XQD Cards

Nikon Rumors just basically reposted what Camera Memory Speed published about XQD cards in the Z7. 

Having now tested a wide set of cards in my Z7—I intentionally bought one of each type of XQD card along the way since the D4 started using them, though not at all sizes—I can say that my results are very similar to Camera Memory Speed's: the Z7 appears to top out near 245MBs with cards that are labeled 400MBs. My D5 has hit ~300MBs on some of those same cards. Put another way: the camera's card slot is almost certainly the limiting factor in how fast the camera performs. 

There are many who are making an assumption that when CFExpress firmware is added to the Z7, it will get faster (ProGrade has a 1GBs card in prototype). I do not subscribe to that theory and expect that CFExpress won't have much impact on speed, if any. History has proven that card slot speed is pretty much hard coded at a maximum by the camera makers, and given that we have underperformance to what the cards can already do, I'd say that the Z7 slot is likely limited somewhere near 250MBs. CFExpress as Nikon is likely to implement with the Z7 is still going to be two lane PCIe, just like XQD, so will be limited by the electronics of the camera to similar speeds.

If you want best performance in the Z7, you need to use a Sony G Series, a Lexar 2933x series, or the new Nikon cards (I'm still waiting for a Delkin card to test, but given their 400MBs rating, I'd expect it to be similar to the Sony G). Simple as that. I was actually surprised at how poorly some of my older XQD cards did in the Z7. I actually stopped using my oldest Sony cards on a recent shoot because I was hitting full buffer and getting poor rollover rates (<1 shot a second once buffer was full; on the fast cards it was over 3 fps). 

Camera Memory Speed found something else I can confirm: while Lossless Compressed writes slightly slower than Uncompressed (<5%), the full buffer clearance speed is faster by almost 20% and produces about 3.5 fps when the buffer is full (the file sizes are smaller). 

So nothing is really going to change from my card recommendations with the D850, I think. 

Well, okay, possibly one thing: I can't reliably track continuous focus on the Z7 as I can on the D850 (let alone the D5), so I'm actually pulling off the shutter release earlier most of the time. Thus, the most important aspect to me on a Z7 is actually buffer empty time. On the Sony G Series cards, that is running a little less than 5 seconds for me. On my slowest older cards, that is running about 10 seconds. Why is this important? Because you generally don't want to start a new sequence with a less than empty buffer.

One thing I note that isn't getting a lot of attention: the D5 with it's USB 3.0 connection is delivering images to your computer at almost twice the speed that the USB 3.0 Z7 provides (neither are particularly fast as a card reader, though). This is yet another example of the camera makers only giving lip service to current state-of-the-art computing standards. They just don't want to use the more expensive supporting parts necessary to provide full capability, and/or are introducing more overhead in the serial transfer chain. In theory, we should get close to 400MBs transfer speeds out of USB 3.0 (as we do from some card readers). Don't use your Z7 as a card reader other than last resort.

text and images © 2018 Thom Hogan
portions Copyright 1999-2017 Thom Hogan-- All Rights Reserved
Follow us on Twitter: @bythom, hashtags #bythom, #sansmirror