I did some tests and thought I would share some of my results on how various settings affect how the buffer will fill and how it clears when writing to the memory card. Together these two man concerns determine how many shots you can take in rapid succession.

There are two things that will affect the number of frames you can take in rapid succession:
1) Camera settings that affect the buffer number (rXX)
2) Card write throughput

Each of these is actually driven by multiple factors.
1) Camera settings

-RAW settings - The best RAW r number I could get was r11
It mattered very little whether or not 12 or 14 bit was selected
or whether lossless was selected. Even RAW + JPEG made an
insignificant impact on the buffer.

-Auto Distortion ON/OFF ON results in a r08 no mater what
other settings are selected! (even for small/basic JPEGs)

-JPEG size/compression
Large/Fine-Normal-Basic = r15, r22, r33
Medium/Fine-Normal-Basic = r20, r33, r41
Small/Fine-Normal-Basic = r33, r58, r99

2) Card write tests
Method -
D7000 Nikkor AF-S 18-105 VR ON Manual 1/200s F/5.6 ISO 800 AF-C
Card used - Transcend SDHC 16Gb Class 6
Set Mode dial to CH.
Take as many shots as required to bring the buffer down to r=0.
Pause shooting until the buffer empties,
Then take another single frame .
Look at the timing of the last frame of the burst to the single frame when the buffer is cleared.

23s - r10 RAW 14BIT Lossless A-Dist-OFF
23s - r08 RAW 14BIT Lossless A-Dist-ON
23s - r11 RAW 14BIT Compressed A-Dist-OFF
20s - r11 RAW 12BIT Compressed A-Dist-OFF
20s - r11 RAW 12BIT Lossless A-Dist-OFF
19s - r08 RAW 12BIT Losseless A-Dist-ON
26s - r10 RAW+FINE L Lossless A-Dist-OFF
26s - r10 RAW+NORMAL L Lossless A-Dist-OFF
22s - r10 RAW+BASIC L Lossless A-Dist-OFF
10s - r15 FINE L A-Dist-OFF
09s - r22 NORMAL L A-Dist-OFF
xxs - r33 BASIC L A-Dist-OFF (Could not run down buffer)
07s - r20 FINE M A-Dist-OFF
xxs - r33 NORMAL M A-Dist-OFF (Could not run down buffer)

EDIT to add****************************************
Different Card- SanDisk Extreme 4Gb 30MB/s Class 10
09s - r10 RAW 14BIT Lossless A-Dist-OFF
***************************************************


As expected, the clearance time seems to be a function of the file size and whether there was a RAW + JPEG shot. The card speed also comes into play here but I only tested one card type. Similar results should be seen even with a faster card but the relative effects should be similar. When I said I could not run down the buffer, this meant that I took many shots and the buffer was emptying but I gave up clicking away to save some shutter activations that would not yield additional information.

If your camera settings are such that you never get a large r number the card speed is almost a trivial discussion as the buffer will fill and slow way down.

With an r08 in the display, I could usually get about 10 shots before the fps dropped as some data is cleared while shooting making room for a couple more frames.

Dean
Phoenix, Arizona USA
Nikonians Team Member
Website: The Splendid Silence of Light

Recent Trips: Grand Canyon 2012 Glen Canyon 2012 West Clear Creek

Visit my Nikonians gallery.

Hi Dean,

Thanks for posting your test results.
Might be a good topic to pin.
Good Luck and Enjoy your Nikons!

Best Regards,
Marty

Thanks very much, Dean.

I'll "pin" this topic for the benefit of everyone

Brian
Welsh Nikonian
Check out the Nikonians Team pages

Hey Dean,

Great idea.

I decided to test my preferred shooting method: R10 RAW 14BIT lossless A-Dist-OFF using your methodology.

I used a 16GB Class 10 PNY SDHC Card. My time from the last shot in the 10 shot burst to the 1st shot after the buffer cleared was 15s.

Jason

Visit my Nikonians gallery.

Visit my Nikonians gallery.

Thanks, Jason. I took the 4Gb card out of my D40 to do a test with a faster card and this makes a huge difference, 9 sec as opposed to 23 sec! The things I have noticed about the auto distortion and the 12/14 bit raw still are relevant, I believe but for high frame rate shooting it would really pay to get the faster series of memory cards. I would be curious to see where the limits of the camera write speed come into play rather than the card speed being the limiting factor.

Looks like there is still something out there for me to buy after all. NAS

Dean
Phoenix, Arizona USA
Nikonians Team Member
Website: The Splendid Silence of Light

Recent Trips: Grand Canyon 2012 Glen Canyon 2012 West Clear Creek

Visit my Nikonians gallery.

So just to confirm if we want to fire a long heavy burst, we should have Aoto-Distortion Off?

Thanks very much for this analysis Dean! And interesting about the card speeds too.

Best regards, SteveK

'A camera is an instrument that teaches people how to see without a camera.' -- Dorothea Lange
My Nikonians gallery
My Nikonians Blog

>So just to confirm if we want to fire a long heavy burst, we should have Aoto-Distortion Off?

Interesting. I have it set in my NX2 batch files to apply when I load images from my D200 and D70s.

I can definitely see where it would slow the camera - watching it in NX2, it moves every pixel there is in the image!

""

Interesting test on this guys site: http://sportsphotoguy.com/nikon-d7000-raw-burst-test/

Appears right now 8-9 seconds (~23 mb/sec) is about as fast as you can get these days. Everyone seems to be releasing UHS-1 cards now, which the d7000 can use, though if the Kingston one used in the linked test is any indication, it might not be better. If they can cut it down to about 4.5-5 seconds, then you are shooting at a true 6 fps.

