#1102
Yes, DP 1.3 will support 120Hz at 4K.
I'm not sure what you'd want a 4K monitor for anyway. For the resolution to be noticable on a 24" (27") monitor you need to be closer than 3' (3.5'), for the full benefit you need to be closer than 1.5' (1.75'). That's awfully close. Of course there's bigger screens but I never understood why you'd use those as monitors.

I'd rather just get an M.2 SSD, Samsung 950 Pro probably.

What are "extended motherboard brands"?

#1103
NVMe needs OS support as well, so Windows 8.1 or 10.

Right now R9 Fury X, R9 Fury, R9 390X, R9 390.
I thought you wanted to wait for the new ones with DP 1.3 though. They aren't released yet, so they aren't named either.

New AMD cards are expected mid 2016 and they should include DP 1.3.
nVidia would be late 2016 (Q4) probably if at all and there's no word on DP 1.3 so probably no on that too.
Why do you insist on nVidia?

#1096
nVidia won't release any "goodies" in April, because they don't have a working GPU yet.
http://semiaccurate.com/2016/02/01/news-of-nvidias-pascal-tapeout-and-silicon-is-important/
While AMD showed a worked GPU, including drivers and everything, nVidia CEO Jen-Hsun Huang waved around a GPU, but didn't actually show it working, because it's obviously amazing and definitely not a year old Kepler GPU like the date code on the chip and the GDDR5 memory around it suggest.
http://semiaccurate.com/2016/01/11/nvidia-pascal-over-a-year-ahead-of-1416nm-competition/

#1093
Only Z chipsets support overclocking.

There was a bug that allowed overclocking on other 8 series chipsets. Intel didn't like that. The next microcode/firmware update in July 2013, a few weeks after motherboard manufacturers had figured that out, "fixed" it. I'm assuming MSI chose to remove the options that wouldn't do anything anymore. Reverting to a BIOS version before (in your case that's only the first (C.0)) that should get you the options back. Whether it'll actually reenable overclocking depends on how Intel blocked it and I'm not sure about that. Also not only using a very old BIOS, but the very first version for a particular motherboard is usually a bad idea.

In any case, if you can afford a -K i7 and a 100€ B85 mobo, but not the 110€ Z87 version of that mobo or the 95€ Z87 version without the glorious Qualcomm Killer E2205 Ethernet swag then something is probably wrong.

#1094
4790K and an overpriced mobo but a budget cooler.
3x4GB RAM which is slower than both 2x4 and 2x8 or 4x4 because it ruins dual channel.
970 whose only advantage over the 390 is efficiency but then you get a 750W PSU for a 300W build anyway.
WHY?!

Just remember, no matter who wins, half the US population will be upset now, the other in a few years.

I like how everyone is ignoring that
1. Most of the time is actually lost because of the routers and other stuff on the way, the packets barely spend half the time actually traveling through fiber and the rest in between.
2. No matter whether the last mile is fiber or DSL anything traveling from Texas to Europe will spend 80% of the way on the same couple of routes, all with similar latency.

Anyway, as much as I'm in favour of derailing a thread when I'm doing it, let's get back to OP's questions.
Short answers:
1. No.
2: Stuff™.

Longer answers:
1. It shouldn't (see #2), but your type of connection and the ISP can.
2. I'm assuming you're asking because you want lower ping, but I'd really need more information to answer this properly.

EDIT: #15
What is the average bandwidth of a mammoth? Are we talking American/Imperial or European/Metric mammoths?

#12
This isn't about performance, TF2 won't suffer from that, rather the mouse will suffer because of TF2.
Your CPU can handle doing this once per millisecond without any problems, but it won't be after exactly a millisecond. Sometimes it'll be 0.9ms, sometimes 1.1ms. This gets worse with higher load (higher polling rate). Here's some pictures.

At 125Hz

http://i.imgur.com/saTBBzh.png

The worst outliers are less than +-0.07ms

At 1000Hz

http://i.imgur.com/xtQFNBO.png

Ignoring the absolute worst at +-0.3ms we're talking about double at +-0.14ms.

Now take a guess what happens at 2000Hz or even 4000Hz. Let it sink in a bit. Updates every 0.25ms plus minus 0.3ms. Yeah, sounds great, right? Now think about what happens when there's actual load on the CPU, you know something that absolutely loves using the CPU. Does anything come to mind? I heard there's a game called TF2 ...

EDIT: #7 ninja'd me

#5
Were you leaning on F5?

Anyway what you get (if your mouse even suppports >1000Hz) is higher CPU load and therefore lower polling precision. Oh wait you were asking for what this improves. Uhm, the number is higher?

Overclocking a mouse usually overclocks the mouse. You do this to get an overclocked mouse.

#1091
-Haswell instead of Skylake
-4690K, but no cooler and a H97 mobo
-BX100 instead of 850 Evo
-960 instead of 380/285
-somehow fries the mobo
Well at least you got 16GB RAM, right?

smh

#1089
Speedfan is definitely incorrect. Unless you are outdoors -8°C isn't legit.
HWMonitor should show some more information like min/max temp. It shows fan speeds as well.

B75 does not support overclocking.
It's not happening.

#10
Sounds like a non-Z mobo.
Model?

#1087
Tbh you can get away with anything below 90 so I don't see the problem. 70-80 definitely isn't overheating.

Other than that:
-Check for thermal throttling to be sure. If there is none, it's not a problem.
-Check the fan rpm. It might to be spinning up all the way to the max. Check the case fans as well.
-Did you apply the thermal paste properly after cleaning both the CPU and cooler and not just put another huge glob on it?
-Is the cooler mounted correctly?