Nikon D4s buffer capacity increased when you use the new Sony G Series XQD version 2 memory cards

After Sony announced their G Series XQD version 2 memory cards, Nikon updated the buffer performance of the D4s camera - in the tables below the left "Buffer Capacity" column shows the approximate number of pictures that can be stored with the older Sony 32GB S Series XQD memory card (QD-S32E) at different image quality, image size, and image area settings. The right (yellow) "New Buffer Capacity" column shows the approximate number of pictures that can be stored with the new Sony 64GB G Series XQD memory card (QD-G64):

Nikon XQD memory card buffer capacity FX 24x36


Nikon XQD memory card buffer capacity DX 24x16 *2


*1 Includes images taken with non-DX lenses when On is selected for Auto DX crop.
*2 Includes images taken with DX lenses when On is selected for Auto DX crop.
*3 All figures are approximate. File size varies with scene recorded.
*4 Maximum number of exposures that can be stored in memory buffer at ISO 100. Drops if Optimal quality is selected for JPEG/TIFF recording > JPEG compression or auto distortion control is on.
*5 Figures assume JPEG/TIFF recording > JPEG compression is set to Size priority. Selecting Optimal quality increases the file size of JPEG images; number of images and buffer capacity drop accordingly

The new Sony XQD memory cards are available for pre-order at B&H and Adorama.

This entry was posted in Nikon D4, Nikon D4s and tagged , . Bookmark the permalink. Trackbacks are closed, but you can post a comment.
  • Bob

    How is the D4 impacted? Improvement?

    • My guess would be yes, but I could not find any data. Nikon only published the results from the D4s.

      • Brent Busch

        I know it’s a year later, but has anybody ever seen whether the D4 benefits or not?

  • itcrashed

    Confused. Isn’t buffer memory a fixed thing? I’m lost on how a buffer can increase with different cards unless the processor actually uses some of the card memory. What am I missing here?

    • itcrashed

      Or is the word “capacity” is used incorrectly. Maybe the proper word is buffer “performance”

      • They mean the capacity of the buffer, not the memory card. This is Nikon’s table btw.

        • MB

          I may be wrong but it seams to me that they are talking about the capacity of the camera as a system … or how fast it is able to process and store images … lets assume that buffer size is large enough to store 50 images, but if they are able to store only additional 10 images before this buffer gets filled total capacity would 60, but if they are able to store 16 images on the card the system capacity would be 66 images before it stalls … or something like that …

          • peterw

            Indeed, it would make sense if Nikon considers photographers as end-users, rather than technicians, and refers to them in terms of the system.
            Considering Nikon F being a system for photography.

          • fiziks

            You are correct. While the photographer is holding down the shutter button trying to fill the buffer, the camera is writing images from the buffer to the card. But you can fill the buffer faster than you can empty it. So the chart is a bit misleading. The numbers aren’t really the buffer capacity, but how many images you can take before the process filling the buffer catches up to the process emptying the buffer.

            Think of the buffer as a buffer as a bucket with a hole in the bottom of it, and another bucket underneath, which is the QXD card. You can dump water into the top of the bucket with the hole in it faster than it can drain through the hole in the bottom, so eventually, you will fill the bucket to the rim, but not before some amount of water has also drained into the bottom bucket. By Nikon’s reckoning, when you reach that point where you have water to the brim of the bucket with the hole in the bottom, they count the amount of water in both bucket’s as the “capacity”… Not a perfect analogy, but hopefully you get the picture.

    • exactly, my thinking is going same way… i am pretty sure that Buffer is still same. But now the DSLR can release alocated internal memory faster thanks to new XQD and therefore is capable manage more frames in a sequence…. tricky from Nikon to claim that 🙂

      • Ken Elliott

        Nothing tricky at all. The faster fill rate means more shots until the buffer is full with the new cards. This means the camera has a high throughput than the previous cards could handle, so Nikon had to (in effect) de-rate the camera because of the cards. They are simply telling us the card was the bottleneck in some situations, and here’s where to expect improvement. It seems reasonable.

