I agree very informative being said that I still like using the M 2 . 

Great review!
I would be curious to see where you DO see a difference.  3d rendering, or video editing/compiling?  I know a lot of that is CPU driven, but I am curious where it will come into play for non-gaming for a home user, even a power user.  Especially things like where does a M.2 bypass a RAM drive, and vice versa.  Not really *needing* the info, I just find these kinds of tests interesting.

EVGATech_DaveB
Great review!
I would be curious to see where you DO see a difference.  3d rendering, or video editing/compiling?  I know a lot of that is CPU driven, but I am curious where it will come into play for non-gaming for a home user, even a power user.  Especially things like where does a M.2 bypass a RAM drive, and vice versa.  Not really *needing* the info, I just find these kinds of tests interesting.

Thank you!
 
For my usage it's very obvious that file transfer speeds between my RAID and the NVME are much better than with my old SSD. So anything that can make use of that would benefit.
 
I know, from reviews, that large databases can greatly benefit from the massive IOPS gains with higher queue depths. I'd hazard a guess that photo/video editing might benefit from an NVME drive as well - yet, as you mentioned, they'd probably be limited more by the CPU. Always a bottleneck somewhere!

It will be rather long response, so here it goes.
 
First I can confirm that RAPID mode works great. It's not a gimmick, not by a long shot if your software can utilize the speed increase. I mentioned EU4 as one of the games which loads ages without it. Other I can think of is Phoenix Rising mod for old SW FoC which has reworked AI scripting and it's damn huge. Also when copying data between NVMe and RAPID enabled drive performance gains are obvious and clear as glass. One of my photo archives in on RAPID drive. There is about 600 directories with I don't even pretend to know how many files. Ha, ha! Hashnig contents takes fraction of the time when compared to non-RAPID 850EVO (I have quite a few 850s to compare).
 
As for NVMe. Certainly VM, but not many gamers use them and I'm doing my best to move away from them as well. Also DB. I have still one old program which rely on off-line db installed locally. Program is no longer updated since beginning of this year, but data is still relevant for fast search of not too new items and DB are still released monthly. Installing them on NVMe is I don't know, 15 second job. 1/4 of normal SSD and on HDD it took 5 minutes easily.
 
So... if you're artist or into photography (even if that only means hoarding pictures) NVMe drive is smooth sailing. I have 2 large directories with thousands and thousands of files. If not for FastPictureViewer software I would have to chop and chop directories. Here I know I can just dump 100 or 200k files and it won't break a sweat. Saving into such directories is much more pleasant experience than even the fastest of AHCI devices. NVMe has no problem reading 100k files and refreshing then almost instantly. SATA SSD choke itself to death and when I was running HDDs for that part of my archive, suffice to say hard drive was close to exploding while waiting to access the data.
 
Recording video to NVMe is just fabulous. You can run few things in the background and record 1440p material with perfect results. Everything is in perfect sync with 90% quality setting and H.264 codec. Ripping DVDs is trivial. Looking back at recording to MPG1 or DiVX3 15 years ago and being always annoyed because something was not right, NVMe is dream come true (given of course powerful CPU which 2011-3 is, but I had some recording done on 1366 and it went equally well, even with reduced transfers). Same goes for recording Audio, but of course here demand for CPU power is not as excessive.
 
And last but not least anything that involves sustained transfers. I'm working diligently to remove HDD ballast from my main rig. If all goes well before end of the year SSD pool for all stuff should be completed. All I need is ~3-4TB more of SSD space (with 1TB safety margin to offload data if something is not right). Here ordinary AHCI SSD caching drive is not ideal since we deal with sustained R/W/balancing transfers of much higher rate than with HDDs. I preparing one of i750 to do just that. Will plug it into  Gen.2 slot as 1600MB/s bandwidth will be enough to deal with 500ish MB/s SSDs. Copying data between NVMe devices is just blink of the eye. Recently did that when transferring stuff from i750 to 950P thinking it's year 2000, still with HDDs in mind. Fired up TotalCommander, selected directories, F5 and went for a tea. Returned maybe 3 minutes later and job was about 30 sec from finishing. It was just like that: jaw on the floor - how the hell it's over already!?! Exactly. Sustained transfer is just [I have to self censor myself and constrain enthusiasm] unbelievably great! Still sustained transfer is not common thing when you deal with gaming, but when moving large amount of stuff at once you will be in awe. Shock and awe*, just without huge explosions! LOL
 
* - military term for guys not familiar with the phrase.