Jason

Visit my Nikonians gallery.

Visit my Nikonians gallery.

Interesting piece on San Disks new UHS card coming out this month:

http://www.photographybay.com/2011/01/27/sandisk-extreme-pro-uhs-i-sd-card-in-the-nikon-d7000/

7 seconds sounds pretty good to me!

Jason

Visit my Nikonians gallery.

Visit my Nikonians gallery.

A change in the firmware in the roll of the second slot could dramatically increase speed (almost double).

Adding Alternate: (meaning)
write First shot to Slot 1,
write Second shot to slot 2,
Write Third shot to which ever is ready.
.
.
.

currently its: (pg 89)

Overflow,
Backup,
and RAW + JPEG

Visit My Website Nikon Fanboy.

Isn't the limitation at this point, given the card speeds available, the D7000 firm/hardware? How fast can it write? The fastest card isn't going to give you much help if the write speed technology in the camera isn't keeping up (or isn't available at the D7000 price range).

Visit <www.carpemoment.com>.

or http://gary-adkins.fineartamerica.com

Visit my Nikonians gallery.

Which only proves my point, currently the D7000 only writes to one card.. the second card is a Overflow or mirrored copy.

Writing to both would give you a 2X in offload speed.

Visit My Website Nikon Fanboy.

>Which only proves my point, currently the D7000 only writes
>to one card.. the second card is a Overflow or mirrored copy.
>
>Writing to both would give you a 2X in offload speed.
>
>
Actually, the D7k will write to both, with RAW and jpeg simultaneously writing...

Visit my Nikonians gallery.

>Isn't the limitation at this point, given the card speeds
>available, the D7000 firm/hardware? How fast can it write? The
>fastest card isn't going to give you much help if the write
>speed technology in the camera isn't keeping up (or isn't
>available at the D7000 price range).


This is a good point. In the electronics world where I live (design ICs) we use a technique called ping-ponging. This simply means alternately writing data bewteen 2 ports. This works if the port is the speed bottleneck and not the device driving the ports. Nikon may have included this ping-pong capability by physical design and will implement with a future sofware upgrade.

I would be interested if anyone knows about this.

When reading this, keep in mind that Nikon recommends (p. 319 of User's Manual) Speed Class 6 or better.

It appears that much, very much depends on the cards's bus speed. As it stands now, there would be little benefit in speeding up the camera's bus interface beyond that of the minimum card supported, which would be Class 6 at 12.5MB/s. A firmware upgrade wouldn't get that bus interface any higher, not unless the hardware that implement that bus already supports the higher rate, and that's doubtful.

From www.sdcard.org/consumers/speed/speed_class/

SD memory card manufacturers use different types of flash memory to create an SD memory card, so actual transfer speed can vary.

The SD Association defines Speed Class standards that pertain to bus interface speeds, not maximum data transfer, or burst, speeds.

The Speed Classes defined by the SD Association are Classes 2, 4, 6 and 10, with separate UHS Speed Classes I and II designed for UHS devices only. UHS (Ultra High Speed), the fastest performance category available today, defines bus interface speeds up to 312 Megabytes-per-second (MB/s) for greater device performance. It is available on SDXC and SDHC memory cards and devices.

Speed Classes available in SDHC cards:
- Speed Class 10 uses a High Speed Bus Interface (I/F) intended for Full HD video recording and HD still consecutive recording
- Speed Classes 4, and 6 use a Normal Bus Interface definition intended for HD through full HD video recording
- Speed Class 2 also uses this same Normal Bus Interface definition, but intended for SD video recording
- UHS Speed Class I uses a UHS-I Bus I/F intended for the full higher potential of recording real-time broadcasts and capturing large-size HD videos
- UHS Speed Class II uses a UHS-II Bus I/F that is about a factor of three (3X) faster than UHS-I

SD memory card manufacturers created “maximum” or “burst” speed designations to support consecutive picture-taking in burst shooting modes on your camera. Maximum speeds indicate read/write speeds in megabytes per second, and is not defined by SD standards. Manufacturers indicate maximum speeds as well as capacity, Speed Classes, and Bus Speeds on the SD memory card and/or the SD memory card packaging.

It was in a press release at LAS VEGAS (CES Booth #14200), January 5, 2011 that Sony Electronics announced its SDHC Class 10 memory cards, which have a 22 MBS transfer rate, the fastest SDHC cards available from Sony.

I can't even find the max transfer rate (burst speed) on Centon's website, nor on its packaging for their Centon 8GB Speed Class 10 SDHC cards (Manufacturer Part # RC8GBSDHC10).


From www.sdcard.org/consumers/speed/bus_speed/

Bus Interface Bus Speed
Normal Speed 12.5MB/s
High Speed 25MB/s
UHS-I 50MB/s (SDR50, DDR50) or 104MB/s (SDR104)
UHS-II 156MB/s or 312MB/s

Would appreciate comments from someone who knows more specifically about the possibility of increasing the D7000's transfer rate.

Take care, ..../Dan

Here is an update to the test results above using the same protocol.


***************************************************
SanDisk Extreme Pro 16Gb 45MB/s Class UHS-1
07s - r10 RAW 14BIT Lossless A-Dist-OFF
***************************************************

The Transcend card took 23 seconds to clear the buffer, the SanDisk Extreme 30MB/s took 9 seconds and the new 45MB/s UHS-1 card took 7 seconds to totally clear the buffer.

Since it takes only about 1.5 seconds to take the 10 shots that fill the buffer in the camera a card that could keep up with that data rate would need to have a data transfer rate of about 90MB/s. And this assumes the internal processing/buffer/write system could deliver data at that rate. In fact a card like that would minimize the need for a large internal buffer.



