Home · Register · Software · Software · Join Upload & Sell

Moderated by: Fred Miranda
Username  

  New fredmiranda.com Mobile Site
  New Feature: SMS Notification alert
  New Feature: Buy & Sell Watchlist
  

FM Forums | Nikon Forum | Join Upload & Sell

1
       2       end
  

D850 buffer much lower at 9 frames per second
  
 
gdsf2
Online
• • •
Upload & Sell: On
p.1 #1 · p.1 #1 · D850 buffer much lower at 9 frames per second


This was posted by another member on a battery thread but I think it deserves its own thread. He stated, correctly, that

Using the EN-EL18b battery greatly reduces the buffer capacity (when shooting in continuous high mode) of the D850 (see page 269 in Nikon Menu Guide).

NEF lossless 12-bit buffer is 170 shots with EN-EL15a but only 54 shots with the EN-EL18b

NEF lossless 14-bit buffer is 51 shots with EN-EL15a but only 29 shots with the EN-EL18b

I was wondering if this is because the faster frame rate is overtaking the camera’s ability to write files to the card and clear the buffer. I find it hard to believe it is a direct result of the battery itself.

Jerry





Sep 23, 2017 at 02:12 AM
Blakehfreeman
Offline
• • •
Upload & Sell: On
p.1 #2 · p.1 #2 · D850 buffer much lower at 9 frames per second


I'm sure the processing etc normally used to clear the buffer is being used to process the extra 2fps. #firstworldproblems


Sep 23, 2017 at 02:18 AM
arbitrage
Offline
• • • • • •
Upload & Sell: On
p.1 #3 · p.1 #3 · D850 buffer much lower at 9 frames per second


Thanks, I had read it on the other thread but I agree it deserves its own recognition as it will be an important consideration for those considering spending the extra $$ on 9FPS. I prefer to shoot 14-bit LLC but 29 shots is getting down into a buffer range I've had issues with on my previous cameras. Still most times it won't matter but 29 shots at 9FPS is only 3s of shooting....I prefer much longer than that for fast pace action, subject interactions and flight.

Has there been much testing of 12 bit vs 14 bit? Do you lose much at mid to high ISO or is it mainly a benefit at low ISO?



Sep 23, 2017 at 02:21 AM
gdsf2
Online
• • •
Upload & Sell: On
p.1 #4 · p.1 #4 · D850 buffer much lower at 9 frames per second


Blakehfreeman wrote:
I'm sure the processing etc normally used to clear the buffer is being used to process the extra 2fps. #firstworldproblems


What is discussed on this forum that is not a first world problem?
#drobvious

Anyway, back on topic, I don’t think clearing the buffer takes much processing power, but I am not an electrical engineer.



Sep 23, 2017 at 02:22 AM
snapsy
Online
• • • • •
Upload & Sell: On
p.1 #5 · p.1 #5 · D850 buffer much lower at 9 frames per second


Here's my best guess, as a firmware engineer who works on I/O systems.

First, the buffer depth's quoted by Nikon are a composite of both the internal buffer size (memory) and the expected throughput from that memory to the XQD. For example, when you half-press the shutter you'll usually see an estimate based on the actual buffer size without the benefit of media-card buffer unloading - on my D850 that produces an r20 in the viewfinder for 14-bit lossless compressed. The higher buffer values quoted in the D850 manual account for the ability to offload images out of that buffer to XQD. An XQD card is slower than the internal memory bandwidth (typically), so eventually the images in the buffer will start collecting faster than they can be written to XQD. The differential between the memory bandwidth and the XQD bandwidth represents the total estimated buffer depth quoted in the manual.

Regarding support of 9fps, there is likely a throughput bottleneck somewhere along the image processing chain or back-end interface to XQD, specifically the buffer memory itself or I/O bus into the memory. That buffer is populated (written) with images by the Expeed 5 ASIC with images read/processed off the image sensor. The buffer is depopulated and unloaded (read) by the XQD interface sending data to the card. This means the memory has two bi-directional consumers of bandwidth. Depending on the bank layout of the memory and ported interface into the memory there is a maximum aggregate throughput for which the memory supports reading/writing, ie total MB/s the memory can sustain. At 7fps there is likely enough idle period between writing images to the buffer memory such that the memory can sustain the expected throughput to XQD for the quoted nominal buffer depths. At 9fps the continuous throughput demands on memory are increased - with a fixed aggregate bandwidth limitation this means either the burst rate writes to memory will slow down (from 9fps), or the read rate from memory to XQD will slow down. I would expect Nikon designed the logic for the latter, which will permit the camera to maintain the advertised 9fps (for bursts), with a lower internal memory-port priority given to the XQD interface. The net result is a lower aggregate buffer depth since there is less memory or I/O bandwidth available to supply data to XQD, thus slowing the XQD writes and causing images to start collecting faster in memory.



