For this test, I fired off 3 bursts of 20 12-bit, lossless compressed RAW files, otherwise using the same settings from the earlier test. The D300′s buffer can absorb 17 files under these conditions, so this is just enough of a test to begin to see major differences in write speeds.
Latest Update (11/20/2010): Patriot LX Series 600x.
SanDisk ExtremePro
Highly Recommended √+
- SanDisk Extreme Pro. The new king of SanDisk’s line, this card reached 6.7 fps with an effective throughput of 34 MB/sec.
- PhotoFast 533x Plus. This little-known card hit 6.7 effective fps with an impressive 31 MB/sec throughput.
- Lexar Professional 600x. Lexar’s current top-of-the-line card reached 6.7 effective fps and 31 MB/sec.
- San Disk Extreme Ducati Edition. This card initially topped the performance list with an effective 6.7 fps and nearly 29 MB/sec throughput.
- Transcend 600x Extreme Plus. This card equaled the Ducati, achieving 6.7 fps and almost 29 MB/sec throughput. A real contender.
- Ridata 600x Lightning Series. A nice entry from Ritek, equaling the Transcend 600x at 6.7 fps and 29 MB/sec but about $10 cheaper for a 16GB card.
- Lexar Professional 400x. A real performer from Lexar that trumps others in its class at 6.7 fps and 28 MB/sec throughput.
- San Disk Extreme IV Series. The two cards I tested – 8GB and 2GB – were just behind the Ducati and averaged an effective 6.3 fps and approximately 28 MB/sec.
Purchasing links (support this site):
Buy Extreme Pro Cards from Adorama: 16GB, 32GB, 64GB
Buy Lexar Pro 600x cards from Adorama: 8GB, 16GB, 32GB
Transcend 8GB 600X CompactFlash Extreme Plus UDMA Card
Transcend 16GB 600X CompactFlash Extreme Plus UDMA Card
Transcend Compact Flash 600x 32GB Memory Card
RiTEK Lightning Series 16GB Compact Flash (CF) Flash Card
Buy Lexar Pro UDMA 400x Cards from Adorama: 8 GB, 16GB, 32GB
Recommended √
- Delkin 625x CombatFlash. 5.5 effective fps and a speedy 31 MB/sec throughput.
- Hoodman RAW 675x. 5.5 effective fps and 30 MB/sec transfer rate.
- Pretec 667x. Respectable 5.5 effective fps with a 25/MB sec transfer rate.
- Transcend 400x UDMA. Distinguishes itself from its 300x sibling with 5.5 effective fps and 25 MB/sec transfer rate.
- Sony UDMA 300x. This card fared somewhat better in RAW shooting than for JPG, coming in at 5.5 effective fps and 24 MB/sec.
- PhotoFast 533x. The standard version of the 533x performed well, with an effective fps rate of 5.5 and 24 MB/sec throughput.
- Kingston 600x Ultimate. Disappointing 5 fps and 28 MB/sec throughput for a 600x card.
- PNY Optima Pro. This card averaged an effective 5 fps and 24 MB/sec throughput.
- Transcend 300x UDMA. Transcend’s 300x UDMA card hit an effective 5 fps and 24 MB/sec.
- Lexar Pro UDMA. The Lexar edged into the recommended performance tier for RAW shooting, with an effective 6 fps and 23 MB/sec throughput.
- Pretec 433x. Despite its official claim, this card doesn’t approach its listed speeds of 65 MB/sec read and 50 MB/sec write in real world testing. It clocked 5.5 effective fps and 23 MB/sec.
- Hoodman RAW 300X. While a marked improvement over its 280X predecessor, this card only managed 5.5 effective fps and 23 MB/sec throughput in the burst test.
- SanDisk Extreme III 30MB/sec. The upgraded, UDMA-capable Extreme III series clocked in at an effective 5 fps and 23 MB/sec throughput.