It also seems that the Transcend card data point does not fit into the graphed data assuming a linear relationship between write speed and time to clear the buffer. There may be some other reason this card write much slower that the SanDisk samples I have tested. I would have expected a time of about 13 seconds for a Class 10 card based on a linear relationship from the SanDisk data.

Dean
Phoenix, Arizona USA
Nikonians Team Member
Website: The Splendid Silence of Light

Recent Trips: Grand Canyon 2012 Glen Canyon 2012 West Clear Creek

Visit my Nikonians gallery.

Attachment #1, (jpg file)

>, the SanDisk Extreme 30MB/s took 9 seconds and the new 45MB/s UHS-1
>card took 7 seconds to totally clear the buffer.
>
>Since it takes only about 1.5 seconds to take the 10 shots
>that fill the buffer in the camera a card that could keep up
>with that data rate would need to have a data transfer rate of
>about 90MB/s. And this assumes the internal
>processing/buffer/write system could deliver data at that
>rate. In fact a card like that would minimize the need for a
>large internal buffer.
>

Thanks again Dean! This time you are saving me money

For the 9 second to 7 second difference I won't rush out just yet and get a UHS-1 card. I know the CF top of the line cards already achieve 90 Mbps so maybe we aren't too far away in the SD format. IIRC even Panasonic (and other) were supposed to announce some 75 Mbps SD cards but I don't recall seeing them released to market.

I am dreaming of a day when we can do what you indicate in your last sentence. No need for a large buffer on the camera if the throughput to the card is fast enough. It will be interesting to see if the buffer emptying can be dropped from the 7 second level with yet a faster card or whether in fact that is now the limit of the speed the camera can do. I think I read somewhere else the 45Mbps cards may now be at the limit of the buffer emptying capacity of the camera, but I prefer your tests as they are clear and concise.

The D7K buffer is comparatively miserly as compared to higher in the Nikon camera food-chain so we have to be pretty careful with our bursts. That is one thing I notice when I still occasionally use my D300, it hammers away merrily in comparison on long bursts.

Best regards, SteveK

'A camera is an instrument that teaches people how to see without a camera.' -- Dorothea Lange
My Nikonians gallery
My Nikonians Blog

>Thanks again Dean! This time you are saving me money
>
Thanks, Steve. Should I pull a 'KR' and ask for paypal donations to feed my starving family since I am saving you money?

Actually, the price increase from the Extreme 30MB/s card to the UHS-1 45MB/s card is only about $15 US. If I had a 30MB/s 16GB card I would not have replaced it but the one card I have is a 4GB Extreme. I'll keep that in my D40. But the performance increase over the Transcend card is truly significant and I'm glad I spent the money. I will keep the Transcend card for overflow use in slot 2 and also send any videos I take to that slot as well.

Dean
Phoenix, Arizona USA
Nikonians Team Member
Website: The Splendid Silence of Light

Recent Trips: Grand Canyon 2012 Glen Canyon 2012 West Clear Creek

Visit my Nikonians gallery.

Does the jpeg algorithm matter much? Size priority vs. Optimal?

Visit my Nikonians gallery.

Visit my Nikonians gallery.

The jpeg settings makes no difference I could detect.

Dean
Phoenix, Arizona USA
Nikonians Team Member
Website: The Splendid Silence of Light

Recent Trips: Grand Canyon 2012 Glen Canyon 2012 West Clear Creek

Visit my Nikonians gallery.

Now that Sandisk have released their 95 MB/s cards (such as the Extreme Pro SDSDXPA-008G-A75 8 GB Secure Digital High Capacity (SDHC))I would be interested in a validation of the data rate you extrapolated on the graph for the "Future Card UHS-?". It looks like the future is here at very attractive prices!

I just ordered myself a couple of the cards since the price was the same as their 45 MB/s version (which seems to be getting discontinued). I paid $32 for the 8GB version. The D7000 specs appear to support the full UHS-I speeds, and the mainline cards appear to converging to that same speed rating.

For those of us wanting the quickest clearing of the buffer (and the subsequent ability to keep shooting), this is an important test and I hope you keep updating it as newer (and faster) cards appear on the scene. Your methods seem to reflect real-world conditions and I appreciate seeing the graphical results.

Any test method is subject to a lot of variables -- but as long as you keep doing the same thing when you run YOUR test, we should have comparable results with the different cards. The graph appears to be a logarithmic curve, which would be expected for this type of process -- diminishing returns as cards get faster -- so I'm very interested to see if the 95 MB/s cards do perform as you expected, and give us the ultimate performance that the D7000 is capable of.

Also, I was trying to replicate your results and noted that having long exposure NR turned on or off does have an impact on the R-xx value -- with it on, I had a smaller R number. It probably doesn't matter for most shooting, of course, but the camera thinks so!

With my current card (Sandisk Ultra Class 6, 20MB/s rated) I get 9s with your test.

Thanks.

I just picked up a couple of Sandisk extreme pro UHS-1 16 GB cards. Time to flush the buffer is 7 seconds (14 bit lossless RAW). With two cards, and slot 2 set to backup, it's 14 seconds. Now, in CL mode for birds in flight, I can shoot intermittent bursts of 4 or 5 frames with in-camera RAW back-up and almost never run out of buffer

My old Transcend SDXC class 10 cards used to take an eternity to flush the buffer and shooting slot 2 as back-up just wasn't feasible.....

Visit my Nikonians gallery.

