It seems like hyperbole to say, but the CFAST vs XQD battle is already over. For the consumer, this is a good thing. For the last couple years we have seen the memory market segmented over storage, and with CFAST being backed by the Compact Flash Association (CFA) (despite it’s weaknesses when compared to XQD), it seemed like we were going to be stuck in a losing fight between two standards, one backed by the CFA, Arri, Canon and Blackmagic, and the other backed by Sony and Nikon. A battle that would continue to drive a wedge between companies, force prices higher, and hurt the end consumer.

I was planning on writing an article that objectively showed the positives and negatives of these two memory types, and looked to inform the community of photographers on how to “make decisions with their wallets.” I wanted us, the consumer, to help push this market in the right direction because let’s face it: the cost of high end cards was rising to outrageous levels.

That was my intention, until some rather unexpected news hit my desk. The article I was going to write no longer matters, because there has been a huge step in the right direction for the future of memory. Last month, the CFA announced a new card type: CFexpress. But to understand why this is important, we need to look at data transmission and the complicated world of politics involved.

But trust me, CFAST has already lost.

CFAST vs XQD

I think to best understand why this is happening in the first place requires the knowledge of each of the two major card types currently out there for high speed, large storage in CFast and XQD. These cards are vastly different in not only how they look, but how they operate.

In order to get a grip on what was going on here, I spoke to Sony, Lexar and SanDisk independently. Initially I wanted to get a firm grasp on the advantages and disadvantages of both the XQD and CFAST formats, but in the end it helped frame a better informed picture on why the CFA made the decision they did.

First, of the major card manufacturers/players in this conversation, here are where each of them stood:

LexarSanDiskSonyCFA
ProducesSD, CF, CFAST, XQDSD, CF, CFASTSD, XQDOrganization for memory producers, not manufacturer
Notably SupportsXQDCFASTXQDCFAST

In the case of Lexar, they “supported” both CFAST and XQD, but in conversations with multiple representatives of the company, it was clear at the time that they saw a greater future in XQD than in CFAST, hence why I have listed them as XQD supporters. The CFA, until recently, backed mainly CFAST, and not XQD. This is likely why we have seen greater adoption of host products from Canon, Arri and Blackmagic, with greater companies producing the standard (like transcend, SanDisk, etc). SanDisk followed the CFA, and chose to not even produce XQD cards over the past couple years, abandoning the format entirely (more on that later).

And here are the major host companies, and the formats they support:

CanonNikonSonyARRIBlackmagic
AcceptsSD, CF, CFASTSD, CF, XQDSD, XQDCFASTSD, CFAST

As you can see, XQD is used in a lot fewer cameras than CFAST today, and especially when you consider Sony still has not put XQD in any of their photographic cameras, electing to only place them in their high-end camcorders. With the backing of CFA and the fact only Nikon has been a loud supporter of the format, it’s no wonder CFAST has been more common in the last year.

But of the two card types, what is the difference and why does it matter? Well, it depends on who you ask (which absolutely should not have been the case). Basically, Sony rode the middle ground, Lexar was highly supportive of XQD, and SanDisk threw their entire support behind CFAST. Confusingly, I even received conflicting information on what should be fact about these devices. Take a look:

LexarSanDiskSony
Format for the FutureXQDCFASTXQD
Max Speed of CFAST600 MB/s600 MB/s, but perhaps greater600 MB/s
Max Speed of XQD1000 MB/s, but perhaps greater500 MB/sCan’t Quite Say

Lexar and Sony told me that they both believed that XQD was the format of the future, while SanDisk said they were putting their weight behind CFAST.

There are two places here that are a bit surprising. First, that SanDisk told us that the limit of speed for XQD was 500 MB/s. This, I have found, is not true. The limit is far greater than that, with Lexar estimating 1000 MB/s (but it is expected to be even higher than that). The second thing is a related note, with Sony’s hesitation on saying the speed limit of XQD, and this is likely because they simply don’t know yet: the limit has not been reached.

In any case, Lexar seemed to be in the strongest position to make statements on these card differences, seeing as they are the only major manufacturer to be producing both CFast and XQD. SanDisk doesn’t produce XQD, which may be a reason for their lack of knowledge on speeds. At any rate, I deferred to Lexar in many places here because they seemed to be more informed on the wholistic picture. 