- SanDisk Ultra 30 MB/sec. New entry-level card gets good marks at 5 fps effective and 23 MB/sec throughput.
ATP Pro Max II UDMA. A pleasant surprise, the ATP card — an also-ran in JPG burst shooting – joins the RAW shooting journeymen with an effective 5 fps and 22 MB/sec throughput.- Pretec 333x. Another also-ran in JPG shooting, the faster of Pretec’s original two UDMA models consistently recorded 5 fps and 22MB/sec in our RAW burst test.
- RiData 300x. 5 fps effective and 22 MB/sec earns RiData its first “Recommended” accolade.
- Patriot LX Series 600x (8GB). Effective 5 fps and 22 MB/sec. Not very impressive for a 600x card.
- Lexar Professional 233x. Lexar’s most affordable UDMA card achieved a consistent 5 fps with 21 MB/sec throughput.
- A-Data Turbo 350x. More proof than manufacturer specs means little, this card reached a merely respectable 5 fps and 20 MB/sec throughput. It’s much better than their “266x” card but not as good as 300x cards from Lexar and Transcend, either.
Delkin 625x (4GB)
Delkin 625x (8GB)
Delkin 625x (16GB)
Delkin 625x (32GB)
Buy the Hoodman 675x from Adorama: 4GB, 8GB, 16GB, 32GB
Transcend CompactFlash 400x 8GB Memory Card
Transcend 16GB 400X CompactFlash UDMA Card
Transcend 400x CompactFlash 32GB Memory Card
Buy Lexar Pro UDMA 300x Cards from Adorama: 2 GB, 4GB, 8GB
Buy Sony UDMA 300x CF cards from Adorama: 4GB, 8GB, 16GB
Buy Extreme III Cards from Adorama: 4GB, 8GB, 16GB, 32GB
Buy Lexar Pro 233x Cards from Adorama: 2GB, 4GB, 8GB
Marginal √-
- Super*Talent 600x & 533x. These cards turned in near-identical performance of around 4.5-5 fps and 16 MB/sec effective throughput.
- SanDisk Extreme. This so-called “60 MB/sec” card reached a measly 4 fps effective and 18 MB/sec throughput, earning this overpriced dog a spot on the Crap List.
- Kingston 266x Ultimate. The Kingston led a pack of cards whose performance was a bit disappointing, at just 4 fps effective and 14 MB/sec throughput.
- Ridata 266x Supreme. Ridata’s second effort reached 4 fps and 14 MB/sec throughput.
- Apacer Steno IV 300x. This entry hit 3.75 effective fps with 14 MB/sec throughput.
- Transcend 266x. Transcend’s faster UDMA card hit 3.75 effective fps and 13 MB/sec throughput.
- Delkin Pro UDMA 305x. Delkin’s card just topped 3.5 effective fps and nearly 12 MB/sec.
- A-Data Speedy 533x. A Crap List awardee due to its ridiculous label claims, this card reached just 3.3 fps effective and 12 MB/sec throughput.
- Hoodman RAW 280X. Hoodman’s card limped in at 3.3 effective fps and 11 MB/sec throughput.
- Calumet ProSpec 305x. This very likely re-badged card has mediocre specs of 3.3 effective fps and 11 MB/sec throughput.
- EP Memory 266x. Ho-hum performance of 3.3 effective fps and 10 MB/sec throughput.
- Calumet ProSpec 420x. Another re-badged “improvement,” it sucks even worse than its predecessor at 3.0 effective fps and 11 MB/sec, and made The Crap List.
- Delkin PRO UDMA 420x. Ugly 2.9 effective fps and 13 MB/sec throughput. Belongs onThe Crap List with its twin the Calumet.
- BiWin 150-300x. This 16GB monster struggled to reach 2.9 effective fps and a 9 MB/sec throughput rate.
- Patriot 266x. 16GB card reached 2.9 fps and 9 MB/sec.
- A-Data Turbo 266x. A-Data was sucking wind at 2.7 fps and 11 MB/sec.
