Ok, I see two possibilities:
1. Overclocking doesn't work like you think it does or
2. Your monitor runs on magic.

You set it to a higher refresh rate and see what happens. That's overclocking. If you meant you can't find the official maximum supported overclocked refresh rate, well that's because it's not supported. There are no specs for running it out of spec which is what overclocking is about. If you meant you set it to higher refresh rate and it worked then congratulations. If you want to go higher then do so, otherwise don't. I can guarantee that unless the monitor runs on unicorn tears it will stop working at some point, you just have to go high enough.

#1132
It's "only" 100Hz but if you don't mind the bezel down the middle it would definitely work. Fov would be 148.11° instead of the 151.93° on 3 16:9 monitors, so almost no difference.

#1129
Fov depends on the game and vertical fov. The formula should be similar for most games, at least all quake based engines (including the source engine).
The source engine specifically uses 4:3 as the standard. So fov_desired 90 on 4:3 gets you 90° horizontal fov. The vertical fov is 73.74° and stays constant unless you change fov_desired, e.g. for fov_desired 75 it would be 59.84°.
The source engine formula is: hfov = 2*atan( tan(fov_desired / 2) * aspect ratio * 0.75 )
Instead of *0.75 (which is the same as / (4/3)) divide by whatever an engine's standard aspect ratio is, if it's not 4:3.
For fov_desired 90 it becomes even shorter: hfov = 2*atan( aspect ratio * 0.75 )

There's calculators for it.
http://www.casualhacks.net/Source-FOV-calculator.html
http://www.wsgf.org/fovcalc.php

No, stacking monitors just reduces your aspect ratio again. Vertical fov stays constant.

#1127
No, you misunderstood again, it's not a platform, it is the actual SSD. Here's a picture of one disassembled:

http://www.pcper.com/files/imagecache/article_max_width/review/2015-04-02/150330-181243_DxO.jpg

The point is that they are using the same connection, just a different form factor. Your own picture actually shows that.

http://i.imgur.com/E1QSuuS.jpg

It's like USB and micro USB. Literally the same, just smaller.

3x2 1440p ultrawide with 4 GPUs maybe 60fps. In theory playable but you'll have the same input lag you'd have with 15fps, so probably not exactly enjoyable.

#1125
1. No, it's about the aspect ratio. fov_desired 90 gets you 90° horizontal fov on a 4:3 monitor, no matter the resolution, but 106.26° on a 16:9 monitor.
2. Yes. Higher aspect ratio = higher fov.
3. fps same as 3 4K monitors (so 60fps maybe, but not 120). fov would be 126.87° because the aspect ratio is 48:18, higher than 16:9 but lower than your current 48:9.
4. 1080p,1440p,2160p are resolutions, not pixel densities. Neither affect aspect ratio. Aspect ratio is the ratio of your monitor's width to your monitor's height.
5. Size also doesn't affect the aspect ratio/fov. On a 10" 480p 16:9 monitor you see exactly the same as on a 27" 4K 16:9 monitor.
6. Afaik it's working on AMD.
7. M.2 is just a form factor. It still uses PCIe. So the 950 Pro uses NVMe via PCIe and the Intel 750 uses NVMe via PCIe, they're just different sizes and use different connectors.
I said RAID0 would be fast, but keep in mind that losing one SSD means you lose all data. As always you should have backups for that case. Do you have backups? I think I asked that earlier when you mentioned 5 year old HDDs.
8. Form factor is irrelevant for RAID. Yes, it will work.
I'm pretty sure 2 M.2 slots are enough for 2 M.2 SSDs.

#1121
Yep, same as G-Sync, except cheaper and Intel will support it and nVidia already uses it in some notebooks (because it's cheaper) while not supporting it officially so they can keep selling G-Sync modules.

In Source games with 3 16:9 monitors (=48:9 total) your fov with fov_desired should be 151.93° (instead of the normal 106.26° with 16:9). 5 monitors (=80:9) would get you 162.94° so it's really not worth it.
5 monitors in portrait mode result in an aspect ratio of 45:16 which is lower than even just 3 monitors in landscape mode. Fov would be 129.27°, the only thing you can is a higher vertical resolution, but you can get that with an ultrawide 4K monitor (slightly lower aspect ratio) or two 16:9 4K monitors next to each other (higher AR) way easier.

Two 512GB SSDs in RAID0 would be twice as fast. There really shouldn't be any relevant quality difference. Samsung promised a 1TB model "next year" back in September 2015 so it should be coming soon-ish.

Ask MSI.

