XQD Discussion

Ray

Looks like Sony changed their XQD cards.

The previous cards looked to be:
H series 125 Mb/s read/write
S series 168 Mb/s read/write

The new ones are
N series 125 Mb/s read and 80 Mb/s write
S series 180 Mb/s read and write

Anyone have experience with the new cards in the D4?
Will the N series at 80 Mb/s write result in a reduced buffer?
Any problems with the new ones?

Thanks

Golf007sd

The faster the better in allowing the body to write faster to the card. I have the 32gig 168 S version and love it.

FlowtographyBerlin

Hasn't been updated for a while, but for most cards except for the newest ones, these test results are quite interesting concerning the actual speeds (vs. the claimed ones, a dramatic difference for most cards) of memory cards. Has also been tested with a plethora of camera models. Check it out:

http://www.robgalbraith.com/camera_wb_multi_page0ea3.html?cid=6007-12444

Ade

Hey Flow welcome back!

@Golf: looks like the D4 gets maxed out at or before 168 MB/s, so the faster 180 MB/s card will not help the body to write to the card any faster.

Golf007sd

All I can tell you is this: I do see a difference in write speed and download to on the S-version of the XQD card. Never the less, XQD cards are FAST, very FAST.

Ade

Golf I gather there is a difference between the H-series (125 MB/s) and the old S-series (168 MB/s).

Now there's a new S-series (180 MB/s). According to Sony's specs, the D4 RAW buffer write performance with the new S-series remains the same as the old S-series, both achieving 108 compressed RAW frames at 12-bits.

paulr

Here is Sony's Press Blog.

http://blog.sony.com/press/sony-expands-line-of-ultra-fast-xqd-memory-cards-for-seamless-and-stable-professional-4k-shooting/

Juergen

All I can tell you is this: I do see a difference in write speed and download to on the S-version of the XQD card. Never the less, XQD cards are FAST, very FAST.

I do not see my D4 improving in it's performance with 180 MB/s write speed. And actually, there is no need for any performance boost. Even in sports I am shooting short bursts at the highest frame rate, then re-focus. I never ran into the issue with the buffer or writing speed.

The only good news to me is, that Sony keeps on developing the XQD format. I wish Nikon had gone all the way with it and put 2 XQD slots in the D4 as well as the D800.
Imaging somebody has both bodies. He then should configure the CF Card slot in the D4 as a prime slot in order to avoid having to carry two card reader all the time. Sick, isn't it?

Or did someone see a combined card reader yet?

Jürgen

Ade

Lexar has a new 4-slot parallel card reader (the HR1 Workflow Hub) which supports any combination of SD, CF or XQD modules (each sold separately).

http://www.lexar.com/workflow

Juergen

Lexar has a new 4-slot parallel card reader (the HR1 Workflow Hub) which supports any combination of SD, CF or XQD modules (each sold separately).

http://www.lexar.com/workflow

Hi Ade,
Thanks for the link. With that my wife probably would not complain so much about the cable mess on my desk :-)
I was more looking into a travel solution, at the size of a normal card reader. But that does not seem to be available.
Jürgen

Golf007sd

Well...well...well... looks like V 2.i is very promising; moreover, the whole argument that XQD will not be used seems to be a bit over stated.

CFA Announces XQD Ver. 2.0 Specification

Key features of XQD Ver.2.0 include:

Dual Interface
-PCIe Gen2 (5Gbps) x 2 Interface
-USB3.0 Interface

Provides an interface speed of 1GB/s (1000MB/s).

This speed enhancement enables a new generation of higher performance cards to meet requirements for professional 4k video recording requirements.

Erase Block Stream Function

A function to ensure that garbage collection does not occur within the media.

Professional Form Factor
38.5mm x 29.6mm x 3.8mm

TaoTeJared

I wish Camera makers would just make the move to XQD - It out performs everything and will be the future. CF Fast is not backwards compatible and is a different connection so adoption would require a new card anyway. The more that adopt it, the quicker the cost will drop. Most of us have been using CF and SD cards for the better part of 10 years and are on borrowed time on both Techs.

Ade

@Golf007sd

The XQD 2 spec is actually two years late, which tells us something about XQD's adoption. Here are a couple posts on our sister site from way back in 2012:

http://photorumors.com/2012/07/28/compactflash-association-announced-xqd-2-0-specifications-cfast-2-0/

http://photorumors.com/2012/09/15/is-the-xqd-memory-format-dead-already-xqd-2-0-specification-announced/

And looks like they've actually "downgraded" the XQD 2 spec a bit vs. the initial design in 2012. Most notably, the XQD 2 spec will now be based on the older PCIe 2 standard (vs. PCIe 3 originally designed). And it looks like they've made some other concessions, such USB 3 compatibility, which has a "realistic" transfer speed of only 400 MB/s.

Hopefully these concessions are enough to convince other camera / device manufacturers to support XQD. Otherwise I think we will continue to see a split between CF / CFast / XQD adoptees.

