Z cameras tested - CFExpress vs XQD

Camnostic just published some test results comparing CFExpress with XQD for a 30 second burst. Their results show an average of 30% more images with CFExpress.

https://camnostic.com/2021/02/nikon-z-series-gains-30-burst-capacity-with-cfexpress/

Keep in mind that there is a lot of variation depending on card brand, model, and card size. The best performance is with larger sizes of premium cards - ProGrade Cobalt, Delkin Power, Angelbird, and Sony Tough. There are also specific camera settings that impact burst length and speed.

Good read for those interested in fast bursts. I did note that the tests were performed using a Z7II, so hopefully someone will test using the Z7 for comparison at some point. How much of the speed increase is due to the 2nd processor, newer architecture, etc?

Thanks for posting.

There are a lot of variables, but in general I don't expect you will see improvement in Z6/Z7 performance using CFExpress cards. There is still variation in speed and burst length depending which card you use. The reports I've seen indicate a slow CFExpress card may be 10% slower than a fast XQD card, but a fast CFExpress card can be a little faster than XQD.

I just did some testing of cards in my Z7II. I tested three cards - Delkin Power 128 GB, Prograde Gold 128 GB, and Lexar XQD 64 GB. I tested with 14 bit Lossless compressed and 12 bit lossless compressed and most other settings optimized for speed.

I was surprised by the differences in write performance - both for a 30 second test but also for shorter periods as well as measurements by the second.

The Delkin Power card was able to shoot 90 14 bit NEF images in 15 seconds and dropped to 63 images in the second 15 second period when the buffer was full. The first 15 seconds the camera was able to take 6 images per second every second, then was more volatile after the buffer filled. Changing to 12 bit allowed faster shooting with 101 images in the first 15 seconds and 72 images in the second 15 second period. This card had the most even performance for the first 15 seconds and the predictability would be helpful.

The ProGrade Gold 128 GB card was able to shoot 52 images in the first 15 seconds, and 29 images in the second 15 second period. There was a big drop off in speed after 7 seconds when the buffer filled. The card was very inconsistent - starting and stopping throughout the test. Seven times the camera paused and went a second or more without taking an image. Dropping to 12 bit helped some with 74 images in the first 15 seconds and 37 images in the second 15 seconds.

The Lexar XQD 64 GB card was surprisingly good. Shooting 14 bit NEFs, I was able to shoot 74 images in the first 15 seconds and 61 in the second 15 seconds. The problem was the XQD card seemed less controlled than the Delkin CFE card. It shot 43 images in the first six seconds, and after that was much less controlled with good and bad variation seemingly at random.

The thing that bothers me about variability is that it is so unpredictable. What if I was shooting a wedding. I'm cruising along with "the kiss" and my buffer is nearly filled. Then just as they start walking down the aisle, I'd reduced to 1-4 fps. I have 2 seconds that are critical, and I don't know whether I'll have 6 frames, 8 frames or zero frames. With the upper end Delkin Power card, I know I get 6-7 frames. With other cards, I might get 5-6 frames, but I might only get 0 or 1.

I can't tell how much of this is Nikon vs. the card manufacturer. The XQD card was about 5-6% faster than the Delkin CFExpress card in my Z6, but slower in my Z7II. The ProGrade Cobalt CFExpress was the slowest card, but in my Z6 it shot only 12% fewer images the Delkin CFExpress. The difference was more meaningful in my Z7II where the Delkin card allowed me to shoot 80% more images in 30 seconds.

I'm not sure how serious your example was. If you need the predictability of 6 shots per seconds during the kiss and a walk down the isle, I'm seriously questioning whether how you shoot makes any sense.

True, predictability is important, but that example is ... aeh ... off.

Hi Eric,

Interesting results.

Thanks for sharing.

Just an example - but there are times when you need a predictable frame rate. The range of 0-7 frames per second is not what I want. Walking down the aisle is a place where you want a high frame rate because the leg and foot position can make or ruin a shot. The wrong position for a full length shot and they look awkward regardless of the facial expressions.

