Fast SD cards

[Dieter Schaefer]

59 minutes ago, ShunCheung said:

Now I use a Z9 at 20 fps with 46MP, those XQD cards seem very slow now, and I have moved to CF Express type B

Until my CFExpress cards arrive, I am using a Sony 120GB XQD card in the Z9. Shooting stills, it's slower speed compared to a "proper" CFExpress card manifests itself mostly in limiting the number of frames until the buffer fills. There are some CFExpress cards that show high write speeds in the specs; however, those speed drop to XQD level when shooting a burst at high fps. It's important to pay attention to the "minimum" write speed of CFExpress cards in order to get the best performing cards.

1 hour ago, ShunCheung said:

However, at 10fps on the D500, the SD card couldn’t quite keep up and I occasionally ran into the buffer full situation. If I used XQD alone, the buffer was limitless on the D500, until it hit the artificial 200 frame limit. Eventually I gave up using backup.

Same here - and the reason I never went for the SD UHS-II cards. My Sony A7RIII can used those faster SD cards as well - but that's not the camera I use for shooting anything that requires a fast frame rate; so the much cheaper SD cards just do fine.

 

1 hour ago, ShunCheung said:

I still have my D500 and D850, and I am happy to use those with only CFx (or XQD).

Is CFExpress actually faster in the D500 and D850 than XQD? Or does the camera limit the write speed?

 

I can't utilize the fast transfer speeds possible from CFExpress cards through a reader to my desktop computer; the limit is dictated by the USB 3.0 ports on my now 6-year old motherboard.

[ShunCheung]

The problem with "fast" SD is that we are still using the same technology as we did 6, 7 years ago. There are new, improved specs and standards, but there is no product yet. And the fragile nature of the casing is not changed. For higher-end products that need the speed, they'll probably stay with CFx Type B or even Type A ALA Sony.

Unless you have a camera that can use XQD but not CFexpress, e.g. a D4 or D4S, there is no point to get any more XQD cards now. They are more expensive and are slower. A D850 and Z6 might not be able to take advantage of the CFx speed, but it is still a more future-proof card. However, if you have a D4, you are stuck with the old CF and XQD.

[Mary Doo]

2 hours ago, ShunCheung said:

the fragile nature of the casing is not changed.

The Sony Tough SD cards have a thick reinforcement at the lower part.  The card has a metal feel to the touch.  Stated write-speed is 299 mb/s.  Not bad - and it seems fast on the field, fast enough.  That said, in my experience I haven't had any SD card that failed, have you? 

I am sure the XQD and CF Express are faster, and I still have quite a number of them which unfortunately will not be used much any more as I am switching to Olympus.  I am not sure where the Olympus (now called "OM") system will lead, but for now I believe it's the right decision for me for many reasons, pros and cons.  Who knows, Nikon may come up with something that will change the trajectory.

[ilkka_nissila]

>There are some CFExpress cards that show high write speeds in the specs; however, those speed drop to XQD level when shooting a burst at high fps. It's important to >pay attention to the "minimum" write speed of CFExpress cards in order to get the best performing cards.

It's a complicated issue as the speed realized in practical use depends on the temperature of the card and may depend on specific camera and card combination. If one purchases the fastest cards available, the benefit from CFexpress should be noticeable over XQD on a camera like the Z9 which is very modern and optimized for CFexpress. There are some tests available (look at the in-camera speed column in the first table and "Nikon Z9 Speeds" in the second which also includes slower cards:

 

https://alikgriffin.com/fastest-memory-cards-nikon-z9/

 

On a D6, my Sony 128 GB CFexpress cards (from 2020-1) are somewhat faster than the XQD cards I bought in 2016-7, but even new XQD cards have gotten a bit faster since then as they started to use newer PCI express 3.0 interface. Reasons still to buy XQD cards might be that one may have existing bundled readers, there are  a few cameras that don't support CFexpress and the performance doesn't drop like a rock after some seconds like it seems to happen with CFexpress once the card warms up.

 

>Is CFExpress actually faster in the D500 and D850 than XQD? Or does the camera limit the write speed?

 

The Sony XQD and CFexpress cards that I have perform very similarly in the D850. Some reports suggests XQD can be a bit faster than CFexpress in these cameras. It would seem that the architecture of these cameras was designed for XQD and it was modified in firmware to support also CFexpress but the architecture is not optimized to take advantage of CFexpress's full speed potential. Newer cameras that were launched after CFexpress cards were actually on the market seem to be the opposite: XQD is somewhat slower than CFexpress on the D6 (and I believe also my Z6 II, but I don't use that camera in a way where it would matter). The Z9 is a faster camera and also achieves faster write speeds on CFexpress, but it's a little surprising they didn't give this camera enough memory for the burst performance to become a complete non-issue. I guess Nikon preferred having efficient raw compression as an option on the Z9 over putting the money into internal memory. In the subsequent steps of processing on a computer, perhaps the benefit of smaller files (rather than using lossless compression in camera combined with a large memory buffer) is apparent. Are the HE and HE* faster or slower to open on a computer than lossless compressed NEFs? I guess this would depend on the computer hardware.

