MB camera formatted 22,23 sec (Scott Anornian)
Simple Tech 48 MB card 25.23 sec (Bryan's)
Lexar 32 MB "4X" card 25.10 sec (Bryan's)
Lexar 80MB "8X" card 23 sec (Thanks Leo Evershed!)
Nikon 8MB stock card 25.33 sec (Bryan's)
Nikon 8MB 22, 23 sec (Scott Aornian)
Sandisk 30MB 20.2 sec (Thanks Indra Laksono!)
CP950 conclusions:
There was a difference between different brands of cards used in the CP950 in the storage of pictures only. Playback speed was the same for all cards. Unlike in the CP900, the Lexar card was faster (but not 4 times faster. Also note that there is no speed difference between the Lexar "4X" and the Lexar "8X" cards. If you read the specs on the Lexar card, you will see that the 4X means "4 times as fast as a 1 times CD ROM drive", not four times as fast as other cards. Nevertheless, it was significantly faster.) Note also that since I do not have any identical size cards of different brands, that I can't test the effect of different sized cards. It has been reported that smaller cards are faster (which should have helped my 8MB card).
Also note that the so called "8X" card is not much faster in the 950.
The Microtech card is the fastest at the moment. I assume that this was a card with the new controller.
Send me your CP950 timings!
This TIFF shot test should be easy to reproduce. Send me some timings and I will add them to the table. I'd especially like to see a timing from the new Lexar "8X" cards. NOTE! The 950 buffers a shot in memory, so to measure the playback correctly you need to switch to Play, switch to M-Rec, then switch back to Play to flush out the buffer. Thanks to David Wei for noticing this.
CP900 timings.
There have been several reports of some Compact Flash (CF) cards being
faster than others in the CP900. I did some testing on the cards that
were available to me, and here are the results. I was interested in the
speeds of the cards when taking pictures.
Executive summary: There is maybe a very small speed difference, but it isn't much. The Lexar "4X" card
was not faster than the simple card.
Here is how I did my timings, (I'm open to suggestions). I think that
others could do this with the CP900 and we could compare.
1) Pointed the camera at lit a white wall so that the timings were all
of the same subject (same JPEG compression time. I don't know how
important the subject is if we are going to compare between cameras.
2) Used "Fine" mode for the largest file and the least compression.
3) Used an empty flash card.
4) Flash off, so charging time does not skew the results.
5) Because of the difficulty in timing a single shot, I set the camera
to continuous, and timed 10 shots. I timed from the first press, until
the 11th click (Start of the 11th picture) this is 10 shots.
6) Rounded to 0.1 second. I'm not that quick.
7) I made several trials for most of the tests.
Simple 48 MB card, white wall target (times for 10 shots)
48.15 sec
47.52 sec
47.60 sec
47.61 sec
Average time per shot... 4.77 sec
Lexar 32 MB card, white wall target
48.73 sec
48.70 sec
48.61 sec
48.44 sec
Average time per shot... 4.86 sec
PNY 8MB card (Sandisk)
51.73 sec
52.17 sec
Average time per shot... 5.20 sec
Lexar card, complicated target (living room, similar lighting)
48.33 sec
49.09 sec
47.70 sec
49.01 sec
Average time per shot...4.85 sec
Lexar card, complicated target, Previous shots left on card.
48.29
Average time per shot 4.83 sec.
Conclusions:
1) There is a speed difference between these cards, but it isn't much.
2) Contrary to what I
had thought, the subject does not change the timings. The camera isn't spending more time compressing shots with complicated subjects..
3) Having a full flash card does not change the timings. I thought that it might.
4)
From the results, I would assume that the camera has a fixed timing cycle for writes, so that there is no benefit from the faster cards.
Notes:
* I couldn't test the Nikon 4MB card that I got with my 900, it was too small to hold 10 fine shots. You think that Nikon could give out a bigger card *grumble*.
* I got tired of taking pictures
with some of the later tests, but the first tests show that there isn't much variation.
* In continuous mode, the picture that you took isn't re-displayed, so we aren't measuring display speed. A slow card might take longer
to display the picture after a shot also, slowing you down. Since the card has to read out the picture when you are using the camera normally, the differences between these cards might be more apparent. A 0.3 second
difference might be 0.6 seconds, or more if some cards read faster.
I'll have to figure out some way to test that, ha.
* April 14, 1999 I have now sold my CP900, and plan to get a 950. When I do, I'll re-test my assorted
cards to see how they behave in the 950.
* I have heard it said that larger cards are slower. I would have to test several cards from the same manufacturer to see if that was true.
I would welcome other test results. If you have a card that can hold 10
shots in fine mode, run through the test and mail me the results.
e-mail me some CF card timings