I have the 240GB version of that drive and I know that I had transfer speeds from that drive over 1.2GB/s and to that drive over 550MB/s.
That is correct. I bought it for a Mac Pro from 2009 but it is currently part of my ryzen system. Difference is I have it slotted into a PCIe card, so the bottleneck is most likely not your drive. I mean it is not that crazy fast like some of todays drives but still more than a gig per second read.
Yup, if he was to put the drive in a pcie to m.2 riser and run it in the second cpu provided pcie slot, he would get full performance. But, the x2 gen 2 connection on his motherboard is most definitely going to limit it for now.
If this helps, I just disabled AHCI mode in my BIOS, rebooted, and lost all my drives except the M.2.
I may reinstall Windows 10 on a Samsung 960 EVO sometime this week, just to see if a NVMe drive makes any difference. If it doesn't, I'll fork over $30 for an adapter card, stick the EVO on that, and turn my GPU down to 8x.
All told, even if I did get what this Kingston M.2 promised (compared to the 750 Reads I'm getting), it probably wouldn't make much of a real-world difference. I was just hoping it was something simple that I'd stupidly overlooked.
Just to be clear, we've come to the conclusion that PCIe 2.0 x2, although claiming 1,250 MB/sec, only really gives you ~750 MB/sec?
Gen 2 x2 only promises 1000MBps theoretical, the drive promises 1290-1400MBps in a by 4. But yes I'd say that's about all there is too say.
Just to put a button on this thread, I went out and purchased a $24 ASUS Hyper M.2 x4 expansion card, and threw a Samsung 960 EVO in my second PCIe slot.
Now I'm getting these speeds: http://i.imgur.com/blWjM8M.png
Yes, it knocks my video card down to 8-lanes, but it only seems to make a difference of <1 FPS in the Heaven benchmark.
Long story short, you don't need to put up with a 2-lane M.2 socket on an older motherboard.
Thanks, everyone, for your advice.
I am not aware of the existence of any video card that can saturate PCIE 3.0 x8, so whatever difference you're seeing is probably margin of error stuff.