        Could have said it better? Yes.

        • Ken, as a software developer and IT expert with 28 years of experience I know how it works. I just tried to explain it in a short way 🙂 It’s clear that there is no bigger RAM memory just by adding new XQD card. What improves is just throughput as you mentioned.

          • Ken Elliott

            This statement let me to believe you didn’t understand:

            “i am pretty sure that Buffer is still same.”

            Since it’s the same hardware – it has to be. This led me to believe you weren’t familiar with the hardware.

            BTW – I’m a C and assembly language programmer, and network architect with over 40 years experiece. Saying the term “RAM memory” will raise eyebrows. Just say RAM (random access memory).

            • well, i am not a camera designer, but i am pretty sure that Buffer in DSLR is RAM type of Memory. OK nowadays you can say DDR3 more or less.

            • Ken Elliott

              It’s like saying “picture camera”. Saying RAM memory is like saying “random access memory memory”. That’s why hardware guys will look at you funny. 😉

              DDR3 is an interface and performance spec, not a memory type. You can’t write to ROM (read-only memory), so it has to be RAM (random-access memory). It could be static RAM (fast/expensive) or dynamic RAM (cheaper and common). So you do indeed understand how it works – no question.

            • Hey, in fact you are right. My expressive capabilities are quite limited as i am not natively english speaking person 🙂 In my language we usualy say “RAM memory” despite the fact its not 100% correct.

            • Ken Elliott

              Opps – after reading my own words, I come across as unduly harsh. You did perfectly fine for a non-native speaker. Please accept my apologies.

            • no problem 🙂 You just tried be scientific perfect :-))

            • MyrddinWilt

              It is certainly RAM in Nikons. The poor Canon chaps are still using magnetic drum memory.

            • good joke, but i didn’t start this kind of discussion and arguments. Anyways it would be funny to see it in real 🙂

            • Andrew

              I have no argument about the increased transfer speed resulting in less demand on the camera’s internal RAM/buffer.
              I guess there is still the possibility that Sony have implemented either SRAM or DRAM in their new cards so as to improve the system’s combined (total) buffer size, thus enhancing the experienced transfer speed from camera to card. The flash in the card remains relatively slow to write to so in-card SRAM or DRAM could help overall buffering and effective performance.

            • Andrew

              Hi Andrew, that makes the two of us 😉

            • Andrew

              […from Andrew]

              Is an infinite buffer is the D4s’ future?

              I don’t think you should look at this as increasing the buffer size in the sense that the XQD card does not expand the systems buffer. The systems buffer stays fixed but the faster XQD card allows the camera to write data to it at a much faster rate than previously.

              Now as data is being written to the XQD card (as aptly described by previous posts), it is emptying the data from the internal system buffer, but not fast enough. And so the internal system buffer gets eventually filled.

              When the internal system buffer gets filled, the camera slows down dramatically and basically its performance is dictated by the speed at which data is written to the XQD card. You would have to stop taking new pictures to allow the system buffer to empty out.

              I think the question now is whether the D4s has a theoretical limit in its interface with the XQD that is slower than the internal buffer speed. If it does not, and if Sony’s XQD specification supports speeds up to the internal buffer speed, then theoretically at some point in the future, you will have an unlimited buffer that is limited only to the capacity of Sony’s XQD card.

            • peterw

              There is nothing theoretical about an unlimited buffer..
              This means that the buffer is not nescessary (for storage, probably for rearrangement it still is). You could slow down the camera to get there…. All data can be writen directly to final storage medium. Storage medium? Ah, we are talking about QXD at the moment.

              It is nice that its there.
              (But I wonder who would need it? A camera on a rail trying to keep up with Usain Bolt doing 200m? Wouldn’t we use video to record that? Who needs to analyse 200 perfect 16 MB pictures of Usain Bolt at 10 fps? Perhaps ask catwalk photographers?)

            • Andrew

              You are missing the point completely 😉
              The image quality obtained from individual shots filling the buffer is orders of magnitude greater than individual frames from a video shoot that is spraying shots at 60 fps (frames per second). And besides, in a 20 second video, you are talking about 1,200 shots to go through, I am certain most people would find a better way to spend their time than sorting through that many pictures.

