Lexar XQD, CF and SD memory cards performance test on the new Nikon D5/D500 cameras

Lexar Japan published a new video where you can see the difference in performance between XQD, CF and SD cards in the new D5/D500 cameras:

The video is over 8 minutes, so here are the actual test results (number of shots taken displayed on the graph):

Nikon D5, Lexar XQD 2933x vs. Lexar CF 1066x memory card test:

Nikon D5 XQD vs CF memory card test
Nikon D5, Lexar XQD 1400x vs. Lexar CF 800x memory card test:

Nikon D5 Lexar XQD 1400x vs. LexaR CF 800x memory card test
Nikon D500, Lexar XQD 2933x vs. Lexar XQD 1400x memory card test:

Nikon D500 Lexar XQD 2933x vs. Lexar XQD 1400x memory card test
Nikon D500, Lexar SD UHS-II 2000x vs. Lexar UHS-I 633x memory card test:

Nikon D500 Lexar CD UHS-II 2000 vs. Lexar UHS-I 633x memory card test

You can see a similar Nikon D4s test here.

Thanks Stefan!

This entry was posted in Nikon D5, Nikon D500 and tagged , , . Bookmark the permalink. Trackbacks are closed, but you can post a comment.
  • ZZ

    Would be nice to see how CFast stacks up against XQD …

    • HotDuckZ

      Size & Design is winner, who will take 200 RAW everyday?

      • Brett A. Wheeler

        I will.

        • John D. Carnessiotis

          I do all the time…

    • PhilK

      Pointless because there is no product that supports both of those.

      This is not just a test of the media, it is a test of the camera I/O performance mostly, and the card performance secondarily.

      Also the difference between Cfast and XQD is not simply media transfer rate, there are various architectural distinctions. For example, XQD is based on PCI-express, just like Thunderbolt, so it should facilitate the production ofThunderbolt-equipped XQD readers with optimal performance at moderate cost.

  • Admin, thanks for the information. So the message remains unchanged: Use the right tool (with the matching Memory card) to achieve the desired rate of capture for the photographic objective. When FPS is necessary, invest in the faster memory cards.

    • yes, pretty much – not everyone needs the fps performance of a XQD card, but it’s good to know that is available

      • Us Birds in Flight shooters will need to spend the money to have the appropriate memory when we choose to resort to high speed burst mode to try and capture the optimum instant!

  • I hope XQD really continues to develop. I don’t like the form factor of the CF card anymore.

    • Jeffry De Meyer

      Cf is horrible one wrong move a prong bends and you can toss your reader or worse your camera

      • PhilK


  • the video also has memory cards reading tests at the end

    • luca

      the shame is that the double cards slots are in different format on D500 … so you MUST buy at least a couple of XQD cards …

  • Coastalconn

    The D500 seemed to start slowing down around 95 with the 2933x XQD. Seems a little short of the claimed 200?

    • IronHeadSlim

      I couldn’t if it was lossless compressed or uncompressed raw. That would make a difference.

  • mansell

    great innovation, the new sd cards can have 90MB/s bitrate, so why they have change them???


    My notebook don’t have this xqd card reader…
    Thanks from the useres…

    • IronHeadSlim

      Get a USB3 reader or a new notebook. You’re welcome. -the useres

      • mansell

        Hello nikon worker.

        • IronHeadSlim

          Your D3100 only takes SD card so no worries.

          The rest of us like the inherent speed in XQD.

          • mansell

            It’s better than your coolpix 😉

            • IronHeadSlim

              You leave my coolpix out of this!!

            • Duncan Dimanche


    • Colin Stuart

      your notebook doesn’t have a CF reader either so…

    • T.I.M

      I remember saying that when they updated the floppy disk to the hard drive disk….