- FileMate UDMA. Drags across the finish at 2.7 effective fps and 8 MB/sec.
The ranks of the also-rans included the following sub-par performances: Pretec 233x (2.3 fps and 9 MB/sec), Ridata 233x Lightning (2.3 fps and 8 MB/sec), Wintec Professional (1.9 fps and 6 MB/sec), and Transcend 133x (1.7 fps and 6 MB/sec). Two non-UDMA cards yielded the following results: Transcend’s 120x 2GB card hit 3.2 fps and 12 MB/sec, while the Kingston Elite Pro 133x 4GB card reached 2.2 fps and 8 MB/sec – quite respectable compared to some of the above UDMA cards.
Latest update 10/16/2010: Added Kingston 600x Ultimate.
Updated 10/6/2010: Added Lexar Professional 400x.
Updated 9/15/2010: Added Ridata 600x 16GB, Patriot 16GB 266x, Wintec Professional 8GB.
Updated 9/6/2010: Added EP Memory 266x and FileMate UDMA.
Updated 6/6/2010: Added results for the SanDisk Ultra (200x).
Updated 4/12/10: Added results for the Delkin 625x CombatFlash, Hoodman RAW 675x, and Delkin 420x.
Updated 1/16/10: Added results for the Lexar Professional 600x, Trascend UDMA 400x, Super*Talent 600x & 533x, and Calumet ProSpec 420x.
Updated 12/26/09: Added results for the RiData 300x and A-Data 533x cards.
Updated 10/28/09: Added results for the SanDisk Extreme Pro and Extreme cards.
Updated 8/27/09: Added results for PhotoFast 533x Plus and 533x.
Updated 8/3/09: Added rating for Pretec 667x, Transcend Extreme Plus 600x.
Updated 4/2/09: Due to the switchover from MLC to SLC memory by many of the smaller card makers, I’ve readjusted the recommendations to reflect the new, generally faster speeds of many of the middle-of-the-pack
cards. The “Highly Recommended” class now truly stands out from the rest, and these are the first cards I personally put into my D300 and D300s.
16 responses so far ↓
1 Presidents Race // May 16, 2008 at 4:09 PM
If I look at Rob Galbraith’s data, the Sandisk Extreme IIIs are near the tops in terms of performance. Given the relatively lower prices and rebates out there now, they seem like a great deal. Have you tested the Extreme IIIs with the D300?
2 The Sports Photo Guy // May 16, 2008 at 5:02 PM
I’m looking for one of the new Extreme III’s to try out. SanDisk upgraded the Extreme III line to UDMA recently, but there are still plenty of the older versions out there.
3 Janko Wodnik // Mar 17, 2009 at 4:30 AM
Some of newer Kingston CF 133x memory cards support UDMA mode 4 but it still sucks in sequential write speed with only 15MB/s on PC IDE bus with proper cluster alignment.
Sandisk Extreme III (new version) is UDMA mode 2 (bus speed 33MB/s) CF card.
Also cluster alignment operations on the partions and file systems of flash media are very likely to add even +50% sequential write speed. It is usefull for MLC flash, low-end storage media are MLC.
Please note that CF speed in cameras are also limited by transfer modes. The way of negotiation transfer mode is a magic in most non-PC devices. Most popular transfer modes for PC are PIO 0-4 nad UDMA 0-6 while cameras utilize PIO 0-6 (and coresponding multi-word DMA 0-6) and sometimes UDMA 0-4.
4 KonstantinMiller // Jul 6, 2009 at 11:28 PM
You know so many interesting infomation. You might be very wise. I like such people. Don’t top writing.
5 Eric Max // Dec 29, 2009 at 9:43 PM
Thanks for testing the A-Data Speedy 533x card. I bought one recently, but hadn’t unwrapped it yet. Looks like it’s going back in favor of something faster!