>Here is an update to the test results above using the same
>protocol.
>
>
>***************************************************
>SanDisk Extreme Pro 16Gb 45MB/s Class UHS-1
>07s - r10 RAW 14BIT Lossless A-Dist-OFF
>***************************************************
>
>The Transcend card took 23 seconds to clear the buffer, the
>SanDisk Extreme 30MB/s took 9 seconds and the new 45MB/s UHS-1
>card took 7 seconds to totally clear the buffer.
>
>Since it takes only about 1.5 seconds to take the 10 shots
>that fill the buffer in the camera a card that could keep up
>with that data rate would need to have a data transfer rate of
>about 90MB/s. And this assumes the internal
>processing/buffer/write system could deliver data at that
>rate. In fact a card like that would minimize the need for a
>large internal buffer.
>
>
>

>It also seems that the Transcend card data point does not fit
>into the graphed data assuming a linear relationship between
>write speed and time to clear the buffer. There may be some
>other reason this card write much slower that the SanDisk
>samples I have tested. I would have expected a time of about
>13 seconds for a Class 10 card based on a linear relationship
>from the SanDisk data.
>
>
>

I wanted to thank you for this thread, and this post in particular. I was shooting with Transcend cards, and happy enough in my ignorance: they worked, never burped. After deciding to replace them, and reading this, I bought the latest & greatest SansDisk Extreme Pro cards, and now am enjoying a noticeably faster buffer clear speed.

Another example of how Nikonians helps me to get the most out of my gear.

I have a d3000 (saving for a d7000 that's why I'm hanging around here) and I have only a r03. I did what dean suggested and it went to r04 then, I switched off active d lighting and it went to r06. I feel I could fix any dark shadows in Light room as I shoot in raw.If any of you have active-d on, you may want to try switching it off to see what you get.

Thanks for the suggestion, David but active D-lighting has no impact on the buffer for the D7000. It may have a faster processor or perform this function in such a way that other parts of the processing/storage system are the limiting factors.

Dean
Phoenix, Arizona USA
Nikonians Team Member
Website: The Splendid Silence of Light

Recent Trips: Grand Canyon 2012 Glen Canyon 2012 West Clear Creek

Visit my Nikonians gallery.

"Real World" test
I just took 32 Large Fine JPEGS of the best quality with my D7000 before it slowed down. Then it took the Sandisk 45MBs card very few (about 2-3) seconds to clear the buffer.

I expect that if it was profitable on the D3, we can expect a buffer upgrade option for the D7000 or/also they may soon come out with a D7000s?
And, if $500 like the D3, I probably won't get a buffer upgrade.
It did not make a huge difference on my D3.
And, I have not had a buffer overrun since buying my D7000.

Visit my Nikonians gallery.

Just looking at your numbers: 6*(32/6)/((32/6)+3) = 3.8fps or lower should go continuous without slowdown. Not great, but may be adequate for a long duration event.

Anyone tried 3 or 4 fps with a class 10 card?

Visit my Nikonians gallery.

What seems so odd about all of this is how "rXX" gets calculated. I get r18 with JPEG Large, Seems strange that this would vary on the same model using the same settings as you specified (1/200, f5.6, AF-C, iso800, A-Dist-OFF).
Any other thoughts on other settings that could effect "r" ?

This is moot of course, but I do not understand why Nikon decided to cripple the D7k with so little buffer and 1080 Video at only 24fps?
Memory is cheap, especially when you buy as much as Nikon does.
Maybe they will offer a buffer upgrade like they did on the D3,
but I doubt it because who would pay $500 to upgrade an $1100 camera unless there were other improvements?
More likely we will see a D7000s.

Visit my Nikonians gallery.

>This is moot of course, but I do not understand why Nikon
>decided to cripple the D7k with so little buffer and 1080
>Video at only 24fps?
> but I doubt it because who would pay $500 to upgrade an $1100
>camera unless there were other improvements?
>More likely we will see a D7000s.

AKA the D400 .... <i>possibly</i> due on 24 Aug.

>
>AKA the D400 .... <i>possibly</i> due on 24 Aug.
>

The, underwater version is available now, really cheap.
Of course they do not works so well.
I would guess the release version will not be before April at the earliest.

Visit my Nikonians gallery.

>What seems so odd about all of this is how "rXX"
>gets calculated. I get r18 with JPEG Large, Seems strange
>that this would vary on the same model using the same settings
>as you specified (1/200, f5.6, AF-C, iso800, A-Dist-OFF).
>Any other thoughts on other settings that could effect
>"r" ?
>
>

Long exposure NR ON/OFF will impact "r" even when you're shooting daylight.

>>What seems so odd about all of this is how
>"rXX"
>>gets calculated. I get r18 with JPEG Large, Seems
>strange
>>that this would vary on the same model using the same
>settings
>>as you specified (1/200, f5.6, AF-C, iso800, A-Dist-OFF).
>>Any other thoughts on other settings that could effect
>>"r" ?
>>
>>
>
>Long exposure NR ON/OFF will impact "r" even when
>you're shooting daylight.

Hmm that is very interesting. I always have Long Exposure NR set to on all my cameras so I guess my "r" is always understated since I rarely actually shoot a long exposure. I'll have to change this just curious to see how much of an effect this has.

On the main topic on this thread, my sense is the buffer emptying will never get faster than 7 seconds, no matter how fast the SD card.

Best regards, SteveK

'A camera is an instrument that teaches people how to see without a camera.' -- Dorothea Lange
My Nikonians gallery
My Nikonians Blog

I just ordered a 16GB SanDisk 90mbs SDXC card for my V1.

I will try it in the D5100 and D7000 and see if it makes any difference.
When I switched from 30 to 45mbs cards, the in camera difference was too slight to notice.
The big improvement was write to computer thru card reader time.

However, the V1 Expeed 3 processor should write much faster.