#1122
New CPUs around the same time. Patience.

#1117
Right now most games still use AFR (alternate frame rendering) for multi-GPU setups. The reason is that you don't have to program anything for it. GPU 1 renders frame 1, GPU 2 renders frame 2 (starts rendering when frame 1 is halfway finished), then GPU 1 renders frame 3, GPU 2 renders frame 4 and so on. That's why the VRAM doesn't double. Also each frame still takes the same time to render so even though you get double the frames your input lag stays the same.

With DX12 and Mantle (soon Vulkan) we should see more SFR (split frame rendering) again. One GPU renders e.g. the top half, the other the bottom half (or if there's really good support for multi monitor setups we might see a "one card per monitor" style). Some things like textures still have to be present in both GPUs' memory but other than that your available VRAM doubles. Also the frametimes halve and your input lag is finally lowered compared to a single GPU.

About CPU/GPU relation: Yep, now you got it.

About the monitors: Do you want Freesync? Also there might be some new monitors by the time the new GPUs get released.

A 3x3 monitor setup would just be like a huge monitor with the same aspect ratio but 9 times the resolution (so halfway beween 4K and 8K). A 3x2 setup would be just like an ultrawide 4K monitor. That's all with 1080p monitors of course. Anything higher than that and there's no way to get playable fps. You lose the only advantage of a surround setup, the higher fov. Unless you were to use a single row but then you'd have to turn your head to see all monitors and you probably don't have enough space to do it, so I don't see the point.

What do you mean "how exactly does the samsung one work?"? What do you want to know?

#1118
Fury X or wait. I'd say wait.

#1119
pcpartpicker does have an option for that to filter anything with less than x M.2 slots. Why do you need multiple M.2 slots though?
There's no point in deciding on a specific mobo already, there should be new mobos by the time the new GPUs are released.

#1115
No. First of all it's not storage, it's not an HDD. Secondly that is just specifically the Fury X vs the 980 Ti. Thirdly I'm not sure how you got to that conclusion:

CAD is all about driver support. http://www.tomshardware.com/reviews/best-workstation-graphics-card,3493.html
Just take a quick look at it. Compare Maya 2013 and OpenCL Video Processing. You'll see what I mean.

VRAM does not stack. You would not have 8+GB with a single nVidia card unless you were to get a Titan X. The 980 Ti got "only" 6GB. All cards need all the information so everything is just doubled. That means right now you got 3GB for 3 1080p monitors. With 6GB and 3 1440p monitors nothing would change, the VRAM per pixel stays the same. 4GB is a bit tight. With 4K which doubles the number of pixels again VRAM is lacking on both. I'm not a fan of overpaying for a Titan X just to get 12GB VRAM so I'd wait. With the next generation of high end GPUs at 8 or even 16GB VRAM there should be absolutely no problems with triple 1440p, although triple 4K might still be a bit tight on 8GB.
Not sure about your distinction between "speeds", "processing" and "rendering power". Either way the new stuff will get you both, enough VRAM even on highest settings for triple 1440p, maybe triple 4K and more speed/power to actually get 120fps (at least on 1440p).

Yes, the memory bus is how much data you can move per clock cycle, no the AMD R9 Fury X is the fastest in that regard at the moment.
The equation is simple: 1/7 the clock speed times 10.66 the bus width equals ~1.5 times the bandwidth of GDDR5. Or in absolute numbers:
980 Ti / Titan X: 3500MHz x 384bit x 2* = 2688Gbit/s = 336GByte/s
Fury / Fury X: 500MHz x 4096bit x 2 = 4096Gbit/s = 512GByte/s
*DDR stands for double data rate, it moves twice the data you'd expect per clock cycle
I think now you see the issue. The Fury X got almost twice the bandwidth of your current 780s, but not twice the VRAM, which is what I complained about. Again, the next gen should take care of that, 2 or 4 times the VRAM and the same or twice the bandwidth of the Fury.
So the "best case" for the new cards would be almost 4 times the bandwidth of your 780s, more than 4 times the VRAM, which fits perfectly with 4K being 4 times 1080p.

Now I can finally elaborate on the 3 reasons why I keep trying to convince you to go for 1440p instead:
1. Even though bandwidth and VRAM might be enough you'll only get a bit more than twice the performance. 1440p are twice the number of pixels of 1080p, add that new games are more demanding and you end up with pretty much the same performance. On 4K your fps would halve unless you get more than 2 GPUs, which I wouldn't recommend.
2. You're close enough to your monitors to see the difference between 1080p and 1440p, but not close enough to notice the difference between 1440p and 4K.
3. 4K 120Hz monitors don't exist yet and even if the become a thing they'll probably be incredibly expensive at first.