    • Piooof

      Why XQD? Easy, look: the UHSII card is supposed to reach 300 Mb/s, and in the field, will give you 68 shots. The XQD 1400x is supposed to reach 210 Mb/s, so 30% lower, but gives you 4 more shots (72) in the same conditions. So the raw bitrate numbers are not enough to judge a memory card (surprise, surprise…).

      • Eric Bowles

        The SD cards use a bus with a limited speed. The XQD bypasses that bus and writes directly. UHS-I had a maximum speed for the bus of 104 MB/s and acgtual performance of about 75 MB/s. UHS-II will be faster, but we don’t have specifics yet. I heard 110-120 MB/s which is a little faster than CF – but I can’t confirm that.

        • whisky

          according to Lexar, their new 2000x cards are capable of up to 260MB/s write speeds and 300MB/s read speeds.

          my understanding is that the SD format is a dumb format, and therefore much of the speed depends on the hardware used to read and write. software overhead throttles the actual data transfer speeds.

          OTOH, XQD cards use a PCIe solid state controller, which rids them of the software overhead, and can thus transfer data much more efficiently.

          • Eric Bowles

            That’s largely correct, but the benchmark speeds are not the same as actual speeds in the camera. As you indicate, the hardware and software used to write to the card provide the limitations.

    • Eric Bowles

      It probably does not have a UHS-II reader either.

    • Chris Daigle

      Some notebooks have express card slots. Sony makes a reader that slides right in. Almost all laptops made today have that same slot. The reader is easy to store inside the laptop, unlike a CF reader.

      • PhilK

        I have had for a number of years a 54mm ExpressCard CF card reader that inserts flush with the laptop slot. Yes, the CF card itself sticks-out, but I wouldn’t have any reason to leave the card inserted after I’m finished with it.

        Unfortunately many laptops today have no expansion slot at all, especially the more compact ones. And the ones that still do have slots, often only have 34mm slots. In that case, I use either built-in USB 3.0 or my flush ExpressCard USB 3.0 card to connect to a small multi-format USB 3.0 card reader I carry with me. (About 1.5×3.0x0.4″)

    • Duncan Dimanche

      what is wrong with innovation ?
      Get over it already. You are waiting your time and.. well just your time really

  • Did I see it right? 2000x UHS-II card slows down at 30 frames? 1400x xqd card slows down around 40? This means if we are using SD card as the backup 30 will be the maximum buffer. Can someone confirm?

    • Eric Calabros

      “your whole system speed is as high as its lowest speed component” they said. sort of.

      • Eric Bowles

        They did not say that. That’s not exactly the way it works. Test your camera. All Nikons in backup mode write the first image to each card, then write the next image. Sop if you have 250 MB of images to write it can write them to a Lexar 2933x XQD in about 1-1.2 seconds, and if you also write to a Sandisk 95 MB/s UHS-I card in your second slot, you are writing the same 250 MB again but at a speed of 70 MB/s and it will take 3.7 additional seconds or 5 seconds total for the two cards. Meanwhile if you keep firing at 10 FPS, your buffer will be full after 6 seconds and you will drop to the write speed of both cards or around 2 fps.

        • Eric Calabros

          If it writes to both cards at the same time, why you add XQD time to SD time? Should be 3.7 not 5. Unless it’s not same time as I thought.

          • Eric Bowles

            It writes sequentially – not at the same time. It writes an image to the primary card then it begins writing to the second card. After writing the image to the second card is completed, it begins writing the next image to the first card. This is only in backup mode.

            • Allen_Wentz

              I see your comment was 2 months ago, and I can test for myself next week, but what happens when writing 14-bit lossless to XQD and JPEG Fine to SD? The much smaller JPEG should complete writing quickly but if the writing is sequential anyway the lame SD card will always be a bottleneck.

            • Eric Bowles

              It always writes sequentially – whether to one card or two cards. A small basic JPEG writes very quickly so it makes less difference than a RAW file such as in backup mode. But if you have a slow SD card, it could still slow down 10-12 fps shooting. I sometimes write a small basic image to my EyeFi card for immediate transfer to an iPad. This works, but it’s an old USB I card and will be pretty slow even for a small file.