Sep 23, 2017 at 03:07 AM
Hardcore
Offline
• • • •
Upload & Sell: On
p.1 #6 · p.1 #6 · D850 buffer much lower at 9 frames per second


snapsy wrote:
Here's my best guess, as a firmware engineer who works on I/O systems.

First, the buffer depth's quoted by Nikon are a composite of both the internal buffer size (memory) and the expected throughput from that memory to the XQD. For example, when you half-press the shutter you'll usually see an estimate based on the actual buffer size without the benefit of media-card buffer unloading - on my D850 that produces an r20 in the viewfinder for 14-bit lossless compressed. The higher buffer values quoted in the D850 manual account for the ability to offload images out of that buffer
...Show more

That's precisely what I was thinking as well Snapsy!



Sep 23, 2017 at 03:16 AM
Imagemaster
Online
• • • • • • •
Upload & Sell: On
p.1 #7 · p.1 #7 · D850 buffer much lower at 9 frames per second


That's exactly what I thought.

Dang, you thought faster than I did.



Sep 23, 2017 at 03:17 AM
gdsf2
Online
• • •
Upload & Sell: On
p.1 #8 · p.1 #8 · D850 buffer much lower at 9 frames per second


Thank you. I followed half of your answer and I think I agree.


Sep 23, 2017 at 03:28 AM
Stoffer
Offline
• • •
Upload & Sell: Off
p.1 #9 · p.1 #9 · D850 buffer much lower at 9 frames per second


arbitrage wrote:
Has there been much testing of 12 bit vs 14 bit? Do you lose much at mid to high ISO or is it mainly a benefit at low ISO?


Yes, Jim Kasson did a test on the Nikon D810:

http://blog.kasson.com/the-last-word/nikon-12-bit-raw-mode/

His conclusions;
ISO 200 and up, 12 bits is fine.
ISO 100, 12 bits is probably OK for almost everything.
ISO 64: use 14 bits. It won’t cost you much, and it’ll give you peace of mind




Sep 23, 2017 at 05:15 AM
snapsy
Online
• • • • •
Upload & Sell: On
p.1 #10 · p.1 #10 · D850 buffer much lower at 9 frames per second


FYI, I sent a message to Peter at NR with a link to this thread. Hopefully he'll post an article so more people become aware of it before they spend the cash for the grip+battery.


Sep 23, 2017 at 02:32 PM
 

Search in Used Dept. 



gdsf2
Online
• • •
Upload & Sell: On
p.1 #11 · p.1 #11 · D850 buffer much lower at 9 frames per second


The increase in speed is the only reason I would get the grip. I am ok with ergonomics when shooting portrait. I was undecided about the need for 9 shots/second for my shooting. The marginal increase in speed would be handy in select situations but the cost is not small. Now knowing the impact on the buffer size I am leaning towards no grip.


Sep 23, 2017 at 03:15 PM
JimKasson
Offline

Upload & Sell: On
p.1 #12 · p.1 #12 · D850 buffer much lower at 9 frames per second


snapsy wrote:
FYI, I sent a message to Peter at NR with a link to this thread. Hopefully he'll post an article so more people become aware of it before they spend the cash for the grip+battery.


Snapsy's explanation seems right to me. If that's the case, you're not actually losing any buffer size. On some Nikon cameras, you can set the frame rate on Ch. If that's the case with the D850, and you're concerned about buffer size, you could throttle the rate when you use the grip.

For me, the grip is going to be a win if for no other reason than to let me use my D5 batteries in another camera.

However, I don't have my D850 yet, so I can't test this.

Jim




Sep 23, 2017 at 03:15 PM
gdsf2
Online
• • •
Upload & Sell: On
p.1 #13 · p.1 #13 · D850 buffer much lower at 9 frames per second


Jim, I am not sure what you mean by “not really loosing any buffer.” I did not read that in his explanation.


Sep 23, 2017 at 03:20 PM
morris
Offline
• • • • • • •
Upload & Sell: Off
p.1 #14 · p.1 #14 · D850 buffer much lower at 9 frames per second


snapsy wrote:
Here's my best guess, as a firmware engineer who works on I/O systems.

First, the buffer depth's quoted by Nikon are a composite of both the internal buffer size (memory) and the expected throughput from that memory to the XQD. For example, when you half-press the shutter you'll usually see an estimate based on the actual buffer size without the benefit of media-card buffer unloading - on my D850 that produces an r20 in the viewfinder for 14-bit lossless compressed. The higher buffer values quoted in the D850 manual account for the ability to offload images out of that buffer
...Show more

I agree with you Snappy and as a one time systems engineer I'd hope that the tasks of focus tracking and exposure get priority over writing images to the card. If writing to the card were given priority then the result would be lots of cruddy images.

