cfe-b card performance test

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).

so you have one 2x2 port
Thanks, I was not sure.

210323 rear of Motherboard Type-C port.
2023-03-21 10_46_06-CrystalDiskMark_40gpsCableMB_2X2Port.jpg
You can only see EXIF info for this image if you are logged in.

180323 IO-PCE3242-1C Card Type-C port.
CrystalDiskMark_8_0_4_x64.jpg
You can only see EXIF info for this image if you are logged in.
 

Yes, John is correct. The rear USB C type port is 2x2.
 
Thanks, I was not sure.

210323 rear of Motherboard Type-C port.
View attachment 57249
180323 IO-PCE3242-1C Card Type-C port.
View attachment 57250
Interesting that the Motherboard port is slower than the port on the PCIE card. Do you have a reader that supports 2x2 by any chance? It would be interesting to see if there is a material speed increase. Also, to see how USB Device Tree describes the reader.

Although my Angelbird reader is described by Angelbird as supporting the 2x2 protocol, USB Device Tree is showing it as a Gen 2 device. So today I have ordered this reader which also purports to support Gen 2x2:


I will be interested to see if it performs any differently to the Angelbird and will report back in due course!

In the meantime I will take a look at the links you have regarding other USB tools. Many thanks.
 
No Reader that supports 2x2, will see if I can find one in OZ that is not too expensive.

No problems about the links they are only a few of the many that came up during USB searches.
Beware some of the USB ones will have scary warning notices during the install phase!
 
Yah, I'm taking these reader claims with a big grain of salt. I suppose there isn't a huge incentive for them to make them better since ~1000Mb/s tend to be fine for most uses. I'm eager to see how that PCI card functions.
 
@Alistair That's not good. Does it perform to 2x2 levels?

USB Device Tree View may not identify each devices information correctly but I have not found anything else yet that does as good or better job.

The different performance for CFexpress cards+reader+USB port is no different to SD cards+readers+USB ports though.
I have been testing some 300M/s type UHS-II SD cards+readers+ports this morning.
The test results did vary considerably depending on which card+reader combination you used.
Speeds range from around 90MB/s to 250MB/s for the same SD card in different readers, same USB port.
No cable attached readers were used in this testing, only blue USB ports on hubs and motherboard ports, hubs no noticable difference in speed.

The card+reader+USB port+cable combination all play a part in transfer rates for CFexpress cards but how to decide which one/s to buy is the real question.

Thanks @John Navitsky for piquing my interest in CFexpress card speeds.
The testing you posted prompted me to test my setup which now allows me to use the highest speed reader+USB port combination when downloading files.

Will be interested to hear about the results from the internal PCI card reader.
 
all numbers WRITE

#1 -- 1319.70MB/s -- DIY Sabrent Rocket 2230 512GB NVME in a CFE-b all-alumninum shell, thermally paste bonded
This is what was of interest to me, with the Sabrent Rocket 2230 1TB selling for about $160 it looks like a cost effective way to create a high capacity CFExpress card.
 
got the sabrent thunderbolt 3 / usb reader. tests out about the same on the old macbook pro. pretty sure the macbook pro is the limiting factor here, so no useful data.

that said, based on limited use, i think this is my fav reader. just about the right size, the rubber cover is nice, and i like the push to release mechanism, and i like that it's both thunderbolt 3 and usb.
 
got the sabrent thunderbolt 3 / usb reader. tests out about the same on the old macbook pro. pretty sure the macbook pro is the limiting factor here, so no useful data.
Is that using the thunderbolt port? I understand your Macbook has both USB 3.2 Gen 2 (10Gbs) and Thunderbolt 3.0 (40Gbs). Its this one? : https://support.apple.com/kb/SP749?locale=en_NZ
If I understood your original post correctly, you connected your OWC Envoy Pro FX to your Macbook Pro via thunderbolt and got > 2113.70MB/s. If I have that right, the Macbook probably isn't the constraint. It is most likely the cards, since you are getting roughly the same performance on USB 3.2 Gen 2 and Thunderbolt 3.0.
that said, based on limited use, i think this is my fav reader. just about the right size, the rubber cover is nice, and i like the push to release mechanism, and i like that it's both thunderbolt 3 and usb.
 
omg, i didn’t know that 🤣🤣🤣. i will investigate

edit: wait, what makes you think the ports have different capabilities? i read this as all four ports can do either:

Charging and Expansion​

Four Thunderbolt 3 (USB-C) ports with support for:

  • Charging
  • DisplayPort
  • Thunderbolt (up to 40 Gbps)
  • USB 3.1 Gen 2 (up to 10 Gbps)
 
omg, i didn’t know that 🤣🤣🤣. i will investigate

edit: wait, what makes you think the ports have different capabilities? i read this as all four ports can do either:
Oh, you are right. All 4 ports must be thunderbolt. Some models had USB on one side and thunderbolt on the other with the thunderbolt symbol on the appropriate ports.
 