              The theoretically limitless or unlimited buffer is for the sole purpose of taking shots at respectable intervals for as long as the photographer wants to without an arbitrary limit being imposed on them.

            • peterw

              Exactly what point did I miss, Andrew 😉

              About arbitrary limits being imposed? 10 fps is rather a strong limit. Set it lower and you can shoot forever. Or do you mean the problem that D4 will stop after 200 actuations? (personaly, that would not be something I would try to find out).

              There will always be limitations when maximising recording interval, amount of data and amount of recording time. But it depends on choices of Nikon/Canon whether the buffer will be limiting factor or the shutter or the processor or whatever. Will those choices be arbitrary?

              For now we are looking at 16 MP, somebody could want to get 36 MP. Why not.
              For now 10 fps is considered satisfactory for stills, future will bring higher rates.

              Perhaps in future someone wants to follow Kenenisa Bekele or Dennis Kimetto instead of Usain Bolt?
              (It’s a good thing these guys keep on running faster and faster: you and me wouldn’t like to analyse 1200 shots of a 20 second sequence, I certainly would hate to analyse a 2 hours and 3 minutes sequence.)
              (For training purposes, I am quite sure that there are 20s video recordings of Usain Bolt, at perhaps even higher rates than 60 fps)

            • Andrew

              Good points 😉

              I think a possible limiting factor may be heat generation/dissipation and so the number of actuations at a high frame rate may be restricted before a critical threshold is reached. An incident too close to home occurred recently when a laptop was running intensive graphics operations (for no good reason other than fun) for a prolonged period of time. It practically fried the motherboard. The laptop (computer) never fully recovered and had to be discarded.

              Sometimes we think that the camera manufacturers are holding back certain features or performance values when in some instances there are real engineering limitations in terms of reliability. I have at times overlooked these physical constraints thinking that the manufacturers are imposing marketing policies by doing things like deliberately restricting the buffer size. But what are we to do when they are withholding such important (smile!) information, it leads us to speculate endlessly 😉

            • peter w

              “… it leads us to speculate endlessly 😉 ”
              yeah, I love it…. 😉

              (expensive fun 🙁 … )

            • Me

              Why are you talking about physical constraints to hardware and things like heat? What does that have to do with anything. It’s DIGITAL.

              The next thing you know, you’ll be saying that there are limits imposed on the system by the battery or something.

              (Yes, I’m kidding.)

            • TheTruth

              @marcelspeta:disqus Have you considered the fact that MAYBE, just MAYBE the initial limit on the buffer size was not limited by the buffer capacity itself. It could be the buffer could already physically hold that number of extra shots, BUT because the slowest link in the chain was writing to the card, the firmware would limit the user to X number of shots stored in the buffer.

              Why would they do that I hear you ask?
              Well, you have a buffer that can hold 200 shots, but Nikon limit it to 100. Why? because your memory card is slow to write to. If it took 10 seconds to clear those 100 shots then it would have taken 20 seconds to clear down those 200 shots if you allowed it to work to full capacity. and Nikon didn’t like making the user wait 20 seconds it was deemed unacceptable.

              So now that the write speed has increased on the new memory cards, suddenly it only takes 10 seconds to now write 200 pictures to the card, and it’s done within the acceptable ‘user waiting’ limit. So yes from the user point of view the capacity has in fact INCREASED.

              All numbers are just pulled out of thin air as if by pure magic!

            • RBR

              I doubt that is the case because I do not believe Nikon would have shipped a camera with grater buffer capacity than could be used. They would not have installed the “unusable” buffer. They are a cheap bunch as the users of the lesser bodies will tell you. Nikon keep “cheating out” on buffer capacity which limits otherwise decent cameras in a number of applications.

              I think everyone believes that there is some sort of translation problem that Nikon needs to clarify. Until they do so, all this is speculation. It’s unfortunate that this statement wasn’t checked by someone in the U.S. before release to avoid this confusion.