I also wonder why people want such large buffers. At 10 FPS 30 frames is 3 seconds which is a very long time for things to happen.

Morris



Sep 23, 2017 at 03:41 PM
chez
Online
• • • • • •
Upload & Sell: On
p.1 #15 · p.1 #15 · D850 buffer much lower at 9 frames per second


morris wrote:
I agree with you Snappy and as a one time systems engineer I'd hope that the tasks of focus tracking and exposure get priority over writing images to the card. If writing to the card were given priority then the result would be lots of cruddy images.

I also wonder why people want such large buffers. At 10 FPS 30 frames is 3 seconds which is a very long time for things to happen.

Morris


Depends on how long it takes the buffer to empty. 3 seconds all at once is a long time however 4 or 5 1.5 second sequences during the action is very common.



Sep 23, 2017 at 04:54 PM
bs kite
Offline
• • •
Upload & Sell: On
p.1 #16 · p.1 #16 · D850 buffer much lower at 9 frames per second


"I also wonder why people want such large buffers. At 10 FPS 30 frames is 3 seconds which is a very long time for things to happen."

This is what I have been thinking Morris.

But after an outing, I've never checked the time-length of my camera's bursts, or counted the frames.

Will try to note such things today.

Hope to be out for last couple+ hours before sunset.




Sep 23, 2017 at 04:57 PM
snapsy
Online
• • • • •
Upload & Sell: On
p.1 #17 · p.1 #17 · D850 buffer much lower at 9 frames per second


Here's a very rudimentary data-rate flow model I created in Excel to roughly depict how the D850's buffer depth might look over a continuous burst sequence. To follow it, scan across the "seconds elapsed" to see how the D850's internal buffer gradually starts filling up as the XQD gets backlogged with images to write. The point at which the number of images in the buffer reaches 20 (actual memory buffer depth) marks the effective buffer depth (memory + XQD).

Note the D850 manual says the 14-bit lossless buffer depth is 51 images with the EN-EL15a (7fps) and 29 images with the EN-EL18b (9fps), whereas my rough model shows 52 images @ 7fps and 38 images @ 9fps. The difference of my model's 38 images @ 9fps vs Nikon's quoted 29 images @ 9fps may be the effect I discussed earlier about the bandwidth bottlenecks.








Sep 23, 2017 at 05:01 PM
gdsf2
Online
• • •
Upload & Sell: On
p.1 #18 · p.1 #18 · D850 buffer much lower at 9 frames per second


Gotta love a spreadsheet model. Thanks. This is helpful.

Jerry



Sep 23, 2017 at 05:24 PM
la puffin
Offline
• •
Upload & Sell: Off
p.1 #19 · p.1 #19 · D850 buffer much lower at 9 frames per second


morris wrote:
I also wonder why people want such large buffers. At 10 FPS 30 frames is 3 seconds which is a very long time for things to happen.


Since it's written up as a "commercial sports" camera (whatever that means), three seconds is not very long. Take for an example, a runner sliding into home plate. If you start shooting as he becomes airborne and follow through to the slide and reaction, that's a lot longer than three seconds, and any of those frames could be "the one". During the Memorial Day fight between the Giants and the Nationals, I hammered away on my D5. That went on for a long time and I got a lot of great shots out of that.



Sep 23, 2017 at 06:11 PM
ilkka_nissila
Online

Upload & Sell: Off
p.1 #20 · p.1 #20 · D850 buffer much lower at 9 frames per second


Compressed NEF (instead of lossless compressed) is also a good way to increase burst depth. Nikon implements the (lossy) compression in such a way that the least significant bits are left out in highlight pixels. The LSBs are basically noise in such pixels (photon shot noise is proportional to the square root of the detected number of photons which is high in highlights). Raw files store the data without a tone curve applied. When the image is used in practice, typically an s-shaped curve is applied which compresses highlight contrast. In the final converted image it would be extremely difficult to see any difference due to the (lossy) compression of NEFs. If one is shooting action, in most situations one is using a moderate to high ISO and there the use of compressed 12 bit NEF is perfectly reasonable. One can then switch to 14-bit lossless for situations where there isn't a need to shoot bursts and when the subject either isn't moving or there is bright light. The main problem is remembering to do so.


Edited on Sep 23, 2017 at 06:27 PM · View previous versions



Sep 23, 2017 at 06:26 PM
1
       2       end






FM Forums | Nikon Forum | Join Upload & Sell

1
       2       end
    
 

You are not logged in. Login or Register

Username     Reset password