 

>I can't utilize the fast transfer speeds possible from CFExpress cards through a reader to my desktop computer; the limit is dictated by the USB 3.0 ports on my now 6-year old motherboard.

 

My guess is that you'd still see some improvement in speed of transferring from CFexpress compared to XQD. I don't recall the initial figures but my computers also have older ports and it still gives about 30% speed of transfer advantage to CFexpress over XQD when transferring large numbers of images. Somehow in my need to have one PCI slot to support my LS-9000 scanner's fireware interface resulted in ports that are suboptimal for my CFexpress reader, but still it's fast enough and I can easily live with it.

Edited November 20, 2022 by ilkka_nissila

[ilkka_nissila]

29 minutes ago, ilkka_nissila said:

Somehow in my need to have one PCI slot to support my LS-9000 scanner's fireware interface resulted in ports that are suboptimal for my CFexpress reader, but still it's fast enough and I can easily live with it.

This could be seen as a confusing statement. What I meant is that I chose a computer which has a single PCI slot and several PCI express slots in order to be able to work with the fireware scanner (which works with some fireware cards but not all) and thus I didn't pay as much attention to the USB ports as I might have to be sure that the USB interface would support fast transfers. As it turned out there is a single USB-C slot and several USB-A slots (some USB 2, some USB 3) but the USB-C slot does not work with my CFexpress reader, and I have to use the slower USB-A ports. If I were to buy a new computer today I would somehow try to find out how well the ports work with my hardware before making the purchase decision. But as it turns out I am happy that my scanner works with the computer and I can proceed to digitize my medium format film (hopefully will finish in a couple of years). I kind of need the space that my negatives and slides take for other things and my plan is to scan the good images and then dispose of the originals to recover the storage space.

Edited November 20, 2022 by ilkka_nissila

[Dieter Schaefer]

3 hours ago, ilkka_nissila said:

https://alikgriffin.com/fastest-memory-cards-nikon-z9/

That's the table I was referring to. It shows that on the Z9, the write speed can drop to XQD levels for some CFExpress cards. Here's a link to Prograde's current CFExpress compatibility chart: https://shop.progradedigital.com/products/prograde-digital-cfexpresstm-2-0-memory-card-cobalt-1700. One can see that there's an entry "min sustained write" which for some cards shows transfer rates in XQD category and nowhere near what the advertised "max write rates" are. I checked with Sandisk and Lexar but couldn't find similar information about the "min sustained write speed" (Sandisk uses "up to" to cover themselves); I haven't checked other manufacturers.

 

 

[ilkka_nissila]

There is also "Guaranteed write speed" that some report. E.g., Nextorage B1 Pro has read and write speeds up to 1950 and 1900 MB/s, minimum sustained write speeds of 1800 MB/s, and finally these cards have a VPG400 rating which indicates video recording guaranteed minimum write speed is 400 MB/s.

 

Quite a different number that last one!

Edited November 20, 2022 by ilkka_nissila

[ShunCheung]

A really fast write speed that is not sustainable is kind of meaningless. Currently I only use ProGrade Cobalt and Delkin Black CFexpress cards, with sustainable write speeds in the 1500 MB/sec range.

Again, max 300 MB/sec for USHii SD seems very out of date.

[Mary Doo]

12 hours ago, ShunCheung said:

Again, max 300 MB/sec for USHii SD seems very out of date.

Not the fastest available indeed.  Hwvr, it seems to fill my need.  So I'll live with it.  😬

[Mary Doo]

A faster speed than 300 mb/s (which is already very fast) is nice on paper.  But, really, is there much practical need for these external hoopla to produce good results in real life?  300mb/s is likely already an overkill for my photographic ability and opportunities out there.  But people enjoy the hypes.  Just some observation and thought.

Edited November 22, 2022 by Mary Doo

[ShunCheung]

1 hour ago, Mary Doo said:

A faster speed than 300 mb/s (which is already very fast) is nice on paper.  But, really, is there much practical need for these external hoopla to produce good results in real life?  300mb/s is likely already an overkill for my photographic ability and opportunities out there.  But people enjoy the hypes.  Just some observation and thought.

As I pointed out earlier, 300 MB/sec cannot keep up with the 20MP D500 shooting at 10 fps so that I ended up using XQD only. Now I use a 46 MP Z9 at 20 fps. And sometimes I shoot 8K video.

But I really dislike the small size of SD cards and how fragile they are. I have lost several SD cards since they are small enough that can fall into gaps. CFx Type A has that same issue.

