My 840 EVO looks faster than my 950 Pro- what have I got wrong?

These are the Crystal Diskmarks comparing my Samsung 840 EVO boot drive and a newly formatted 950 Pro 512 GB . The 840 looks to me like it's performing much better than the 950, but I may be misinterpreting the numbers or using the wrong test setup. Are the results about what's to be expected or am I missing something?
The 840 has the OS and program files. The 950, the F drive, has no data on it yet. The processor is an Intel 5930K at stock 3.5 GHz on an Asus X99-E WS motherboard with 32GB ddr4 RAM. OS is Win7 Pro.

Here they are. C is the 840, E is the 950. Any help interpreting them would be very much appreciated.
I see the rate of the 950 starts going up at the 64 Mb transfer size point. Should I up that upper limit to get a better comparison? To be honest, I don't know how that influences the result. I have a lot to learn.

Yes, that's exactly the Samsung 950 Pro NVME drive I have, but I'll have to open the case and check the manual to verify the 840 EVO is on the Intel controller. I'll check the drivers too. Not sure what you meant earlier by changing the benchmark, but I wouldn't know how to anyway.

In order to compare my disc speeds to a known source, I went to the Crystal Diskmark website* and clipped out their Intel 750 example test result and pasted it into a Photoshop image. Then I ran the same benchmark on each of my drives- same queues, same threads- and after each test, I used a snipping tool to capture the result and pasted it side-by-side below 750 test for comparison.


Before anyone suggests that these results could be faked, let me just say I wouldn't disrespect the forum by doing such a thing, even if I knew how. It's just cut and paste; no manipulation.

As you can see, the 950 Pro pretty much held its own against the Intel 750, but the 840 EVO seemingly, (and I stress seemingly) way outperformed it — by it is much as 20 times in the case of 4K reads. I'm under no illusion that I this is the fastest SSD in the universe, but either a driver or controller, or maybe quantum foam, is messing with me. Anyone have a recommendation?


* Overview

Are you testing the 2 disks on the same port ? I recently changed my 2600K system to SSD and was disappointed with the speed tests ( < 50% expected ). I had the SSD plugged in to a Marvell port on my Z68A-GD80 mobo, so I tried an Intel port and the speed went up to the correct figure.