D850 Buffer

Killerbob

I just got my D850, including battery grip and EN-EL18. Now, however, I find that at least one thing does not live up to my exceptions...

The FPS is great at 8-9/sec. I shoot RAW (FX, L, 14-bit, lossless compression), and JPG (fine compressed). RAW on XQD, and JPG on SD. The buffer in this config is supposedly 51, but I am finding that when I just press the shutter, it takes 24 shoots, and then it starts slowing down.

What can be causing that? The cards, settings, the lens?

snakebunk

How is it working if you only write raw files (no jpg)? From what I've read it only says that it can buffer 51 raw files.

Killerbob

I thought about that, and I just ordered a new SD card, UHS-II. In the meanwhile, I will try your suggestion out.

Pistnbroke

Make sure your SD card has two rows of connectors on the back and use JPEG basic.

BVS

In addition to what others have said, in earlier cameras settings that add extra processing to JPGs like High ISO NR, Active D-Lighting, Distortion Control, etc. were know to limit buffer depth. While you're waiting for your new card you might try disabling all this stuff and see if it makes any difference.

Killerbob

Even with shooting RAW only, the buffer i merely 14 shots. There is something funky going on...

BVS

What XQD card are you using? Aren't there slower ones and faster ones?

Maybe also double check you're set to 14-bit compressed and not uncompressed. I know you said you had it set to compressed though. Or, maybe try changing to uncompressed and see what happens. Would be funny if Nikon got the names backwards or something. I'm just grasping at straws though.

Killerbob

I am using the M series XQD card from Sony, which is supposedly one of the fastest, and I have tried most combinations of compressed vs. uncompressed.

Killerbob

Does anyone have a D850 I can compare with?

Sports

Sony M series is fast for reading but not for writing, I'm afraid. It could easily explain the problem.
I'm using Lexar's fastest cards for the D500, and have never tested the buffer depth.

Killerbob

Sorry, I was wrong. That was my old XQD card. The one I have in the D850 is the newer G series, 128GB...

BVS

Yeah, I think that's the problem. The M series is only 80 - 150MB/sec write speed. Compare with the G series which is 400MB/sec.

Killerbob

As I already said, I have the G series XQD card...

WestEndFoto

Did you try XQD only?

Killerbob

Yep - I reset the camera all the way back to square one, and with only an XQD card (Sony G series), FX L, RAW L, Lossless compression, 14-bit, the display says "r22", and I actually get 24-25 shots before it starts slowing down.

I talked to to Nikon support, and they actually were able to recreate on their D850, and hence confirm that there is an issue. They unfortunately only had a slower XQD card, but there is indeed an issue, and they will keep me updated on the progress. They also said;

The Rxx buffer capacity displayed on the camera is based on an average memory card speed (approx. 100Mbit/s), while the capacity listed in the Manual was calculated with the fastest available card (aprox. 440Mbit/s). This means that it is expected that these numbers do not match in some cases - however you should be able to take approx. 50 images with your fast card in Ch mode before the camera slows down, even if the screen shows R22.

Also, please note that the total active buffer capacity depends on the buffer fill / write rate balance, and hence the file size and fps speed you are shooting vs. card speed. When you attach the EN-EL18 battery the Ch speed increases, which is reflected in the reduced buffer capacity display. This would be normal then.

So I guess there is nothing to do than wait. In the meanwhile I can confirm this is a very nice camera, and that it is ALL THAT...

WestEndFoto

They said you should be able to take 50 but your test got to 24-25. How do you reconcile that? Or can you?

Killerbob

Sorry, something was funky, so had to copy/paste into new comment...

Yep - I reset the camera all the way back to square one, and with only an XQD card (Sony G series), FX L, RAW L, Lossless compression, 14-bit, the display says "r22", and I actually get 24-25 shots before it starts slowing down.