I have never lost a single CF card or XQD/CFx Type B.

[Mary Doo]

5 hours ago, ShunCheung said:

As I pointed out earlier, 300 MB/sec cannot keep up with the 20MP D500 shooting at 10 fps so that I ended up using XQD only. Now I use a 46 MP Z9 at 20 fps. And sometimes I shoot 8K video.

But I really dislike the small size of SD cards and how fragile they are. I have lost several SD cards since they are small enough that can fall into gaps. CFx Type A has that same issue.

I have never lost a single CF card or XQD/CFx Type B.

Shun, your response motivated me to do a simple test with the Sony Tough SD card in my Olympus OM-1 camera.  (Note that I normally hate testing and equipment statistical stuff,  but my curiosity got the better of me because you mentioned your SD card could not keep up with shooting at "10 fps" and it seemed to be at odds with my experience.  Now my test results are "scary" to me because I never expected it to be so fast and I never shot so many frames - it would be crazy.  I don't know whether it's because of the camera or the SD card, or both.  Here was how I tested:

The Olympus OM-1 is 20.6MP - equivalent to D500.  Shooting RAW + JPG, so every shot produced two frames.

Test 1:  Mechanical Mode:  I kept my finger on the shutter until I felt the first sign of hesitancy:.
            Number of frames:  291 RAW, 291 JPG.

Test 2:  Silent mode.  I did not complete the test because it seemed to go on forever and there was no sound and no indicator (or I did not recognize the indicator).  I was a little worried so I stopped it. 
             Number of frames:  289 RAW, 289 JPG

I don't know what to make of this.  I take your word for the under-performance of the SD card with the D500.  But I am glad I can now understand why I feel the Sony Tough SD card is fast enough for me - because I never really tested it.  LOL.

[ShunCheung]

Mary, there are a lot of variables such as frame rate, the degree of RAW compression, bit depth, high-ISO noise reduction, etc., as well as the size and compression of the JPEG files you are getting. On the D500, I typically use 14-bit, lossless compressed RAW at the fastest 10 fps available on the D500. However, back in 2016, the XQD cards I was using were merely 400 MB/sec, theoretically only 33% faster than the 300 MB/sec UHS II SD cards, but 400 MB/sec is sufficient to give the D500 effectively an endless buffer. Part of the problem maybe Nikon is having a somewhat limited amount of real buffer (RAM) on their camera. Nikon is more dependent on using fast memory cards to unload from the buffer.

But all of that is now "ancient" technologies. Today I stick to CFx Type B cards that can sustain 1500 MB/sec.

[Dieter Schaefer]

12 hours ago, Mary Doo said:

The Olympus OM-1 is 20.6MP - equivalent to D500.  Shooting RAW + JPG, so every shot produced two frames.

 

52 minutes ago, ShunCheung said:

Part of the problem maybe Nikon is having a somewhat limited amount of real buffer (RAM) on their camera.

A google search shows the D500 to have 1.3GB memory and the OM-1 1.8GB.

Also, the Sony Tough card is currently the fastest writing SD UHS-II card for the OM-1: https://www.memorycard.expert/-Cameras/OM-System-OM-1

I don't own one (I don't own any SD UHS-II card actually), so I can't perform a test (and I have not find one online). All the D500 memory speed tests I can recall used older SD UHS-II cards that maxed out at around 160MB/s write speed; the best write speed in the OM-1 is around 215MB/s.  From this table one can see that the same card used in the D500 and the OM-1 always performs better in the OM-1: https://www.memorycard.expert/-Cameras/Nikon-D500

Both those tables also show the enormous write speed spread of the SD and SD UHS-II cards available.

 

[Mary Doo]

7 hours ago, ShunCheung said:

Mary, there are a lot of variables such as frame rate, the degree of RAW compression, bit depth, high-ISO noise reduction, etc., as well as the size and compression of the JPEG files you are getting. On the D500, I typically use 14-bit, lossless compressed RAW at the fastest 10 fps available on the D500. However, back in 2016, the XQD cards I was using were merely 400 MB/sec, theoretically only 33% faster than the 300 MB/sec UHS II SD cards, but 400 MB/sec is sufficient to give the D500 effectively an endless buffer. Part of the problem maybe Nikon is having a somewhat limited amount of real buffer (RAM) on their camera. Nikon is more dependent on using fast memory cards to unload from the buffer.

But all of that is now "ancient" technologies. Today I stick to CFx Type B cards that can sustain 1500 MB/sec.