            • Andrew

              Transitioning from Borland C and AT&T Unix C on their 3B2 Minicomputer to Visual Basic 1.0 was a major step forward using Visual Basic’s visual programming tools and its event driven paradigm. But Microsoft Visual C++ 1.0 was a big disappoint. The Integrated Development Environment (IDE) for Visual C++ was nothing like the elegant Visual Basic IDE. So I abandoned it and moved on to Power Builder who was later acquired by Sybase.

              Fortunately Microsoft fixed their IDE in Visual Studio for C#, though not completely in Version 1. Visual Studio 2005 for .NET was when their development tools became elegant again. Visual Basic though did lose its way in the earlier versions of Visual Studio for .NET and many VB developers did not adopt .NET and stayed at VB 6.0.

          • TheTruth

            OH SNAP.
            I do agree with your assessment on the situation, BUT when people start spouting of ‘I do this job for this many years’ it means SHIT ALL!
            Seriously I moved into IT 5 years ago and know lots more than most software devs in my business. Especially since most dev’s know NOTHING about memory control these days because they are fart around in highlevel stuff and think they know their stuff.

            You could be doing the same job for 100 years but if you don’t learn and simply potter along the number of years experience means NOTHING!

            • Andrew

              Let’s not be too hard on them. People have pride in what they do and what they have accomplished. Just like you have pride in knowing and learning things in the years you have been in the IT field. Sometimes we overstep our bounds in this area, I included. So all we have to do is give a gentle reminder that our views should only be supported by evidence and sound reasoning and not because we are some authority figure.

        • stormwatch

          Analogue to CPU’s microarchitecture this is technically speaking an very good L2 cache upgrade, directly helping to the L1 – but in the reverse way!

    • it takes less time to dumb the buffer to the memory card and the buffer can process more images

      • *dump 😛

      • milos

        has to be true for other models as well. correct?

        • it should be

          • me

            only if the camera is fast enough to take advantage of the extra speed of the new card.

        • RxGus

          It’s true on my D750. Using a 95 mb/sec card vs a 45mb/sec card increased my buffer by 5 shots in 12 bit NEF (11 to 16)

          • Jay Doggett

            How can one measure this?

            • RxGus

              I reformatted my card; put the camera on continuous high shooting; and fired until the buffer filled and the firing rate slowed down… 11 shots

              I put in the 95 mb/sec card and did they same thing and got 16 shots before the firing rate slowed down.

            • Jay Doggett

              Thank you!

        • RBR

          Tests with the D7000 with different speed cards showed that it did not write fast enough to take advantage of the full capabilities of the Sandisk 95 mB/sec cards. The knuckleheads at Nikon never did say what the fastest write speed any of the DX bodies could benefit from.

          There is one matter to be aware of though. I don’t have the links handy, but the Lexar SD cards specifically wrote at speeds significantly slower than the Sandisk SD cards so you have to check what the actual capability of a media is.

          • Jay Doggett

            I wonder if that’s true of the D7100. I’m going to perform the experiment RxGus describes over the weekend, I have a 95mb/sec card

    • The buffer is fixed however the camera is capable of writing to the card at a much faster rate.

      The effective change here is that the buffer will fill up just the same but it can dump faster to the card.
      Therefore the camera will reach these new maximum shot capacities as the buffer can now clear to make way for more images.

      Previously the camera has been limited to the card’s ability to handle large amounts of continuous data from the buffer without choking.

    • You confused yourself, because you’re talking about buffer memory while the post is about buffer capacity. The former is a fixed value measured in mega/gigabytes. The latter is measured in the number of pictures that can be stored before the buffer is full. Since this is I/O (input/output) speed dependent, with the faster cards this capacity has increased.

      Think of it this way. Your buffer has incoming and outgoing data and this actually happens in parallel. Writing out to the card starts as soon as there is incoming data. Say in 5 seconds you have 1000Mb coming in and with previous XQD in the same time you get 500Mb out. With the new card, you get 800Mb out. So in the same time frame (5 secs) you have 300Mb more space for incoming data -> higher buffer capacity.