Yes, the new stuff will be a large milestone. nVidia does have a press conference but they don't have a GPU to release yet, I linked that before. http://semiaccurate.com/2016/02/01/news-of-nvidias-pascal-tapeout-and-silicon-is-important/
AMD is going for a mid 2016 release and is expected to release their new GPUs before nVidia does.

As for SSDs I'd consider the Samsung 950 Pro because it doesn't need a PCIe slot, it uses an M.2 slot instead, is quite fast* and cheaper than the Intel 750.
*http://www.storagereview.com/samsung_950_pro_m2_ssd_review
http://www.storagereview.com/intel_ssd_750_review

As for CPUs, yes you can get away with a normal i7 or even an i5 (although that's pushing it) if you stick with 2 GPUs. For 3 or 4 GPUs you need an i7 Extreme because only they support it. Mobo choice obviously depends on the CPU.
New CPUs, Intel both the standard desktop quad cores including i7s and i5s (although it'll only be a refresh (Kaby Lake), same architecture as the current ones (Skylake (e.g. i5-6600K and i7-6700K)) only with slightly higher clockrates and/or at slightly lower prices) and i7 Extreme (6, 8 and maybe 10 cores), and maybe AMD Zen (GET HYPE) should be released around the same time as the GPUs, the standard back to school sale window.

Wait until the GPUs are released, then you'll see if you want/need more than 3 GPUs and with that information you can decide which CPU to get. I'll probably still be here in a months so just ask again when that time comes around.

#1113
Yes, DDR4 is out now.

I don't know why HDDs that were never wiped would be an issue, but you should always have backups, no matter how new an HDD is. 5-6 years is about the average life expectancy so if you don't have any backups yet you should change that asap.

Fury or Fury X? The Fury X should be faster than the 980 Ti now after the driver updates and it's cheaper, not a hard decision. In games there's no such thing as "superior rendering", in CAD there might be differences due to the drivers. While AMD cards usually have more raw power (as in number crunching performance) and recent nVidia cards have been a bit more efficient (that might change with the next generation of GPUs) neither of it actually matters to you, because you shouldn't worry about how the GPU gets the fps, just how many. And efficiency is hardly a concern when you're doing a 1000W build.

The memory on the Fury X deserves a bit of elaboration though. It's HBM (High Bandwidth Memory) and it does exactly what the name suggests. The basic idea is that running GDDR5 at 3000+MHz is a pain in the ass (high speed is always a pain in the ass) and takes a lot of power. Due to that the Titan X is sometimes slower than the 980 Ti because even though it is the faster card the 6GB more VRAM take so much out of the power budget that it has to hold back. Going from a 384bit bus (780 ti/980 Ti) to a 512bit bus (290(X)/390(X)) with the memory at lower speed costs die space and therefore reduces the space you can use for stuff that does actual work so it's not ideal either.
The "obvious" solution: Go crazy on the bus width. 256bit per chip. The Fury (X) got 16 that means a 4096bit bus. So even with the memory running at only 500MHz (1/7 of the 980 Ti's 3500MHz), which means extremely low power consumption, having more than 10 times the bus width gets high you 1.5 times the bandwidth. The problem is routing a 4096bit bus. You can't get that many traces that close together on a normal PCB so you have to use an interposer. Those are very expensive so to keep them as small as possible you have to keep all the memory chips close to the GPU. The solution: Stack 4 of them on top of each other. Now the only problem left is that since HBM is a new technology and the chips are smaller than GDDR5 chips you can only get 0.25GB (=2Gb) per chip which yields a total of 4GB (4 stacks is the max before the interposer becomes too big, 4 high).
Due to that the Fury X is a bit of a weird beast. It got the power and memory bandwidth for 1440p/4K but only 4GB VRAM.

That brings me to the conclusion: Wait a bit.
New AMD GPUs should be released mid 2016. They should fix the only problems the Fury X has. They'll be more efficient (and obviously faster) and they'll get HBM2. 1GB (=8Gb) max per chip so a total of 16GB if AMD wanted to (although 8GB is more reasonable and probable) and 1000 instead 500MHz. For 4 stacks that means 1024GB/s, triple the bandwidth of the 980 Ti and obviously double of the Fury X.

There'll also be Zen later which might finally be some competition for Intel again and there'll definitely be Broadwell-E (10 cores rumoured, but at least 8 cores). There'll also be a refresh of Skylake named Kaby Lake.