Some folks were over on HardOCP were wondering if just using Steam Mover might have bottlenecked the game load times as Steam would go through the original drive to find out where the files/folders have moved to.
 
Well, I redid the one test that showed the most variation:
 
 
Fallout 4 1080P (no mods)
Used a save game from one hour thirty minutes into the game. (1920X1080 and all game settings manually set to max)
Timer started when press Enter to Confirm (load) was pressed and stopped when the game loaded in.
 
Tested with steam install + FO4 moved (no other common folder steam games). Verified that other games could not be opened by trying to open one within Steam.
 
Order of testing: NVME, SSD, RAM, RAID. RAM drive only enabled for RAM drive test.

8 Hard Disk RAID 6 Array (WD Black 5TB 128MB Cache 7200RPM drives) (18.1TB of 27.2TB free)
30.25
23.34
21.74
 
512GB 950 Pro NVME in PCIE 3.0 X4 slot (with passive heatsink) (boot drive, ~135GB free)
7.54
7.26
6.9
 
512GB 850 Evo (empty except for Steam + FO4)
7.33
6.80
6.95

48GB RAM drive (3200MHz DDR4 16-18-18-32)
7.06
6.76
6.89
 
 
 
FOR REFERENCE HERE IS THE ORIGINAL TEST
Fallout 4 1080P (no mods)
Used a save game from one hour thirty minutes into the game. (1920X1080 and all game settings manually set to max)
Timer started when press Enter to Confirm (load) was pressed and stopped when the game loaded in.

8 Hard Disk RAID 6 Array (WD Black 5TB 128MB Cache 7200RPM drives) (18.1TB of 27.2TB free)
29.11
22.61
22.91

512GB 950 Pro NVME in PCIE 3.0 X4 slot (with passive heatsink) (boot drive, ~75GB free)
7.65
7.01
6.68

512GB 850 Evo (empty except for FO4)
7.49
6.81
6.67

42GB RAM drive (3200MHz DDR4 16-18-18-32)
6.9
6.54
6.63
 
 
It doesn't seem to have changed the results at all, wouldn't you say?

Interesting tests.
 
The RAM drive test, was that loading it into RAM drive (from a HD) then opening the game?
 
Or were you setup to automatically populate the RAM drive at boot, and it took that long for the game to open from within the RAM drive (memory) ?
 
If the RAM drive is that slow, the CPU / or coding of the game, must be limiting the game opening

Cool GTX
Interesting tests.
 
The RAM drive test, was that loading it into RAM drive (from a HD) then opening the game?
 
Or were you setup to automatically populate the RAM drive at boot, and it took that long for the game to open from within the RAM drive (memory) ?
 
If the RAM drive is that slow, the CPU / or coding of the game, must be limiting the game opening

The RAM drive was created solely for the test, and the games (and for the one test, Steam plus the game) were moved onto the RAM drive and opened from the RAM drive. I was checking save game load speeds only, and not "open the exe, wait for cut-scenes, etc.")
 
I'm not sure what the bottleneck is exactly but these tests show that after seek times are eliminated, game load times plateau - regardless of how fast a drive is.

Thanks for the clarity and sharing your results with the Forum community. 
 
Glad I did not buy the 64 GB RAM kit to make a RAM drive; though it still might help with large CAD files

Are all your testing platforms connected together at the same time or are you separately connecting them to the MB for testing purposes?

GTXJackBauer
Are all your testing platforms connected together at the same time or are you separately connecting them to the MB for testing purposes?

Everything is connected, with the exception of the RAM drive that is only turned on during the RAM drive test.
 
I completed read/write tests, using ATTO disk benchmark, on all drives and they are all performing as expected so there isn't a bottleneck there.