Lastly, from what's been announced so far, it's not clear to me that they've preserved full "forward compatibility" of XQD 1. I.e., there is a possibility that XQD 1 cards may only work in some (but not all) XQD 2 cameras / devices.

Golf007sd

If memory service me right, CFA announced the specification of XQD back in December of 2011. At that time they said it would have the "Scalable High Performance Interface" via "PCI Express: 2.5Gbps today and 5Gbps Future."

Whatever "concessions" have or are being made, the outlook is promising. Let's see where the chips fall.

TaoTeJared

Hell CF "specs" allow it to store over 100PB (petabytes) of data.
Using the same rational Ade, I guess CF was really slow to be adopted.

Value Metric
1000 kB kilobyte
1000^2 MB megabyte
1000^3 GB gigabyte
1000^4 TB terabyte
1000^5 PB petabyte

Good thing about not being "really" adopted is that if "gen 1" of XQD is phased out, it really won't make a difference. a D4 and a few Sony higher end video cams are the only things using them. 6 years time they will be at end of life cycle anyway.

It's the available Head Room for XQD that is promising and makes it the best choice for what we have out there now.

Ade

CF's adoption was one of the fastest in the history of consumer electronics, especially considering that CF was a proprietary SanDisk format at the time. Within one year of its announcement, CF was backed by 12 major companies, including Apple, Canon, Kodak, HP, Panasonic, Polaroid, Seagate and SanDisk. These 12 companies formed the CF Association to takeover stewardship of the CF standard.

As a comparison, four years after XQD was first announced in 2010, only Nikon and Sony have cameras using the format.

XQD 2 was announced in the summer of 2012. The planned technical revision was to replace PCIe 2 with PCIe 3. The XQD 2 standard was supposed to be finalized in the Fall of 2012. That didn't happen. By mid-2013 it was pretty clear that XQD 2 as originally proposed was dead-on-arrival. SanDisk, Canon and ARRI decided to back CFast 2 instead.

Now we have a new XQD 2 proposal using PCIe 2 x2 plus a USB3 mode. Some technical (and cost) concessions were clearly made, hopefully enough to entice other players to finally back XQD.

The XQD 2 final spec won't be generally available until November 2014 so we'll have to wait until then for a more in-depth analysis.

TaoTeJared

The XQD 2 spec is actually two years late, which tells us something about XQD's adoption. Here are a couple posts on our sister site from way back in 2012:

Hell CF "specs" allow it to store over 100PB (petabytes) of data.
Using the same rational Ade, I guess CF was really slow to be adopted.

CF's adoption was one of the fastest in the history of consumer electronics, especially considering that CF was a proprietary SanDisk format at the time. Within one year of its announcement, CF was backed by 12 major companies....

Thanks for proving my point.

Golf007sd

Correct me if I'm wrong (again), but SanDisk was very much involved in the creation, adoption and implementation of XQD technology in the first place. In fact, "SanDisk participates in many standards bodies and has contributed to a variety of new standards that allow for options in the marketplace..."

I do not know why SanDisk did not bring to market their own version of XQD memory cards, but perhaps that was an agreement they reached with Sony. Never the less, XQD memory cards are far better, in more ways than one, when compared to CF memory cards.

XQD is much like Thunderbolt is in relation to USB...and look how long it is taking for hardware companies to adopt it...even with Intel and Apple behind it.

Ade

@TTJ Really, it's easy. CF: 12 companies adopted the CF standard in 1 year == fast adoption rate. XQD: only 3 companies adopted the XQD standard in 4 years == slow adoption rate.

@Golf:

The problem with XQD right now is that it offers very little benefit vs. the cost to implement it. XQD's main advantage is supposed to be speed, but in reality, here are the fastest transfer speeds for different type of cards as of today:

CFast -- 500 MB/s (Lexar 3333x CFast 2.0)
SD -- 280 MB/s (SanDisk Extreme Pro UHS-II)
XQD -- 180 MB/s (Sony new S Series)
CF -- 160 MB/s (Lexar 1066x CF UDMA-7)

So basically XQD today offers no performance advantage over SD or CFast, and only a marginal improvement over CF.

I think the re-designed XQD 2 spec is an attempt to solve the cost issue by: 1) leveraging the older PCIe 2 standard vs. the current PCIe 3 standard; and 2) providing an alternative (cheaper) interface using USB 3. These design concessions might be enough to entice other players to finally support XQD.

Ironically I believe the only way CFA can guarantee full compatibility for existing XQD 1 cards is to require all future XQD 2 host devices (e.g., cameras) to support both the PCIe interface and the USB 3 interface -- which adds costs and complexity. This does allow cheaper XQD 2 cards to be made available. Or, they could drop full compatibility between the two versions.

Golf007sd

@Ade: Speed is just one benefit of XQD technology, as well as, CFast 2.0 and USH-II. Durability, capacity options, over all functionality are some of the other benefits. XQD has proven it can do all of this.