Why XQD was the Smarter Choice

XQD is going to be the winner here for a number of reasons, but the most basic one is at the core of the two products: the type of interface. XQD is PCIe, while CFAST is SATA.

PCIe is a more “scaleable” type of interface, while SATA was never really designed to be scaleable at all. SATA has one definition and is capped at 600 MB/s speeds. This is not a “theoretical” cap, this is a true cap. It can’t go faster. As you can see, this is problematic since the requirements for speed and efficiency on the side of cameras in the last two years has gone well beyond SD, and with 6K and 8K just around the corner in consumer devices, there is just no way that 600 MB/s is going to be enough.

But on PCIe you can add what are known as “lanes” and therefore it can scale. You can’t do that on SATA, it’s just not part of how the interface is built.

Because SATA cannot scale, is perhaps why CFAST was doomed from the start.

PCIE can be combined together to expand upon each other. For example, if something required more speed beyond what one card could do, PCIE allows to combine two cards to reach those speeds (create more lanes) and then software could be developed to read both cards together as one unit, which allows you to go way beyond the rates of what you see right now. Think of it like RAID 0: more drives for more speed. It’s a bit dangerous, as if one drive fails, the entire data packet is lost, but at least it is an option.

Another benefit to PCIe is backwards compatibility. Let’s say you built a host platform (like a camera) on PCIe Generation 1. If you then created PCIe generation 2, the reader for Generation 2 would be able to read Generation 1, and Generation 1 host platforms could accept Generation 2 with a simple firmware update.

XQD is a PCIe Gen 2, but if a new host which is PCIe Gen 3 was created, it will be able to read media on PCIe Gen 2. It can support Gen 3, Gen 2 and Gen 1.

So why does this matter? Well, because the CFA’s new format, CFexpress, is PCIe. 

The Advent of CFexpress, XQD Compatibility, and Why You Should Stop Buying CFAST

On September 7, 2016 the CFA decided that competing standards was not the way of the future, and likely realized that their continued majority support for CFAST over XQD was not a winning strategy in the long run.

standards

Luckily not the future here, via xkcd

Rather than allow the two competing standards to drive up the price of memory, the CFA and its members developed a new type of memory: CFexpress. The best part? Because it’s PCIe and because PCIe is backwards compatible, XQD will continue to work on CFexpress cameras… and, theoretically, any camera produced to work with CFexpress will work with any XQD card currently on the market.

Which means that buying CFAST is a dead-end decision. Buying XQD? That’s actually future proof.

CFexpress is the next evolution in an open standard format for a broader set of applications, such as professional imaging, personal computing, servers, IoT and industrial. CFexpress differs from prior CFA standards in that CFexpress is a specification spanning multiple form factors and performance levels to match various market requirements while maintaining the same electrical/physical interface and memory/storage protocol. The new CFexpress specification is under development within the CFA CFexpress Working Group.

CFA’s goals for the CFexpress family are:

  1. Unify the ecosystem around a removable storage specification that has longevity in the market
  2. Provide a platform that scales with performance increases in the PCIe interface
  3. Leverages the technologies and skillsets of the larger volume compute markets
  4. Seamless compatibility with the capture ecosystem and workflow tools.

The CFA is putting its backing to this new format for three major reasons:

Key FeatureBenefit
PCIe Gen 3 InterfaceProvides a max theoretical interface speed of 1.0GB/sec per lane of PCIe. This speed enhancement enables a new generation of higher performance cards to meet requirements for professional video recording.
Family of Form FactorsProvides multiple form factors to address multiple market segments with a unified physical interface and memory/storage protocol. 2 lane and 4 lane form factors will be the initial focus but anything from 1 lane to 8 lanes is possible. This provides a range of up to 8GB/Sec with PCIe Gen3.
NVM ExpressA storage interface designed from the ground up for NVM storage. The NVM Express protocol features fixed length commands, an efficient queuing model, and very low noncacheable PCIe accesses per command, delivering reduced I/O overhead and performance improvements in comparison to legacy storage device interfaces. NVM Express is gaining popularity across various flash-based storage solutions and has driver support on all leading OSs.

The decision to move forward with CFexpress has already been made. The spec for 1.0 should be ready by November/December. Samples should be ready for testing with host companies (that means anyone who wants to build hardware to accept the memory can have a sample to begin testing) by the first half of 2017. That means, theoretically, we will see CFexpress hit the market and have host products available by as early as 2018. That is much, much faster that I ever anticipated a “winner” between CFAST and XQD being decided, let alone a new format being adopted by the CFA.