I talked to to Nikon support, and they actually were able to recreate on their D850, and hence confirm that there is an issue. They unfortunately only had a slower XQD card, but there is indeed an issue, and they will keep me updated on the progress. They also said;

The Rxx buffer capacity displayed on the camera is based on an average memory card speed (approx. 100Mbit/s), while the capacity listed in the Manual was calculated with the fastest available card (aprox. 440Mbit/s). This means that it is expected that these numbers do not match in some cases - however you should be able to take approx. 50 images with your fast card in Ch mode before the camera slows down, even if the screen shows R22.

Also, please note that the total active buffer capacity depends on the buffer fill / write rate balance, and hence the file size and fps speed you are shooting vs. card speed. When you attach the EN-EL18 battery the Ch speed increases, which is reflected in the reduced buffer capacity display. This would be normal then.

So I guess there is nothing to do than wait. In the meanwhile I can confirm this is a very nice camera, and that it is all that...

Killerbob

I won't be happy if it turns out to be that the D850 buffer is only capable of holding 24-25 shots.

Nikon confirmed that the D850 they had access to had the same issue. Since Nikon obviously had it hold 51 shots at some point (otherwise they wouldn't have said that was the limit in the manual), I am sure they will get it fixed. It could be an inconsistency with the Sony XQD card, or something else, which will probably get fixed in the next FW.

BVS

Hopefully it's not that you only get 50 when doing 7fps and less when doing 9fps.

Killerbob

No, it’s just the display which is awfully off. In reality you get more shots when using 7fps.

BVS

What I mean is that Nikon advertised the camera as being able to do 50 shots and never said anything about fps affecting that. If it turned out that the 50 shots was only at 7fps, and you get significantly less when doing 9fps, then there may be a lot of pissed off people, especially if they paid $1000 or whatever for the grip, battery, and charger. Hopefully that's not the case though.

Coastalconn

After your reset did you check to make sure everything was turned off as BVS mentioned above.. active D, distortion control, etc...
Have you formatted the card? If I. Have a glitchy card I format first in the computer and then in camera.
Could you tell me the r number in 1.2x crop mode both in lossless 12 and 14?
If the buffer isn't as good as Nikon stated I will not be upgrading to the D850 anytime... Especially if 9 FPS makes it worse...

Killerbob

I did reformat, and I did check that all was set as explained on page 362 in the D850 manual... NEF (RAW), Lossless compressed, 14-bit, Large, ISO100, XQD card, No JPG, Sec. Card is set to Overflow.

Nikon support did exactly the same, and experienced exactly the same on their setup. Difference was that my XQD card was significantly faster, and I had the EN-EL18 battery in the MB-D18.

The buffer gets the pictures faster with the MB-D18 battery, and hence fills up faster, and hence cannot offload these on the XQD card fast enough to be able to hold 51 shots. Pure math...

BVS

Killerbob said:

The buffer gets the pictures faster with the MB-D18 battery, and hence fills up faster, and hence cannot offload these on the XQD card fast enough to be able to hold 51 shots. Pure math...

While I agree with your logic, if 9fps is about 29% faster than 7fps, shouldn't you still be getting around 35~36 pics? Only getting 25 or so is a 50% drop which seems like something is wrong. Of course, math isn't necessarily my strong point though.

Just curious, how many shots do you get at 7fps? Is it close to the advertised 51?

snakebunk

If the camera can write 5 fps directly to the card, then 9 fps should fill up the buffer twice as fast as 7 fps. Maybe this is the case.

Killerbob

BVS said:

While I agree with your logic, if 9fps is about 29% faster than 7fps, shouldn't you still be getting around 35~36 pics? Only getting 25 or so is a 50% drop which seems like something is wrong. Of course, math isn't necessarily my strong point though.

Just curious, how many shots do you get at 7fps? Is it close to the advertised 51?

No, you're right, and if that was the only thing I guess that's OK.

Without the MB-D18/EN-EL18, at 7fps, the buffer probably should then be around 51, and it's not - it's close, perhaps 43 before starting to slow down. There is a problem, verified by Nikon, which means the buffer is either not filling up correctly, or not writing to the card correctly. With the MB-D18/EN-EL18, the issue is only exacerbated, because the fps is soo much higher...