      • UA

        Well explained.

        Like, you pour water with 5 litre pot in to empty 50 litre well and your friend takes it away with 2 litre pot. If you work at the same phase, you will fill the well with 17 pots. If your friend changes to 3 litre pot, the well would fill up after you have poured 25 pots. It’s still the same 50 litre well.

        Simply put: the buffer cleans up faster from the existing frames ->
        increase in capacity. So earlierly D4s was able to fill the buffer
        faster (with less frames), as the frames spend more time in the buffer.

      • itcrashed

        Thanks for the clarifications. Semantics confused me. But I’m totally onboard with it. Time for Sony to take my money! I was ready to change out my OG N series for some S series cards but it appears these new ones are roughly the same price!

        • Marcelo Tezza

          I think the camera makers should have in cameras manual’s the maximum transfer speed that the camera is capable of in MB/s, and that speed will decrease based on the cards speed been used and
          I think on every memory card should have minimum and maximum write and read speeds in MB/s wrote on the card itself.
          I have not found any ISO norm wich specifies memory cards speeds.
          Some brands have missing specs, and this nomenclatures of SD SDHC SDXD UHS-I UHS-2, X1000.
          Why so many?
          This seems to me is like a bad intention to make consumers confused. Too much information when this could be simple as MB/s only. If has any specific information this should be noted in the box. Just simple as that.

    • Desmond

      ask canon, they increased the buffer by about 60% on their 5d2 and 7d on their 25th firmware update. none of the canonites raised an eyebrow.

      • peterw

        … Canonistas can’t raise their brows 😉
        something with DNA

        (sorry, couldn’t resist… my best friend owns a Canon… my daughter owns a Canon, I myself have owned Canons)

      • RBR

        The one that Nikon really should be trying to figure out is how Canon increased the frame rate of the 1Dx from about 10 to 14 by a firmware update.

        • Neopulse

          Umm…. no? It’s 12 fps normal and 14 fps with the mirror up. You are gravely mistaken.

  • Eric Calabros

    133×15.4= 2101mb >> 200×15.4= 3080mb %46
    78×19.3= 1505mb >> 102×19.3= 1968mb %30
    There is enough room for additional 50 frames for 14bit, I dont iunerstand why its just 102

    • Marcelo Tezza

      Maybe it’s a processor limitation, even if it is by temperature security.

      • peterw


        it makes sense,
        it is clear
        (clear as a drop of water)

        • Marcelo Tezza

          Thx peter!

  • Robert

    I comment Nikon for equipping the D750 with 2 SD Card slots.
    However, when will Nikon realize that UHS II is on the market. Some cameras could reveal the same effect as the D4s with the new XQD with fast SD / UHS II cards. Does Nikon again fear, that this improvement would nag at another camera’s sales?

    • m35g35

      The issue has to due when UHSII was developed versus when Nikon started development of the D750. I suspect Nikon had access only to UHS I SD cards at the time. Secondly, the D750 probably can’t saturate the bus to take advantage of the UHSII. Besides, as in the computer world you are only as fast as your weakest link. Which is it here? I don’t know. I would guess the CPU.

      • Robert

        How long does the Fuji XT1 with UHS II already exist? I guess at least a year or so as Fuji recently launched the second iteration. And they had to do previous development too, not just Nikon.
        The D750 certainly is a great camera, but since it is distiktly above the D610 and different to the D810 due to the sensor, the questions are: why an AA-Filter? Why no 1/8000 sec? Why no AF-On button? Why no X-Contact?
        In the recent past the impression manifests in me, that Nikon often develops with the handbrake pulled. Or, as far as many lenses are concerned, with primarily one customer group, namely wedding photogs (all about bokeh), in mind.

  • Too bad no Nikon can click more than 99 images in a row, due to the stupidest feature ever created… 🙁

    • pforsell

      What feature is that? My D3s and D3x both shoot more than 99.

      • Sorry I mis-spoke, I meant 100 images max.