Visit my Nikonians gallery.

>I just ordered a 16GB SanDisk 90mbs SDXC card for my V1.
>
>I will try it in the D5100 and D7000 and see if it makes any
>difference.
>When I switched from 30 to 45mbs cards, the in camera
>difference was too slight to notice.
>The big improvement was write to computer thru card reader
>time.
>
>However, the V1 Expeed 3 processor should write much faster.

Do you have a USB 3.0 port in your computer? The faster cards coupled with a USB 3.0 reader really makes a difference.

>
>Do you have a USB 3.0 port in your computer? The faster cards
>coupled with a USB 3.0 reader really makes a difference.

Depends on how fast your FireWire or USB 2.0 reader is.
My Delkin USB 3.0 is only slightly faster

Visit my Nikonians gallery.

I had to return my Delkin USB 3.0 reader (Model DDREADER-42) because it would not reliably recognize the new 95MB/s Sandisk cards. (SanDisk Extreme Pro® SDHC™ UHS-I Memory Card - 8GB SDSDXPA-008G-A75). I returned it to the store and purchased a Sandisk Imagemate Model SDDR-289-A20 to replace it.

Plugged into the USB 3.0 port of my computer and using Nikon Transfer 2 as the transfer program, I was able to move 100 12-bit RAW files (average size or around 14.5 MB each) in just under 23 seconds. I timed the transfer from when I first clicked the transfer button until the transfer queue read 0.

While some people are reporting only minimal speed increases from the card and USB 3.0 readers, my experience is quite different.

>>
>>Do you have a USB 3.0 port in your computer? The faster
>cards
>>coupled with a USB 3.0 reader really makes a difference.
>
>Depends on how fast your FireWire or USB 2.0 reader is.
>My Delkin USB 3.0 is only slightly faster
>

Visit my Nikonians gallery.

I hope Nikon:
1) buys your test results and methodology -- really valuable, and
2) comes out with a clear explanation of all of the factors that influence r###


Meanwhile, please keep this up. I'm learning a lot just thinking about what you've done here.

Take care, ..../Dan

Excellent analysis Dean.

This helped me to finally choose Auto Distortion OFF as i can do that in LL.

Thanks.

-------------------------------------------------------
Nikon D7000 + 18-105 VR + 35mm f/1.8 + 70-300 VRII + AF SP 90mm Macro + SB-700 + TT SW Harddisk + Vanguard Alta Pro 263 AB 100
https://secure.flickr.com/photos/prakhardeep/

thanx for sharing the info!! very useful info!!!

Has anyone got any performance analysis using the new SanDisk 16GB Extreme Pro SDHC UHS-I 95MB/s Card in the D7000 yet?
I thought I had read somewhere within this forum that the D7000 doesn't benefit from any cards faster than about 45M/Bs but I can't track down that post - so it would be nice to get some feedback.

Thanks,

Grant W.
NZ

>Has anyone got any performance analysis using the new SanDisk
>16GB Extreme Pro SDHC UHS-I 95MB/s Card in the D7000 yet?
>I thought I had read somewhere within this forum that the
>D7000 doesn't benefit from any cards faster than about 45M/Bs
>but I can't track down that post - so it would be nice to get
>some feedback.
>
>Thanks,
>
>Grant W.
>NZ

I've tried the 95MB/s, 45MB/s and the 30MB/s cards. The 30MB/s cards are very fast, the 45MB/s cards are marginally faster - the 95MB/s cards don't perform any faster in the camera than the 45MB/s cards, but read faster in a high speed (eSATA) card reader.

I think the D7000's write speed is about 25MB/s, so there comes s point where a faster card doesn't flush the buffer any faster.

Visit my Nikonians gallery.

>Has anyone got any performance analysis using the new SanDisk
>16GB Extreme Pro SDHC UHS-I 95MB/s Card in the D7000 yet?
>I thought I had read somewhere within this forum that the
>D7000 doesn't benefit from any cards faster than about 45M/Bs
>but I can't track down that post - so it would be nice to get
>some feedback.
>
>Thanks,
>
>Grant W.
>NZ

I own 4 of the 8GB version of the cards and couldn't be happier. I strongly encourage you to test the cards yourself as you will find wildly different opinions here. Buy some cards from some place that allows returns, borrow from a friend, or something - spend a few $$ on shipping costs to save you real $$ and time in the long run.

The most dramatic performance increase that I noted was the time to transfer the card's contents to my computer. A USB 3.0 card reader is a must for ultimate performance - the USB 2 readers max out at speeds lower than the card is capable of.

I went with the UHS-I 95MB/s cards because the price difference was negligible to the 45MB/s version at the time I bought them. Note that different manufacturer's cards perform differently, so you really must do your own testing to get a feel for what is possible.

Your shooting needs also matter - are you shooting action of some kind where you need the maximum performance and need to optimize the D7000's capabilities (Just can't afford a D4, right?)? Do you want the fastest transfer time to read all the cards you shoot? In other words - why does speed matter to YOU? If you want the maximum performance from your D7000, buy the 95MB/s cards - maybe you can compromise and get the 8GB version at half the price of the 16GB you are looking at -- do you REALLY need 16GB cards?

Your choice of a reader is important -- in a previous post I stated:

"I had to return my Delkin USB 3.0 reader (Model DDREADER-42) because it would not reliably recognize the new 95MB/s Sandisk cards. (SanDisk Extreme Pro® SDHC™ UHS-I Memory Card - 8GB SDSDXPA-008G-A75). I returned it to the store and purchased a Sandisk Imagemate Model SDDR-289-A20 to replace it.