Shun, I think there is nothing to pooh-pooh about this "ancient" technology when, combined with superb camera innovation, it can write much more files in a much shorter time span than your whatever-"super"-technology.  The OM-1 files uses 12-bit lossless compression which is way good enough, as evidenced by the output that is the bottom line.  You love your camera and your cards and there are good reasons, but Nikon's results are not better than this "ancient" technology for now.  😬

Edited November 23, 2022 by Mary Doo

[ilkka_nissila]

11 hours ago, Mary Doo said:

Shun, I think there is nothing to pooh-pooh about this "ancient" technology when, combined with superb camera innovation, it can write much more files in a much shorter time span than your whatever-"super"-technology.  The OM-1 files uses 12-bit lossless compression which is way good enough, as evidenced by the output that is the bottom line.  You love your camera and your cards and there are good reasons, but Nikon's results are not better than this "ancient" technology for now.  😬

Shun is using 20 fps with 45 MP files, so there is a lot of data moving to the card. Olympus reportedly allows up to 50 fps bursts in electronic shutter mode, is that what you were using, or was it a slower speed (10 fps)?

Nikon seem to be using a slightly smaller buffer memory in the camera and thus if using the largest files, long bursts require fast cards (alternatively, a compressed RAW file can be used). On the other hand, the Z9 although it has a built-in vertical grip included, is less expensive than Sony A1, so maybe that smaller memory was a conscious choice to keep the purchase cost down. At some point we had cameras that couldn't record 10-image bursts (D7100) comfortably so we've come a long way from that (although in a more expensive camera). A lot of people who shoot sports still use JPG which makes these things non-issues. In many sports venues the lighting is set very evenly and quite brightly for television cameras and this makes it unnecessary to have a lot of extra dynamic range or colour adjustment options available, and JPG is a perfectly valid choice.

I think it's great that these cameras allow relatively liberal use of burst capabilities for those that need them. There is no need to defend one's choices any further than noting that the capabilities are sufficient.

For me I have a similar view to Shun's about SD cards and I generally avoid using them. I haven't misplaced a card but four or five of them stopped working (or had corrupted files) in a relatively short time period of less than 10 years, while my CFexpress, XQD or Compact Flash cards never had any failures. This is the main reason I avoid using SD when possible. At work I have cameras that I use SD and I try to keep relatively new cards in them and avoid handling those cards unnecessarily. This doesn't mean others will have data loss or card failure, it just is my own experience that this has happened much more often than I would have hoped.

Edited November 24, 2022 by ilkka_nissila

[Mary Doo]

4 minutes ago, ilkka_nissila said:

I think it's great that these cameras allow relatively liberal use of burst capabilities for those that need them. There is no need to defend one's choices any further than noting that the capabilities are sufficient.

You always do deeper analysis than most on this forum and that is thoughtful.  I was pointing out the at least some (obviously not all) SD cards, when used with superior camera technologies, are currently producing results that are incredibly efficient and effective.  I have Nikon cameras that use XQD and CFX as well.  

[thomasx]

If you're referring to the OM system used in Olympus cameras, the best memory card would depend on the specific camera model you have and your intended usage.

In general, it's recommended to use SD (Secure Digital) cards with a UHS-I or UHS-II speed class rating for Olympus cameras. These types of cards offer faster read and write speeds, which can be especially beneficial for capturing high-quality photos or videos, or for continuous shooting.

Some popular brands for memory cards include SanDisk, Lexar, and Kingston. It's important to choose a reputable brand and ensure that the card you select is compatible with your camera model.

In addition to the speed class rating, you may also want to consider the storage capacity of the card. The amount of storage you need will depend on how frequently you use your camera and how often you transfer files to a computer or other device.

Overall, the best memory card for your OM system will depend on your specific needs and preferences. Be sure to do your research and choose a high-quality card that is compatible with your camera.

[mike_halliwell]

Well yes.

🙂

[Mary Doo]

On 3/21/2023 at 7:32 AM, thomasx said:

Be sure to do your research and choose a high-quality card that is compatible with your camera.

Did.  Thanks.

[rodeo_joe1]

On 3/25/2023 at 9:05 AM, Mary Doo said:

Did.  Thanks.

Sounds like you're thanking an AI 'bot for its regurgitated and generic' 'advice'. 

I wonder if it can lie? 

So, thomasx - are you an AI 'bot? 

[mike_halliwell]

Aren't there 'rules' about AI telling lies...? 🤣

I wonder if this is a recognised way of beta-testing new AI tech.... bit like the Turing Test?

[Mary Doo]

9 hours ago, rodeo_joe1 said:

So, thomasx - are you an AI 'bot? 

Dunno.  Bot, AI, updated Turing, breathing human, whatever.  Anyhow thomasx's post appears to be the scheduled follow-up on the project - to double-check and advise on whether I did do my "research to choose a compatible high-quality card".  😁

I am glad to report "Mission Accomplished".  I did my "research" right here on this thread, and happily completed the project with the advice, recommendations, and references offered by helpful and knowledgeable members of this forum.  Thank you. 

[mike_halliwell]

Don't mention it. Happy to help, to help, to help, to help...... brrrrrrrrrrrr.... klunk.

🤣