    • Eric Bowles

      The buffer is fixed in size. It clears at the write speed. If you are writing to a second card, it clears at the write speed of both cards combined – writing a file to each card sequentially.

  • Eric Calabros

    Their XQD 1400x tops at 72, and their SD 2000x at 68, but both are around $100. so even if you don’t need the maximum rate, you’d better get lower priced XQD.

  • T.I.M

    XQD cards look great, I hope we will have the choice CF/XQD with the D810’s successor. (48mp is a lot of data)

    • 24×36

      Yes, but TWO of them (XQD) – ditch the stupid second slot SD!! SD dumbs down the speed of the camera, sort of like putting one sneaker and one lead shoe on a track runner.

      • Allen_Wentz

        Agreed 100%. Sell different optional card modules if necessary for folks who want to cripple their cameras’ performance by using SD cards, but give us the ability to choose a camera with two XQD.

        I use the second slot for backup, and the D500’s performance is crippled by the backup slot. Dumb.

  • doge

    I always shoot 100 photos at a time so this is great news for me.

    • Captain Megaton

      I actually like to depress the shutter in Cont-H to the rhythm of “Flight of the Bubblebee.” It is very annoying if I can’t get to the end of the piece before the buffer locks up.

  • Plug

    I hope Sandisk start to compete with XQD cards. Might tend to lower the price.

    • T.I.M

      still way cheaper than 35mm film…..

    • Allen_Wentz

      Sandisk and Lexar both said they would not do XQD back in 2012. Lexar obviously changed their mind but it seems Sandisk has not, instead choosing to promote CFast instead (so far anyway). Too bad.

  • JimP

    I’m amazed that people are commenting on a test that is so opaque. Looks to me like they are shooting 14 bit uncompressed RAW.

    • Eric Bowles

      I was trying to figure out what they were shooting. It’s definitely 14 bit. Was it uncompressed? If so, the files are around 48 MB each, so actual performance will be much faster. That also means the D5 has a 2.5 GB buffer.

      • Eric Calabros


        • Eric Bowles

          200 is the number of frames you can shoot including those written from the buffer to the card.
          The D5/D500 14 bit lossless compressed file is about 24 MB – the uncompressed version is nearly twice the size or 48 MB and the D800 file is 70 MB.
          200×25 is the amount of data generated if you are shooting 14 bit lossless compressed. Of that amount, up to 1.3 GB goes into the buffer leaving about 3.7-4 GB to be written to a card in 20 seconds. That requires a card with a write speed of 180 MB/s.
          There are only two XQD cards that have a write speed of 180 MB/s or higher – the Lexar 2933x and the Sony G. The UHS-II SD cards cannot achieve that speed and the other Lexar and Sony cards are slower. That means those cards cannot achieve a 200 frame burst with 14 bit lossless compressed RAW at full size.
          I just left a D5/D500 demo and the Nikon rep confirmed these statements and demonstrated both the ability of XQD to write 200 frames without filling the buffer and the inability of the UHS-II SD card to get anywhere near that number.

          • JimP

            According to Ole J. Liodden the uncompressed RAW on the D5 is 42-45MB. He wasn’t surprised by the size, but I am. Obviously the RAM must be at least 1 GB.

          • Eric Calabros

            1.3GB is about 52 frames, at 12fps its a bit more than 4 seconds. What happens if someone doesn’t go beyond 4 seconds? I mean how many seconds should wait to shoot another 4 seconds?

            • Eric Bowles

              The camera resumes at peak speed of up to 10 fps as soon as you have room in the buffer. So you could clear just a portion of the buffer and be able to shoot again at peak speed. The time it takes to clear the buffer is important. With a slow card like a 45 MB/s UHS-I SD card (actual write speed 35 MB/s) is could take 7-8 seconds to free up enough space for a one second burst at the full 10 fps capability of the camera.

