Lexar CF 800x vs 1066x

Rob Galbraith has published results of 1000X memory and 600X memory in the 1D4...
There is a 20% improvement in writing RAW data to the 1000X, compared to the 600X.

The data in the table was derived by timing how long it took the Canon EOS-1D Mark IV to write 25 RAW CR2 photos to the memory card. Timing commenced when the camera's buffer status light illuminated, and stopped when the light went out.

The figure in the Burst column represents how many CR2 + JPEG pictures the camera was able to take in 30 seconds. Higher numbers are better, and a relatively small difference in the number of frames captured translates to a noticeable difference in how much faster one card feels than another, when shooting frequent extended bursts. Burst count increased by 13%.

SilentRampage34 wrote in post #17504761
Per Lexar's website.

Granted, the VPG spec as defined by the Compact Flash Association is intended to define the minimum write speed for sustained video capture, it stands to reason that if they card can continuously write video at a given speed, it ought to be able to write photos or other files at that speed. And to be clear, I never claimed the 1066x could write at 160MB/s, only that Lexar claims it meets the requirement of VPG-65, which states it can write video files at 65MB/s. I thought I was very clear that the 160MB/s was in reference to read speed as a maximum, which doesn't matter much beyond about 110MB/s unless you're reading from the card and writing to an SSD.

It is a incorrectly assumption that if a card can do 65MB/s with continuous data, it will do the same for other files. In fact, it is direrctly the opposite. Memory card, such as CF, SD, USB are optimized for large data. They are the most efficient when it come to continuous data. This is even true for SDD but to a lesser extended. If you were to continously write 4KB files onto this CF, you will probably get 5-10MB/s on this card if you are lucky. You can try it with Cystal Disk Mark measurement on a USB3.0 card reader.

You can look into VPG-65 spec if you have it availble. It defined a specific data pattern. It is not entirely a continuously data. Because if it is, it will do 160MB/s. It is a sequential data stream of files of certain minimal size, If you write that pattern, you will get 65MB/s. Similar spec is also defined for SD speed class as well.

What you said is correct tho, there almost no point of going beyong 110MB if you are using a HDD on the other end. But these days, who doesn't have a SSD.

I just looked up VPG data pattern size. It is defined as a random stream of 2MB sequential data. Meaning each 2MB is pure sequential with no breakage. Then from one 2MB to the next 2MB can be random. I just did CDM measurement on a both Lexar 1066 and Sandisk Extreme pro. Results are similar to each other with in 5%. The slower than spec sequential write is probably due USB3.0 card reader's overhead. Nonethless, it shows data pattern and size has everything to do with write performance.

Pure sequential read: 151MB/s
Pure sequential write : 130MB/s
512K read: 145MB/s
512K write: 40.45MB/s
4K read: 9MB/s
4K write: 2.4MB/s

eelnoraa wrote in post #17508061
It is a incorrectly assumption that if a card can do 65MB/s with continuous data, it will do the same for other files. In fact, it is direrctly the opposite. Memory card, such as CF, SD, USB are optimized for large data. They are the most efficient when it come to continuous data. This is even true for SDD but to a lesser extended. If you were to continously write 4KB files onto this CF, you will probably get 5-10MB/s on this card if you are lucky.

Surely writing digital image files (at 4 to 30MB a go) is going to be much closer to sequential speeds than 4kB speed. Especially when you consider that fast card users will be more likely to shoot in raw, so will be towards the top end of that range.

eelnoraa wrote in post #17508061
It is a incorrectly assumption that if a card can do 65MB/s with continuous data, it will do the same for other files. In fact, it is direrctly the opposite. Memory card, such as CF, SD, USB are optimized for large data. They are the most efficient when it come to continuous data. This is even true for SDD but to a lesser extended. If you were to continously write 4KB files onto this CF, you will probably get 5-10MB/s on this card if you are lucky. You can try it with Cystal Disk Mark measurement on a USB3.0 card reader.
.

The Rob Galbraith data in the table was derived by timing how long it took the Canon EOS-1D Mark IV to write 25 Large (Quality 8) JPEG and then 25 RAW CR2 photos to the memory card. This is not 'continuous' data, but data in 25 chunks! Nevertheless, 25 shots of RAW data could achieve 66MB/sec, and 25 shots of JPG could achieve 46MB/sec.