We have been told that CFexpress’s form factor is going to be identical to the XQD form factor. This is huge news for Nikon and Sony users who have already invested in XQD. Their cameras will be able to be ready for the future via a simple firmware update. With that, the cameras will be able to accept both XQD and CFexpress. Unfortunately for those who backed CFast, and those who invested thousands of dollars into cards, the format will undoubtably 100% die, just as CF is.

XQD Won, CFAST Lost

From the get-go, XQD was going to be the winning pony. With the PCIe interface instead of CFAST’s SATA, it was undoubtably more future-proof than the aging infrastructure of CFAST.

That’s why I feel confident in recommending anyone who wants to be financially responsible to not buy CFAST anymore. Here are the facts and recommendations:

  1. The CFA is backing a different card type as the “future,” which means it’s highly unlikely that any camera developed after 2017 will take the CFAST card format
  2. CFAST will die. It’s not an if, but a when. And based on the speed at which CFexpress is to be rolled out, I predict before 2020
  3. CFAST is not backwards compatible (why do you think you needed a new card reader when you “upgraded” from CF?)
  4. CFAST pricing is already absurd. Don’t waste your money on something that’s not going to last

We got to this point because Sony and the CFA managed to come to terms on something. Sony before owned the format, and licensing of the format was theirs, and theirs alone. This attitude is likely a major reason why the CFA supported the CFAST card, as the format really was far more “open.” But with the announcement of CFexpress, it’s pretty clear that Sony loosened up a bit, and that was a good thing for us as the consumer. More manufacturers, more people on board and everyone using one system means better product, and lower price.

And that’s a win for us.

[All information above courtesy of Lexar, SanDisk, Sony and the Compact Flash Association.]

  • Erik Stenbakken

    +1 for Sony users!

  • Steve Simon

    seems like big news but haven’t seen this information circulated around the web yet…good job! (if it all proves true of course)

    • Not many people follow what the CFA does, considering they spend more of the time rather silent because what is there to say? But given the conversations I’ve had with Lexar and the fact that the CFex information is public on the CFA website, the card is real and coming, that much is certain.

  • I would love for this to be true but I lived through the VHS-BETA era and know that the best man doesn’t always win #feelthebern
    Canon still have the greater market share of worldwide camera sales and are in a position to get their way if they push hard enough. If Lexar are forced to choose between the two formats and CFast sells 60% more just because Canon sells 60% more, they’ll choose CFast.

  • AStarbucks

    So typical of Canon to back a loser antiquated standard like CFAST…

  • And what about now that Lexar sold its XQD division?

    • Chaitanya

      Lexar XQDs are dead.

      • Marty Haas

        Not so fast. The Phoenix is rising.

  • eeddcc

    I wouldn’t agree to “already over”.
    A PCIe interface is nice if you want to use it in a desktop PC because you have direct access to its internal bus. So you can get more speed because there’s no limiting SATA controller between CPU and flash memory.

    But: PCIe is meant as general extension bus. In the past there have been several compaibility issues with PCIe SSD maybe because the standard is too general.

    However, SATA has been designed as storage interface. There are proved standards for internal and external connections, hotswapping and you can even connect SATA devices to an older PATA system using standard adapters. So I would expect fewer compatibility issues.

    Here in Europe you can get Cfast card readers at a quite reasonable price. They are desined for fixed installation in the 3,5″ bay and have to be connected to the internal SATA port of the mainboard. This is an effcetive and logical setup IMHO.

    On the other side I’ve only found an super expensive external USB 3.0 XQD card reader. I would suppose that it’s high price is caused by a kind of USB 3.0 to PCIe-bridge chip inside the device.

    To you have a pc system with a kind of [system’s-PCIe] –>(USB controller)–> [USB 3.0] —>(bridge chip)-> [card reader’S PCIe] –> [XQD-card] configuration. Which is IMHO quite extravagant and costly, and not as compatible as SATA. If it can take advantage of the PCIe interface strongly depends from the USB 3 configuration on the host computer side.

    SATA ist the more evolutionary standard to me. The PCIe connection of XQD cards is in contrast revolutionary, but the way it is currently accessed by computers seems odd to me…