Oh, you are right. All 4 ports must be thunderbolt. Some models had USB on one side and thunderbolt on the other with the thunderbolt symbol on the appropriate ports.
that said, i wonder if the auto detection of thunderbolt vs usb is working with my card readers. you guys are getting better performance with usb
 
Last edited:
I don't know what the x1 means in this context. Possibly doesn't refer to lanes as the x2 in Link Status above it is not a reference to lanes. As you point out, that port has already supported >2100MB/s to the OWC. So either the card or the port are maxing out. The CF Express B PCIE slot will provide some answers (probably along with a bunch more questions!).
 
yah, i'm tired of dealing with usb/thunderbolt/reader in this chain. it makes me think basically everyone's benchmarks are suspect with regard to understanding the speed of the actual cfe-b being tested

unfortunately, the eta for the card is april 10
 
Following up the idea that there may be big differences between all parts involved I tested the laptop I have.
Using the same card/reader/cable for my initial tests (see above) I tested the 2 USB-c ports available on the laptop and was pleasantly surprised with the results of both ports.
Results were close for read/write to the main computer setup (see above).
I now know at least the parts I am using give respectable results so I can have some confidence in the card/reader/cable/computer combinations and the respective read/write performance results.
Not sure what implications the lower values of the last 2 tests may mean, hopefully not much.

2023-03-26_Laptop_s.jpg
You can only see EXIF info for this image if you are logged in.
 
Just to be clear, how much actual time does a faster card save? I suspect that the only test which counts now is heat. I am guessing that heat is some sort of indicator of durability, but that just a guess.

I bought 2 Delkin Power cards 128, they seem just fine. I also bought Delkin Black 150 and it is fast enough. When it was on sale I bought a Delkin Power 512Gb and it seems a bit slower but not enough to actually delay anything when using the camera. It clearly states it isn’t designed to do 8K video (which I don’t plan to do anyway).

I upload my files via Sony card reader (MRW-01) using the cable that came with it, into PhotoMechanic and that lets me pick the images I want to keep. There is no real waiting for images to appear on screen. It’s actually pretty good. Loading the images into Capture 1 after placing them into relevant folders is slower than PhotoMechanic but that has nothing to do with the card.

The computer is a late 2017 iMac using a Thunderbolt USB C port, so nothing special. ( I am not sure when I will update to a faster computer).

I do have one Sony Tough 128Gb CfExp card and a Lexus 128Gb. There may be some tests which suggest these are slower but I just haven’t seen much evidence that they are slower.

I have read several reviews of card speed but I don’t see the point of taking too much notice if the maximum speed depends on the card and cable. The tests which show big differences have been done using the cards as a direct SSD card.
 
it depends on your objectives. personally i’m looking for cards that enable potential performance. that is to say, cards that may allow for 20fps sustained capture in lossless raw, or higher on one of the other raw formats.

note there are some aspects of speed that may not be as obvious like if you are in backup mode you need to write twice as much data, so if you can sustain 20fps to one card, that may mean you can only sustain 10fps in backup mode, so it isn’t necessary as fast as it may seem

in addition i simply want to know how a given card performs in a objective way since manufacturers numbers tend to be opaque or suspect

but yah, there are diminishing returns in our current environment. with our current cameras/firmware, we can only go so fast, and on the desktop, most usb/readers only go so fast, and also, even if they were faster it wouldn’t make a huge difference
 
Last edited:
I understan the 2 card thing. The innovative use of a dual channel system on the Z 9 may eventually lead to simultaneous dual card recording if a 3 channel bus can be set up.
The question is whether it is just smarter to improve reliability and cost of the single card or maybe just make backup to card 2 happen automatically when the camera is not being used to focus or shoot.
I am also wondering if 2 cards is as important with CfExpB compared to the SD card era.
 
i would love it if they could have the backup to the second card happen when you are NOT shooting and i've submitted that as a suggestion via Brad Hill's FW suggestion list.

it seems like right now you have a fixed amount of bandwidth. and it seems like that tops out right now at like 12 fps if you're shooting lossless compressed raw. so that pretty much means if you are mirroring that you can probably only sustain about 6 fps.

that said, it didn't have to be that way, it comes down to whatever decisions the mfg makes. and we also don't really know exactly what the bottlenecks are. we know that 20 fps is about 1300MB/s and we know that some of the cards can keep up with that or come close. but we also know that it seems like we can only write about 12-13 fps sustained. so there are some bottlenecks there but we don't know if it's in the processing, or if there are limitations in getting the data to the card (like how you are limited when you write via usb).

as for the single vs dual card thing, it's an age old issue. it really just comes down to how bad would your life be if your card died. and if the answer is REAL BAD, then the second card makes sense.

the other gotcha is that the bigger the card, the more pain you might feel if the card died. so, if i've been out shooting all day and i have 400GB of stuff on my card, that's going to hurt a lot more than if i shot all day, but that's on four cards and one of my four cards died.

over all, i find the cfe-b cards very reliable, but also, everything fails
 
Back
Top