As for "the cost" of implementation, I do not see that being the drawback for its adoption. It is more of having the producers redesigning their gear. Moreover, the cost of either the CFast 2.0 and Extreme Pro USH-II memory cards are not cheep either to say the least.

Lastly, at the moment only two cameras take advantage of CFast 2.0 or the USH-II: the newly released Fuji XT-1 can take advantage of the SanDisk Extreme Pro USH-11 and the ArriI Amira for the Cfast 2.0. I would argue that this is very much like the D4 market.

Personally, I welcome all these forms of fast memory interphases...

Ade

Golf, unlike XQD which is a net new interface, all the other technologies are extension of existing standards, and therefore have a cost advantage to implement.

E.g., the added cost for Fuji to implement UHS-II vs UHS-I is likely negligible. It's a given that in the next year or so most new cameras with an SD card -- including Nikon's -- will adopt UHS-II.

TaoTeJared

Ade you may want to check this article out:
http://www.xbitlabs.com/news/storage/display/20120726143026_Sony_and_Nikon_Begin_Development_of_XQD_2_0_Memory_Card_Format.html
I found 12 others that have all said XQD 1 & 2 will be backward/forward/whatever-ward compatible. I'm also finding everywhere that it is still based off of PCIe 3. Maybe by the time you find the answer CF will finally be adopted. =D> Oh and btw - XQD is only 2 years on the market - specs are 4yrs old.

Correct me if I'm wrong (again), but SanDisk was very much involved in the creation, adoption and implementation of XQD technology in the first place. In fact, "SanDisk participates in many standards bodies and has contributed to a variety of new standards that allow for options in the marketplace..."

I do not know why SanDisk did not bring to market their own version of XQD memory cards, but perhaps that was an agreement they reached with Sony. Never the less, XQD memory cards are far better, in more ways than one, when compared to CF memory cards.

XQD is much like Thunderbolt is in relation to USB...and look how long it is taking for hardware companies to adopt it...even with Intel and Apple behind it.

SanDisk, Sony and Nikon were the three that pushed it. Nikon being the oddball in that group. I'm guessing Sony is making a controller board/card/chip somewhere in the mix and Nikon uses them. What is more odd than SanDisk not making them, is that Sony isn't really utilize the tech at all. I think 1 or 2 vid-cams are it last I looked.

XQD is designed for 2.5Gbs read/write speeds - None other comes close.
I'm waiting for the SD cards to start failing due to heat with that much sustained write times. That is why industrial companies that utilize small computers in harsh environments go with CF - it can withstand harsher extremes. I've tested that, seen it, and just waiting for it to happen in a camera.

Golf007sd

Agreed, XQD does require a new interphase. But so does the new inter working of the CF 2.0 cards; hence new chip within the body, as well as, the connection.

Ade

@TTJ

Check the date on that article you quoted: 2012.

Back in 2012, the original proposal was to use PCIe 3 for the XQD 2 with full backwards/forwards compatibility. But the XQD 2 spec as proposed in 2012 never materialized.

Fast forward to 2014, there is now a new, completely different proposal for XQD 2, based on PCIe 2 (x2) -- not PCIe 3 -- and USB 3. The inclusion of USB 3 into the XQD 2 spec will break compatibility. How exactly the compatibility will be broken is an open question. My guess is some XQD 2 cards which will not work in existing XQD 1 devices like the D4.

I'm not sure why you think XQD is designed for 2.5Gbs read/write speeds and "none other comes close". 2.5Gb/s is about what SD UHS-2 achieves today and CFast 2 is already at 4 Gb/s real-world performance.

TaoTeJared

Yes that was the first release of the standards. You shouldn't have a hard time following the trail from there. I found articles going up to the beginning of this month stating the same specs and none your claims of them dropping things backwards. You have not provided any proof for what you keep writing about.

I'm not sure why you think XQD is designed for 2.5Gbs read/write speeds and "none other comes close". 2.5Gb/s is about what SD UHS-2 achieves today and CFast 2 is already at 4 Gb/s real-world performance.

Then show us the cards.

Ade

From the Golf's original post which started this current discussion:

"Key features of XQD Ver.2.0 include:

Dual Interface
-PCIe Gen2 (5Gbps) x 2 Interface
-USB3.0 Interface

- See more at: http://forum.nikonrumors.com/discussion/1990/xqd-discussion#Item_24"

I already posted cards faster than 2.5 Gb/s. The target design rate for XQD 1 has always been 5 Gb/s from day one -- 5 Gb/s being the speed of a single PCIe 2 lane (technically, 5 GT/s).

Early XQD implementations could only support PCIe 1 speeds, 2.5 GT/s. But the standard itself has always targeted 5 Gb/s since it is based on PCIe 2.

The original XQD 2 spec had a design target rate of 8 GT/s, the speed of a single PCIe 3 lane.