Plugged into the USB 3.0 port of my computer and using Nikon Transfer 2 as the transfer program, I was able to move 100 12-bit RAW files (average size or around 14.5 MB each) in just under 23 seconds. I timed the transfer from when I first clicked the transfer button until the transfer queue read 0."

You're on the other side of the world from me and I do understand that availability of products may be different than what I experience in the US. Best of luck with your purchases.

Visit my Nikonians gallery.

Here are some tests from another reviewer:
http://sportsphotoguy.com/nikon-d7000-raw-burst-test/
http://www.robgalbraith.com/bins/camera_wb_multi_page.asp?cid=6007-12454

Visit my Nikonians gallery.

Today is the 1st day that I found this discussion thread. When I started reading it I was a little concerned that no one knew about the latest cards on the market. Then I saw the 2011 dates.

(Full disclosure--I am about to rant--skip if you wish) I am somewhat of a memory card bigot. When I buy, I always buy the latest highest performing Lexar or SanDisk memory cards. When I purchased a D7000, I bought two SanDisk Extreme Pro 32GB, SDHC, UHS-1 (95MB/s) cards as well. I do NOT want the least expensive piece of the equation to create a point of failure.

I shoot 14-Bit lossless RAW to Card-1 & highest quality JPG to Card-2. My test is shooting sports. At the last track meet that I shot, I set the highest frame rate, and held the shutter release down for as long as 12-15 seconds with no slowdown. I shot 916 photos in less than 2 hours. I wouldn't mind a slightly faster frame rate, but for what I can get, I'll still go with the highest quality, fastest memory card with a capacity so that I have absolutely no worries that I'll run out of room before I get back to my PC.

Why people spend so much time trying to find the cheapest card that will get by is something that I do not understand. I can understand that some won't need 32GB capacity. But why do those with $2-4,000 investment in camera, lens, & flash (Not including those who buy the Nikon Grip & Gitzo tripods) futz around trying to save $5-10 on something as important as a memory card.

Sorry, I don't get it. Perhaps someone on here can enlighten me.

Charlie

Visit my Nikonians gallery.

>(Full disclosure--I am about to rant--skip if you wish)
>Charlie

I agree with you completely! I bought the highest speed card I could with the capacity I could afford. I added a second pair of identical 8GB cards when I could afford it. If I was working professionally, I would not have hesitated in buying the higher capacity version of the same cards. They are winners in my mind.

Visit my Nikonians gallery.

nice job

you will find that the speed of the card does not affect the number of shots in the buffer BUT using a class 10 does speed up the download into your computer

Obviously, the card's write speed does not affect the size of the camera buffer - but it certainly DOES affect the rate at which the buffer can be cleared by writing images to the card. Thus the card speed can impact the maximum frame rate over a long burst.

Brian
Welsh Nikonian
Check out the Nikonians Team pages

not been my experience as the camera writes to the card at a much lower rate than even a class 4 can accept.

>not been my experience as the camera writes to the card at a
>much lower rate than even a class 4 can accept.

No, certainly not class 4. I recall pretty clearly that I obtained a benefit in in-camera buffer clearing speed, right up to 45mbps cards. After that yes, there is no improvement (but that speed is well beyond Class 4 and 6 and beyond many early Class 10 cards too).

For me that's what this thread is about, the cutoff point.

Best regards, SteveK

'A camera is an instrument that teaches people how to see without a camera.' -- Dorothea Lange
My Nikonians gallery
My Nikonians Blog

There are strongly differing opinions and personal experiences regarding the card speeds. I encourage running your own tests, in your real-world shooting conditions. It really doesn't matter what i think, or what someone else thinks, is the optimum card for the D7000. It's what works best for YOU.

Beg, borrow, or buy some different cards and test them yourself. Costco and a few other stores have a no-question-asked return policy that might help. You can also buy 8GB cards at a reasonable cost before investing in big ones.

Visit my Nikonians gallery.

>Obviously, the card's write speed does not affect the size of
>the camera buffer - but it certainly DOES affect the rate at
>which the buffer can be cleared by writing images to the card.
> Thus the card speed can impact the maximum frame rate
>over a long burst.

Brian, I absolutely agree. Plus when you can buy a SanDisk Extreme 16GB (45GB/s) at $16.99 or the Pro version (95GB/s) for $40.38 on Amazon, why on earth would anyone spend valuable time trying to save a couple of bucks when they could be fulfilling their passion shooting fabulous photos with their multi-thousand dollar camera/just right lens from among several/just right filter/perhaps just right flash/just right tripod/just right ball head/ etc., etc.

I don't want to waste money, but at my age, I don't want to waste what little time I have left even more.

Charlie

Visit my Nikonians gallery.

thank you for sharing these useful!

It's more clear now what cards have to be chosen.

Roberto

A lot of useful information here, from which i can draw conclusions, but i still have questions.

I am using the D7000 with a SanDisk UHS-1 95 MB/s Extreme Pro card, 32 MB, shooting RAW, 14 bit, lossless compressed, for sports & ballet. (Or RAW + JPEG)

Question 1: From the D7000 manual, page 320: RAW lossless compressed, 14 bit, file size = 19.4.MB. Is that megabits, or is it mega bytes. If megabytes, how many bits? (Wikipedia is confusing on that.) The buffer capacity is 10 of those files. That should yield the buffer size, should it not? Yet Nikon support said that Nikon does not publish the buffer size.


Question 2. I have read that the transfer rate to the card from the buffer is about 25 MP/sec. Another post referred to 45 MP/sec. Does anyone have further data?

Question 3. In either case, the 95 MP/sec card transfer rate would exceed the internal transfer rate. There seems to be a consensus that the excess of 95 over 45 is wasted in the camera, but not in a USB 3 card reader. Any further comment on this point?

