CFExpress 4.0 Question

If you would like to post, you'll need to register. Note that if you have a BCG store account, you'll need a new, separate account here (we keep the two sites separate for security purposes).

Has anyone tried a CFExpress 4.0 card in their Z8/9 yet? For example this one:
https://shop.progradedigital.com/products/prograde-digital-cfexpress™-type-b-4-0-memory-card-gold

I recently upgraded my PC to support USB 4.0 and got a card reader that also supports USB 4.0. This allowed me to read/write to my current CFExpress 2.0 cards at their maximum speeds which in my case is between 1400 to 1700 MB/s. Before I was limited to 1000 MB/s due to the USB 10Gb/s limit of the card reader interface. The CFExpress 2.0 specification theoretical maximum is 2000 MB/s. Obviously these 4.0 cards cant run at full potential in a Z8/9 as the camera interface only supports v2.0 cards. BUT since these new cards support sustained writes of 2400 MB/s would it be possible for the Z8/9 to saturate its interface and write at 2000 MB/s in 2.0 mode using these cards? And if they could saturate the interface would it improve the 20FPS burst performance enough to make a difference in real world shooting?
 
You ask a great question and so far, I haven't seen any tests or information which provides any insight. Right now, the 512 GB Gold V4 is on sale for $162. Add in a discount code and these cards are some of the best bargains out there. According to my conversations with ProGrade, the cards should be able to run as fast as the camera output (v2) will allow. I have not been able to substantiate this or find any theoretical output "benchmarks", though I've come across some real-life testing with high quality cards which suggest sustained write speeds for the Z8Z9 are around 700'ish MB/sec. Whether this is limited by the firmware or hardware is likewise uncertain. My informal testing of the V4 Gold in the Z8 have been favorable and the cards have delivered everything that I've thrown at it. 8k video in different codecs for at least 15", no problem. > 5 sec of shooting at 20 FPS stills, no problem. Heat, no problem and they actually feel "cooler" than my 325 GB cobalt cards by comparison, though again I haven't performed any rigorous testing. For now, I feel rather confident in these cards and no longer feel compelled to shell out exorbitant $ for Delkin Blacks/ProGrade Cobalts.
 
There was commentary about this on another thread within the past two months. As I recall one person’s experience was that there was no discernible difference in data transfer to a 4.0 card on a Z9. That’s what I’d expect because the camera’s firmware (and possibly its hardware, too) is designed for writing to a 2.0 card.
 
BUT since these new cards support sustained writes of 2400 MB/s would it be possible for the Z8/9 to saturate its interface and write at 2000 MB/s in 2.0 mode using these cards?
i’ve had the same thought and my testing of nvme drives vs pci version has suggested the answer is no.

i suspect the performance will be half the rating of the v4 number because the v2 bus runs at half the speed of the v4. ie, 1200MB/s
 
And if they could saturate the interface would it improve the 20FPS burst performance enough to make a difference in real world shooting?
my testing of the fastest v2 cards suggest that there is a camera limit that kicks in before the theoretical limit of the v2 cards. it seems the camera with the current fw can’t sustain more than about 15fps in lossless compressed raw which is lower than the measured speed of some of these cards

so basically, with today’s firmware, there is no speed difference in-camera between the fastest v2 cards
 
Well its a bit disappointing to hear that the camera writes to the 2.0 cards at a somewhat lower rate than they are rated. Wish I knew what that actual camera write rate was. I might just snag one of the cards just to do some more testing as they seem reasonably priced and of course post processing will be way faster.
 
Well its a bit disappointing to hear that the camera writes to the 2.0 cards at a somewhat lower rate than they are rated. Wish I knew what that actual camera write rate was. I might just snag one of the cards just to do some more testing as they seem reasonably priced and of course post processing will be way faster.
Well, they do and most resources on the web suggest looking for a card which will write with sustained speeds of 800 MB/sec to support 8k video. From the limited testing that I've seen, and again the methodology has not been disclosed or validated, most high performing cards - such as the Delkin Black, Prograde Cobalt, etc. - deliver real world speeds in the Z8/Z9 of around 700 MB/sec and they do support 8k. For the card in question, the ProGrade Gold V4, my limited testing suggests that it functions well in terms of video, stills, and termperature. I've included a chart that ProGrade supplies regarding its cards and have highlighted in red the Gold V4 card.

Screenshot 2024-02-04 144141.jpg
You can only see EXIF info for this image if you are logged in.
 
Here’s a solid review of the Cobalt V4 with some testing in an R5 and V4 reader along with a reference to the corresponding Gold V4. My take is that these cards far exceed what current cameras demand and they will likely perform well in the future provided that the form doesn’t change (who doesn’t have a whole array of CF’s sitting around gathering dust?).

 
also, i just wanted to point out the real, measured performance potential of some of the faster v2 cards:

#1 -- 1666.42MB/s Lexar Diamond 128GB (est 25fps*) [SSD Review=1683.23]
#2 -- 1629.71MB/s Nexstorage B1 Pro 165BG (est 25fps*)
#3 -- 1541.40MB/s Prograde Cobalt 650GB (est 23fps*) [SSD Review=1420.52]
#4 -- 1527.24MB/s Delkin Black 650GB (est 23fps*)
#5 -- 1543.18MB/s Delkin Black 512GB (est 23fps*) [SSD Review=1412.97]

* Estimated FPS *potential* based on lossless compressed raw at 65MB per image.

So the higher end cards are definitely all capable of more then we can realize in-camera at this time.
 
With file transfers there is overhead in creating packets and then disassembling them at the receiving end. There is also the overhead of verification that the data was received which is commonly done when the speed of the devices permit this. With floppy drive the data was written and no verification that the data could be read back. With hard drives this changed and write verification would take place. I would expect this to be the case with file transfer software as well. I know that if I move an entire folder to a different device the process takes far less time than if I select all the individual files in the folder and move them to the same device. This is with the Windows operating system and it may be different with a more efficient OS like that from Apple.

Keeping the data on the device is the most efficient way to go. With 4K 60p video and the Z9 I have moved away from external recorders and use very large capacity memory cards instead. On my workstation I have dual NVMe memory modules inside and connected directly to the motherboard and in a RAID1 array to maximize data I/O and have redundancy. Unfortunately NVMe M.2 modules have proven to be less reliable than either hard drives or standard SSD drives so always compromises involved.
 
VPG 400 (Video Performance Guarantee) is probably as relevant for card performance, notably video. This recently updated certification underwrites the video performance of CFexpress cards to sustain at least 400 MB/s write speeds without dropping a single frame.
Not all vpg 400 rated cards are CFexpress 4 rated, however.




 
Last edited:
Back
Top