6 Lostgallifreyan // Apr 24, 2010 at 5:48 AM
Good to see that Transcend do ok there. I found them to be by far the most compatible for building silent machines with. I don’t know if the other firms have such full and easily found public disclosure of highly detailed engineer’s specifications for their cards, but if so you might find that to be useful to gauge quality, though it might be a heavy-going foray at best. I found all Sandisk cards I’d tried (up to ‘Extreme III’) to be horribly disappointing, and buying them on eBay to be a minefield due to extreme amounts of counterfeiting fraud. Transcend cards were harder to find but sources were small and more direct, and the difference in everything I saw was dramatic. The prices were lower, speeds better, compatibility higher… I don’t doubt that Sandisk’s best are good but the cost and difficulty of being sure that’s what I actually GOT was insurmountable! The Transcend cards were so good I ended up getting spares just in case the word gets out and they become as hard to get at good prices as Sandisk’s have. The thing that really decided it for me was that Transcend’s engineering data was so complete that I could go far deeper than I’ll ever need to go in verifying theuir claims, and examining details of other stuff to look for causes of any incompatibility that might arise. (haven’t seen any yet). CF cards for me suddenly went from an exotic curiosity, only needed for a small PDA, to something to build silent machines with, for high speed, and a deliberate intention to avoid a hard disk. This goes way beyong photography now, it opens the field for high performance industrial machines that almost anyone can put together, saving power, making no noise, standing extreme physical shock… Even now it seems few people have got this, but when they do this market will soar, and hard disks might start to show their age as a species. (Though it’s still astonishing how well those have survived and increased capacity too).
7 David Cree // Sep 15, 2010 at 2:35 PM
Forgive my ignorance—I still am having difficulty understanding the raw burst data.
How many photos can I take in a single burst, holding the shutter button down the entire time in NEF/Raw (12 bit) before the camera STOPS taking pictures using a San Disk Extreme Pro???
You say 6.7 fps –fine, but for HOW LONG and HOW MANY shots can I get on my D300 before it stops ?? 15 photos 20 photos 30???
Thanks
David Cree
Peachtree City, Georgia
8 David Cree // Sep 20, 2010 at 1:16 AM
I have gotten my D300 to register “r17″ in the top display to show that the buffer will take 17 Raw 12 bit photos non-stop at 8 fps (battery pack attached).
So, that’s about 2 full seconds, then the buffer is full and data begins being transferred to the flash card.
With the fastest card known to man at this time—let’s say 600X -90MB/sec – for argument sake, HOW FAST can the D300 write to the card cleaning the buffer out??
25Mb/sec, 30 Mb /sec, 35 Mb/sec?????
Even if there was a 1000X CFC @ 150 Mb/sec speed, the D300 has to have a limit on how fast it can transmit data. I would like to know what that LIMIT is. Do you know??
Thanks.
David Cree
Peachtree City, Georgia
9 The Sports Photo Guy // Sep 20, 2010 at 8:06 AM
David,
With the fastest 600x cards, the D300/D300s can write ONE 12-bit NEF to the card from the buffer by the time the buffer is filled at 8 fps. That’s a big reason why I chose 20 frames for my “real world” test – it’s going to fill the buffer and have to wait for it to clear space for the last 2-3 frames.
The MB/sec numbers from my test are a fair approximation of how fast the camera writes to the card – it is the total time from triggering the first frame until the card activity light goes out at the end of the burst. While I can’t control the final file size – even shooting the same “target” under the same conditions, there will always be variations – it’s pretty close.
The limiting factor in terms of overall speed is definitely the D300 at this point; it doesn’t appear capable of more than 35 MB/sec or so write speed.
10 iluv4n6 // Dec 13, 2010 at 7:00 AM
A bit off topic, but Is it normal for a larger capacity card to have an effect on how long it takes for a picture to appear on the LCD display after it is taken?