Question 4. In post #20, I see the formula: 6*32/6)/(32/6+3) = 3.8 fps. Can the author please explain what the arguments of this formula rpesent (units, for example)

Question 5. Is it true that the camera is writing to the buffer, and from the buffer to the card at the same time? This appears to be true, from the very first post.)

Comment: ping-ponging between the two cards looks like a great idea. But i don;'t suppose we can expect Nikon to update the firmware for free, when they can expect to sell us a new camera with tat feature. or a bigger buffer. Or does someone know how to modify the firmware to achieve that result?

Question 6. I find the overall average frame rate computations interesting, but as a practical matter, what is important to me is: how much time does it take before the buffer fills and the camera drops to the "crawl" rate. The manual has specs as to how many frames this is (depending on shooting mode (p. 320), but not timing. If I am shooting 6 frames/sec in RAW, it would take 10/6 = 1.67 sec to fill the buffer -- unless the buffer is writing out to the card at the same time, in which case it would take longer to fill the buffer, depending on both the write rate to the buffer and the write rate to the card. Right? The object is to estimate how much time it takes to avoid shooting long enough for the buffer fill and drop the exposure rate to the crawl rate. This would, for me, be the practical limit on the number of burst shots I could take at 6 fps..

Thank you all.

Visit my Nikonians gallery.

I'll answer as many questions as I can from my own experiences:

1. The image sizes are given in Megabytes, although they are approximations. Actual image size varies with the range of tonal and colour variation in each image. The buffer size is tricky to calculate precisely as the camera is writing to the buffer at the same time as the buffer is writing to the card.

2. From my own experiments, buffer flush speed is about 25 MB/s. 10 frames written from buffer to card in about 8 seconds.

3. The camera can only write to the card as fast as the buffer allows - 95 MB/s cards have little margin over 45 MB/s cards, so the buffer write speed is the limiting factor. Where the ultra-fast cards do gain is in reading to a computer via a USB or e-SATA card reader. The high read speeds reduce copying time.

4. I'll leave that to the original author.

5. Yes, it does, but it fills faster than it can empty.

6. My own experiments yielded a result of about 11-12 frames. 10 frames fill the buffer in about 1 2/3 seconds, but nearly two frames can be written out in that time, so about 12 frames before fill rate falls to write rate which is 2 seconds at 6 fps.

Visit my Nikonians gallery.

Thank you.

Re point 5:

buffer fills faster than it can when it is empty? this is counterintitive. why would that be so?.

re point 6:

"result is about 11-12 frames" (before crawl): so if 12 frames, at 6 fps, that is about two seconds of continuous shooting before hitting the crawl. Right?

But compare point 2: 10 frames written from buffer to card in about 8 seconds -- but the camera is writing to the buffer at the same time, so this 10 seconds does not mean much more than how long it takes to empty the buffer when not writing to the buffer (i./e, when not shooting). Right?


"fill rate which is 2 seconds at 6 fps". Do you mean two seconds per frame?

Visit my Nikonians gallery.

> buffer fills faster than it can when it is empty? this is
>counterintitive. why would that be so?.

That's not what Graeme said. His point was about the rate at which the buffer is filled (by taking more shots) being greater than the rate at which the buffer is emptied (by writing to the card).

Brian
Welsh Nikonian
Check out the Nikonians Team pages

>> buffer fills faster than it can when it is empty? this
>is
>>counterintitive. why would that be so?.
>
>That's not what Graeme said. His point was about the rate at
>which the buffer is filled (by taking more shots) being
>greater than the rate at which the buffer is emptied (by
>writing to the card).
>
Thanks Brian.

Visit my Nikonians gallery.

>Thank you.
>
>Re point 5:
>
> buffer fills faster than it can when it is empty? this is
>counterintitive. why would that be so?.
>
>re point 6:
>
>"result is about 11-12 frames" (before crawl): so
>if 12 frames, at 6 fps, that is about two seconds of
>continuous shooting before hitting the crawl. Right?

Correct.
>
>But compare point 2: 10 frames written from buffer to card in
>about 8 seconds -- but the camera is writing to the buffer at
>the same time, so this 10 seconds does not mean much more than
>how long it takes to empty the buffer when not writing to the
>buffer (i./e, when not shooting). Right?

No, in point 2, I've run the camera in CH until the buffer indicator shows r0. I then immediately release the shutter to the half press position and time how long it takes to get back to r10. During this period the buffer is only flushing to the card. No data is being written to the buffer.
>
>
>"fill rate which is 2 seconds at 6 fps". Do you mean
>two seconds per frame?

No, I meant here that you'd get 12 frames (2 seconds at 6 fps) before the capture rate slows to a crawl. At that point, data can only be written to the buffer as fast as it can write to the card (25-28MB/s) so the camera will drop to about 1.3 -1.5 fps.

Visit my Nikonians gallery.

Thank you for the clarification.

Visit my Nikonians gallery.

OK, to summarize: 12 frames(2 sec at 6 fps) before capture rate is limited by the buffer-to-card flush rate (1.3 - 1.5 fps, at 25 - 28 MB/sec). Was that for raw lossless compressed or lossy compressed? And does that distinction make any difference in this context?

Also, 10 frames from buffer to card in 8 sec, with no capture in progress.

Visit my Nikonians gallery.

And, if the buffer flush rate is about 25 mB/s, and the buffer flushes 10 frames in 8 sec (when there is no capture), then the buffer flushed 8 sec x 25 mB/sec = 200 mB, averaging 20 mB/frame.

Visit my Nikonians gallery.

Great sharing....! Thanks for that

Really, very professional tests.

