I just ran it again three times with all settings turned off and just the body cap on the camera shooting RAW - 23 frames on the 1000x Lexar, 35 frames on the 1067x Sandisk. This is the number of frames on the 5DIII before it begins slowing from the max frame rate. Yes, you can keep shooting at a reduced frame rate and eventually it will slow even further, but I'm measuring (and interested in) the number of shots at full frame rate.
Also, given these results, I can confirm that aberration correction, vignetting removal, and standard levels of high ISO noise reduction don't impact frame rate.
mack..I didn't leave the body cap on, I shot at a a fixed subject - large RAW - I'll also try it again and check back. do you have image quality set at "10"
edit...I wonder if this is the reason for the diff....from Adorama website:
Lexar Pro 1000x 16GB UDMA 7 CF Specifications
Read Speed 1000x (150MB/s)
Write Speed Max 966x (145MB/s)
Compatibility UDMA CF and CF devices
this would mean that there are two different lexar 1000x cards, both at 150 read, but with 95 and 145 write.
I started a thread about the Lexars as when I bought the ones I currently use they were advertised as write 145 MB/s write but if you look at them online now they say the 1000X is now only 95MB/s write, and if you want the 145 or 150MB/s write you need to buy the newer 1066X.
This only happened quite recently, so I wonder whether people that bought the 1000Xs some time ago have got 145MB/s write cards but those that bought them much more recently only have 95MB/s.
I just ordered a new 64GB card and nearly bought the 1000X, it was only when I noticed the 1066X that I noticed the difference, so I went for the latter.
It is very misleading by Lexar, because I - and many others here and elsewhere - have recommended the 1000X, but that was when they were 145MB/s write.