      • JimP

        I’m not sure what you mean – as by buffer we’re always talking dynamic buffer – but my guess is that the D500 has a 512MB RAM chip.

    • Captain Megaton

      That’s exactly what’s written on the screen caps, so, I guess so!

  • Aldo

    There you go paparazzies, birth shooters, alien hunters… mocking aside, I can understand these people sit around for hours waiting for ‘the’ shot… you wanna increase your odds of having that million dollar photo.

    • TheInfinityPoint

      birth shooters…? lol

      • Eddy Kamera

        Births resulting from rapid shooting. LoL

      • Aldo

        hahaha… omg… well them too… gotta shoot fast you know …

  • jeverton1

    What happen to the Lexar and Sony XQD comparison on the D5?

    • I have not seen one.

    • Eric Bowles

      The video was just using Lexar products.
      The Sony G XQD is supposed to be about the same speed as the Lexar, but don’t get sucked into the new M card unless speed is not important. The Sony M will be about the same speed as CF – 100 MB/s in the camera. It’s optimized to be cheaper – not faster.

      • PhilK

        As I recall, the READ speed on even the 2nd-tier XQD cards is still quite a bit higher than CF maximum. So no it won’t help you shooting 200-frame RAW sequences in the camera, but it should speed up the process of viewing or copying them off of the card.

    • Allen_Wentz

      Thom Hogan did a quick test at

  • ZeBäst

    Focus issues and oil splatter fest at 14 fps incoming loooooooooooooool

  • usa

    First 50 shots appear capture at the same fps rate. 50 shots is enough for even the most casual NBA layup.

    As an aside the cps rate (complaints per second) about anything new from Nikon on this blog seems to be accelerating!

  • eric

    I don’t understand, The D5 and D500 are suppose to have a 200 picture buffer with 400 mb/s qxd. So you get more picture with a 160 mb/s cf card vs a sd uhII 260 mb/s??? doesn’t make any sense. https://www.youtube.com/watch?v=H1TD1Px5oVY&feature=youtu.be

    • Eric Bowles

      The buffer size is apparently twice as large on a D5 compared to a D500. The buffer is fixed in size and hold a limited number of images depending on the format you select.

      There is a lot of marketing speak claiming a 200 image buffer and none of these cameras have a buffer that is actually capable of holding 200 14 bit lossless compressed images. The best is the D5 which holds about 95 images. The D500 buffer holds about 55 images. The 200 reference actually describes maximum burst length, and it assumes you are able to write a large number of images to the memory card in the time it takes for a 200 image burst.

      Card speed and file size make a huge difference. The D5 camera can shoot 200 frames in 14 seconds if the buffer is not filled (in 14 fps mode). It can’t achieve that rate if the buffer fills and drops the frame rate to the write speed. If you are writing at around 260 MB/s, you have 3 fps going into the buffer representing about 75 MB/s. In 20 seconds the buffer is only half full. With the D500, the frame rate drops, so while the write speed is constant, you have less data going into a smaller buffer so it is still not full. But with a card any slower than 250 MB/s, the buffer will fill and frame rate drops to write speed for the remainder of the burst.

  • QuantalQuetzal

    Great thing is, that my notebook’s internal SDHC reader will also be able to read SDXC cards. That’s the major convenience advantage here, and SDXC UHS-II will certainly be fast enough for me. Also, seeing that UHS-II is only slightly slower than the cheaper XQD cards, the case is clear for me.

  • David Swager

    Very interesting, but can anyone decipher the hieroglyphics and tell me is that 14bit lossless compressed or uncompressed. The other issue never mentioned by anyone, anywhere is what the buffer size is. That is, how many images of lossless compressed 14bit for example can be shot before the camera has to actually start writing out to the card. Depending on your own specific burst patterns, this alone my obviate the need for superfast cards.

  • Back to top