From Imaging Resource:

"The Nikon D7100 stores images on dual Secure Digital card slots, as did its predecessor. There's an important change, though. They're still compatible with the higher-capacity SDHC / SDXC cards, and the higher-speed UHS-I cards, but the latter have the potential for much greater speed. That's because the UHS-I mode supports a clock frequency of 198 MHz, versus just 81 MHz in the D7000. That allows a maximum transmission speed of 99 MB/second, versus the D7000's 40.5 MB/second. You'll need a UHS104-compatible card to support this speed, though."

Jim

According to the SD card spec, what Imaging Resource states is not accurate.

SD speeds are either 100mhz-45MBytes/Sec for UHS-1 or 208mhz-95Mbytes/sec for UHS-104, which is an enhanced version of UHS-1. I expect, though, Nikon used the chip from the D600 teardown by ifixit, which leaves us with a max speed of 45Mbytes/sec. That yields 1 frame/sec once the D7100 has a full buffer.

SD I/O chip highlighted in dark blue...
http://guide-images.ifixit.net/igi/GpMPdA4bDfTXCARM.huge

Chip specs
http://product.inrevium.com/sd/index.html

David

"Sawed that board three times and it is still too short... "

Visit my Nikonians gallery.

IR reports that after the buffer is full the D600 shoots at 1.4 fps in 14 bit lossless. Those files are 29.2 MB, so we're looking at 41 MB/sec. Based on using 12 bit lossless in the D7100 that would yield 41/22.7 = 1.8 fps. Which is consistent with the buffer (7 shots) clearing in say 4.25 seconds. Which again is consistent with the D600's buffer clearing in 9 seconds after 14 14 bit lossless compressed RAW with these cards, as reported by IR. Rob Galbraith tested these cards in a D7000 and only achieved write speeds of 26 MB/sec.

Jim

SO Friends Help me out here.
Should i Buy a:

1) Sandisk Extreme Pro 95MB/s UHS-1 SDXC Class10 or

2) Sandisk EXTREME 45MB/s SDHC Class10 SDHC™ UHS-I
for my D7000

I can say that for the D600, the 95MB cards are a waste of money from a performance perspective - they can't run faster than the SD I/O chip will allow.

With the D7100, I have not yet seen a performance test or teardown to indicate which chip it has in it for SD I/O. In the other thread running about buffer size and card performance that Jon Bloom started, BillD80 has bought the higher speed cards and hopefully will run an informal test to give us an indicator.

David

"Sawed that board three times and it is still too short... "

Visit my Nikonians gallery.

In the D800, Rob Galbraith measured the RAW throughput of the 95 MB/s cards at 42 MB/sec; the 45 MB/s cards were well down the list at 29 MB/sec. http://www.robgalbraith.com/bins/camera_wb_multi_page.asp?cid=6007-12451

Jim

>In the D800, Rob Galbraith measured the RAW throughput of the
>95 MB/s cards at 42 MB/sec; the 45 MB/s cards were well down
>the list at 29 MB/sec.
Yeah, I might buy some of the higher speed cards, I think the D600 would benefit and hopefully the D7100 would as well.

David

"Sawed that board three times and it is still too short... "

Visit my Nikonians gallery.

>I can say that for the D600, the 95MB cards are a waste of
>money from a performance perspective - they can't run faster
>than the SD I/O chip will allow.
>

No the D600 can utilize these cards. The D600 is quite fast. See my test in D600 section - http://www.nikonians.org/forums/dcboard.php?az=show_topic&forum=435&topic_id=5385&mesg_id=5385&page=). 600x or 95mb/s cards give faster buffer write time, slightly longer maximum burst duration and slightly better speed when the buffer is full.

There is a measurable difference between 300x and 600x SDHC cards in the D600. Not twice as fast, but still noticeable.

>SO Friends Help me out here.
>Should i Buy a:
>
>1) Sandisk Extreme Pro 95MB/s UHS-1 SDXC Class10 or
>
>2) Sandisk EXTREME 45MB/s SDHC Class10 SDHC™ UHS-I
> for my D7000

For a D7000 option 2 is sufficient.

Best regards, SteveK

'A camera is an instrument that teaches people how to see without a camera.' -- Dorothea Lange
My Nikonians gallery
My Nikonians Blog

Thanks for the reply SteveK

I came across this tip elsewhere and thought I'd share it for what it's worth. I haven't tried it yet because I don't have any 95MB/s cards. Has anyone else come across speeds like this?


"Here’s a tip for sports/wildlife photogs: set the camera to 12-bit compressed RAW, pop in a Sandisk 95MB/s card, and go into crop mode, and you now have a 14-15 frame buffer that almost instantly regenerates! If you need an even bigger buffer ( 0.0 ) and can spare a couple FPS, some have said the settings above, paired with Continuous Low at 5fps, can get you around a 50 frame buffer!"


Scott

www.ScottAshleyPhotography.com

I have dual 95MB/s cards... I haven't shot in Crop mode yet, but can say I get quite a few before any shutdown when shooting 14 big lossless RAW's...

http://www.wlwakefield.com

http://www.nikonians.org/forums/dcboard.php?az=set_threaded_mode&forum=329&topic_id=25668&prev_page=show_topic&gid=25668#25674

In actual tests I have done, I got 17 shots at 6fps before slowdown, and 26 shots at 5 fps before slowdown....

Visit my Nikonians gallery.

I added a link to your tests in the pinned Resources Links thread.

Hanks for all you hard work making this info available.

Dave
Quad Cities, USA
A Nikonians Team Member
My Gallery
SummersPhotoGraphic.com

Nikonians membership -
"My most important photographic investment, after the camera"

Share, Learn and Inspire