If nothing else waiting for the AMD GPUs with DP 1.3 might tell you whether or not there'll be any 4K 120Hz monitors at all or even some that are worth buying (though I still think that 1440p 120Hz IPS monitors would be a better idea).

Ah yes the true professional approach to diagnosing performance problems:
"Why run a benchmark, I can see the fps are lower than the should be. Well at least lower than I've heard other people say the should be. They said other people with better hardware are getting more fps."

Simply benchmarking and then comparing the result to the results of similar hardware with similar settings would've told you right off the bat that nothing is wrong and you just have to use even lower settings.

#1111
I checked your profile and now this finally makes sense. You're reusing old parts.

I don't expect any 4K 120Hz monitors before DP 1.3 and even then there's basically no market for since we don't have the hardware yet to get 120fps on 4K.

Well you can't boot from an NVMe SSD until you get the drivers so if you want to use it as boot SSD things get interesting.

The viewing distance where you can take full advantage of 1440p is about 2.34' for 24" and 2.63' for 27" monitors. That's pretty much exactly what you're looking at so no need for 4K. I'd say 1440p >120Hz IPS panels would be a nice upgrade. You'll still need double the GPU power just to keep your fps at the same level but that's actually doable.

What are your current specs?

#1108
https://en.wikipedia.org/wiki/NVM_Express
You can get it to work if you add the drivers before installing but honestly when you're doing a fresh install why not move to a newer OS?

I guess now is the time for me to crush your dreams.
4K or 120Hz. Choose one. Apart from the fact that there are no 4K 120Hz monitors yet, current and even next gen GPUs are going to strugge to push 60fps on high settings on a single 4K monitor. If you were to get 3 GPUs and 3 monitors the problem stays the same, with only 2 GPUs it's even worse.
There's no point in buying all this stuff now if
a) You can't get 120fps in triple 4K unless you drop to pretty low settings and
b) You can't get 120Hz 4K monitors so you might as well wait until those are available and get more powerful GPUs for less money once that happens.

If those 4-5K are including the monitors I'm pretty sure the budget won't cut it either. 120Hz 1440p monitors are >500$, 4K will be more expensive than that. GPUs I'm guessing you're going for flagships or slightly below. Add waterblocks and you're already easily over budget without any other hardware. For 2 cards you could get away with LGA1151 (barely) but who are we kidding? You'll want LGA2011-3 with those nice 40/32 PCIe lanes. Check the specs there should be something like "PCIe configurations: 16/16/8, 16/8/8/8, 8/8/8/8/8". For 3 GPUs you'll want at least 3 entries with 8 or higher in one of those lists. Like I said 2 shouldn't be a problem, most full ATX Z170 boards can do 8/8.

What do you mean you already got most of the watercooling parts? You can't even guess if radiators can handle the wattage if you don't even know what parts you'll get. You don't know if the pump(s) can handle the blocks if you don't know which and how many blocks you'll get. You can't know what blocks you'll get if you don't know which parts you'll get.

This all seems incredibly rushed.

#1109
No ninjas please.

Sorry, but I have to do this:

No, you don't. There is no i5-2600K.
It could be an i7-2600K or an i5-2500K.

You could run the benchmark demo to see if your performance is in the right ballpark.

#1105
No, not quite. There is no reason to buy a PCIe SSD that doesn't use NVMe (it would be hard to find one anway). Windows 7 does not support NVMe. Therefore either don't get a PCIe (or M.2) SSD or stop using Windows 7 and upgrade to at least Windows 8.1.

At 2-3 foot 1440p is more appropriate, I don't see the need for 4K.

That's just a normal motherboard.
Are you sure about dual GPUs though? A single, more powerful GPU is always better (unless there isn't a more powerful one). What's your budget?

#1
Yes, your cooler should be good enough.
The FX-9590 is basically and overclocked 8350 and it's not a problem to reach the same clockrates with an 8350 and an air cooler (see #4).

Also, never buy an All in One Liquid Cooler.
Never.
They are more expensive and don't perform any better than a high end air cooler. If they do it's because the manufacturer slapped high rpm fans onto them so they sound like a jet engine. If you turn the fans down they perform worse than air coolers, if you don't you could just get an air cooler and high rpm fans and end up with better cooler at lower noise levels anyway.
http://www.relaxedtech.com/reviews/noctua/nh-d15-versus-closed-loop-liquid-coolers/2
http://www.relaxedtech.com/reviews/noctua/nf-a14-industrialppc/7