        Try holding down the shutter for more than 100 images without letting up. It’s a menu option limitation, and the greatest option is 100, when it REALLY ought to be “unlimited”…

  • KnightPhoto

    Please put this in the unicorn D400 😉

  • Confused

    Under what conditions is this true, when using only the XQD card slot? Are the XQD and CF bus separate, so a slower CF card won’t slow down writing to the XQD card?

  • Spy Black

    That’s misleading. The buffer capacity hasn’t been increased, merely the throughput rate.
    The frame stills only goes to 11…

    • ArrGeeBee

      The buffer’s throughput capacity has increased, rejoice! Oh, but then nothing Nikon does works for you does it?

    • Spy Black

      No, the buffer’s throughput RATE has been increased, not the buffer’s capacity.

  • torwag

    Nikon (and other brands) are cheating here, we might need some clear definition of buffer capacity.

    Take this to the extreme:
    If Nikon releases a Nikon D1000 with only 12MP and only 2FPS, but using the same card interface, they could claim unlimited buffer with the new cards, however this camera does not at all have any physical memory buffer. They could simply dump the images directly to the card at that speed.

    It would be more fair to give the real buffer size. Say, how many images with best quality settings can be taken in demo mode, with no card inserted. That would tell that one can at minimum make X pictures of the highest quality in a row at highest speed, independent of any card combination.

    Everything else is a lucky combination of card speed, interface speed, image size and max. framerate. Hard to predict any real number from that.

    @NR: Any chance that we can find out which current (and older) camera models have the card interface abilities to make use of faster cards?

  • Tippy35075

    Wish the SD card slot would have the same potential.

    • Neopulse

      Well UHS-II technically is on it’s way.

  • RBR

    This makes no sense. Changing a card can not change the buffer size of the camera. Improving the write speed might help clear the buffer more quickly if the camera is capable of utilizing that speed. A firmware update is potentially capable of increasing the frame rate (e.g. Canon 1Dx), but it doesn’t change the physical buffer.

  • What’s more curious to me is why there’s a ceiling limit of 200 shots regardless of type (e.g. 12-bit loss-less compressed NEF has same max number as Small Basic JPG).

    One could say that it takes as long to write a larger raw file as it does to process & write a smaller jpg but I’d have thought there’d be at least /some/ wiggle room for a difference in writing speeds of the differing JPG types (i.e. from Large Fine to Small Basic).

    I wonder if they’ll come up with a FW update to make better use of the improved throughput? I’d guess not as they’d have already done it & shouted from the roof-tops about it…

  • Bellisimo

    Wish the D7100 could be improved too…

    I am primarily a sports photographer. To improved the IQ I got with my D300s I had to switched for a K-3 for this buffer problem. If you want to shoot wildlife or sports, and you are Nikon, you stick with a D300s and D700 or you go BIG MONEY FF with a D800 or D4 … too bad !

  • Neopulse

    I swear, sometimes if you don’t make a post showing the benefits of XQD, people will still believe it’s crap and prefer CF over anything.

  • Long ago, (reminisce the old days), I found that with a different Camera and Memory Card than what is discussed here that a faster Card does give you more Shots (and fewer lost Frames when recording a Video).

    It is the Card’s WRITE speed and not the READ speed that is what matters. If the Card will not accept Data quickly enough then the Memory Buffer starts to fill, when it is full your burst is finished (or your Video Codec must miss that Frame).

    Note in this old Article that the Author mentions the Cards he used for testing and that slower Cards would not produce the same result: .

    When you buy a New Card for your Camera always buy the fastest you can afford since when you get a new Camera you will want to be able to use the old Card — UNLESS you only buy one Card for each Camera and keep it only with that Camera, then you ought not to pay more for something you do not need.

    Camera Manufacturers _could_ have made the Camera’s Software allow you to simply sit on the Shutter Button and that when Shots _could_ be recorded then they would be. I guess they thought a stop-start action would disconcert the Customer so they made it that when the Buffer is full you are done (how quickly it fill is a function of your Memory Card’s write speed).

    IE: This is not new Info, but the “Chart” is “new”.

  • Back to top