With my style of photography I rarely fill the buffer so speed isn’t the highest of priorities when buying CF cards for my D300. I have three Transcend CF cards: a 4GB 300x, an 8GB 133x, and a 16GB 133x.
the 4GB and 8GB cards display the photos in under 2 seconds from when my finger leaves the shutter button; with the 16GB card in, it can take 4-5 agonizing seconds to display. Strangely enough, it tends to take longer in a studio situation with strobes involved (triggered by Cybersync RF transmitters).
I’m afraid to buy anything over 8GB of ANY speed due to the eternity I have to wait for my 16GB card. Any ideas?
(I should mention that single shot mode doesn’t help at all [i.e. I'm not approaching the capacity of the buffer])
11 The Sports Photo Guy // Dec 13, 2010 at 7:21 AM
Have to say I’ve never encountered this lag. You will find sometimes significant differences between different-sized cards of the same speed manufacturer/speed rating, so it could be that the 16GB Transcend is just inherently slower than its cousin. But let me also ask: do you tend to use the 16GB when shooting RAW (due to its capacity) and the others for JPG? That could certainly be a possible cause; logic would suggest the camera takes longer to retrieve and display an NEF from the card than a JPG.
12 iluv4n6 // Dec 13, 2010 at 6:34 PM
I shoot exclusively in RAW format, so that’s not the issue. I did a very non-scientific test and shot a couple pics in RAW and some set to JPEG basic–same waiting time to show up on the back display.
I’ll just have to buy more CF and compare. Bummer!
Thanks a lot for you’re time!
13 Alexandar // Dec 31, 2010 at 4:12 PM
I have purchased one SanDisk Extreme Pro 16 GB card from Amazon and did some tests on my Nikon D300 the following way:
I covered the lens with the lens cap, put it on manual focus and did a high speed burst for 20 shots, with the display off (12-bit, lossless compressed RAW files) . I measured the time from the first shot until the card light went off and it was approximately 13 seconds.
Each picture size was on average 7.8MB so that was a total of 156MB for all the 20 pictures taken.
Dividing the 156MB with the 13s time i got 12MB/s which i 1/3rd of the time posted.
The same test on the SanDisk Ultra II 2 GB took 25 seconds.
Thinking that i might have purchased a counterfeit Extreme Pro 16 GB, i have send pictures of the card to SanDisk and i was told that based on the pictures and the serial number of the card, it was a genuine SanDisk Extreme Pro 16 GB.
So, either it’s a good counterfeit card or i’m doing something wrong?
I don’t have a good card reader yet, so i only did a benchmark using the card in my camera and got 8MB/s write and 13MB/s read speed when using the ATTO Disk Benchmark program.
Can someone please do the exact test i did (Nikon D300 and SanDisk Extreme Pro 16 GB) and post the results here?
Many thanks in advance.
Alex
14 The Sports Photo Guy // Dec 31, 2010 at 5:39 PM
Other considerations affecting write speed would be Noise Reduction, Active D-Lighting (I turn all of these OFF for tests), Auto ISO (also OFF), and finally if you’re shooting without the grip at 6 fps (I shoot with grip at 8 fps) that will also affect your “effective” throughput.
15 Alexandar // Dec 31, 2010 at 6:02 PM
Thanks for your reply The Sports Photo Guy.I forgot to mention that i use the MB D10 grip with a fresh Nikon EN-EL4a battery. Active D-Lighting, Auto ISO, Noise Reduction are all off.
I will try to get another identical card and repeat the test, not sure if the high street shops have opened one so i can test it in the shop.
In the meantime i wrote another mail to SanDisk customer support to let me know if i can send the card to them for further testing.I registered the card on their website and it accepted the serial and showed that the card is under warranty, so i’ll wait for their reply as well.
Once again, thank you for your reply and i hope you have a great New Year’s Eve.
A.Charvat
16 The Sports Photo Guy // Jan 1, 2011 at 10:29 AM
Sounds like you’re testing under the same conditions as I am, other than putting the lens camp over your lens. I use a consistent target and lock AF rather than using manual focus.
Leave a Comment