I ran more than 6000 test images with two cameras. The intermittent speeding up and slowing down was partially related to the cards.

The cameras track the number of shots remaining as the buffer fills. With the Z6 I was watching the count drop from r20 to r0, and then at times it increased to as much as r9 and dropped again. The Z7II was more predictable and tended to stay around r0 to r1.

Interesting perspective. I shot a few weddings, but it's a long time ago. Maybe times have changed that much and people run more (kidding, of course), but I never had issues with shooting walking the aisle with film and often even manual focus cameras – never had a problem getting the right shots.

I'm happy that I'm not doing this professionally (and never did) and especially not now where it seems expectations have changed radically ...

Eric, thanks for putting this together. It's an eye-opener! I currently don't have CFExpress cards, but was contemplating a purchase soon and looking to purchase based primarily on price (assuming performance was similar).

I seldom have to shoot burst, but when I do (dance, motocross, etc.) it would be good to have predictable camera performance.

I'm going to rerun the test with the camera set for a faster speed. I was using Silent shutter and according to the manual, it's faster with Electronic shutter but not Silent. I was using AF-C Single which is fastest.

One of the reasons why this can be so unpredictable is that modern cards, especially the XQD and CFExpress types, have the same logic as SSDs: they try to spread writes and re-writes over the whole disk to not prematurely wear out certain areas first. When this is done, a mapping of potentially bad sectors can also happen.

Now, this needs a certain amount of logic to be executed, and it's not the camera doing this, it's the card/SSD, this means that quality of algorithms, performance of on-disk compute, as well as wear status of the disk have a massive performance impact.

You want to test cards that have seen roughly the same amount of use, well within their specified wear. If they are in different states of their life, you'll see vastly different results, even within the same line of cards.

This is something that has to be kept in mind with these cards: they are only specified for a certain amount of "re-writes" (amount of data being written on the card as a multiple of the card's size). Lesser quality cards or SSDs have less re-writes. That's why enterprise SSDs are so much more expensive than consumer SSDs.

Assuming that your card size is big enough, it would be great to see the difference in speeds compared by using some higher camera. I mean if you take a CFExpress and put it in a D6 and then compare the same card by putting it in your Z camera, could an argument be made that its the card (or the buffer size) that’s the problem.

"Today is the tomorrow that yesterday you spent money like there was no."

Visit my Nikonians gallery.

It looks to me like you have speed controlled in two places. The D6 has a different processor, more battery power, and a higher frame rate, so I expect it would right more or longer at peak speed. It would be interesting to see if the card slows and operates in a similar manner.

At the same time, if the card is limiting speed, performance may be slower, more variable, or maybe has some other impact.

My other post on this topic shows the second by second frame rate for the Z7II and Z6 with some specific cards. I'm interested in what the D6 or D5 looks like with a couple of different cards.

This was a review I came across a couple of months ago after I bought my Z6II. The camera store manager recommended the Delkin 128gb card with the Delkin card reader bundle he had. Coming from a D700, I never factored in the need for such speed, but also the need for temperature control. The Delkin cards smoke the competition in these tests on both speed and temperature control.

https://www.thessdreview.com/our-reviews/cfexpress/delkin-devices-64gb-128gb-ands-256gb-power-and-prime-cfexpress-type-b-cards-reviewed-great-sustained-speeds-at-low-temps/

The catch is that test was not with a camera. The speeds are much slower when you are using a camera, camera processor, and limits the camera engineers placed on the camera to prevent issues with different models of CFExpress cards.

I do like the Delkin Power 128 GB CFExpress card and reader. The bundle with a reader is a great deal, and Hunt's still has it available.

< also the need for temperature control. The Delkin cards smoke>

JIM-

Really? ;P

"Today is the tomorrow that yesterday you spent money like there was no."

Visit my Nikonians gallery.