#1222
In terms of GPU it's doable, you'd need something around a Fury X / 980 Ti, maybe a bit better, so that should line up perfectly with the new GPUs (Pascal/Polaris) in the 300-400$ bracket.
I'm really concerned with the CPU requirements for 144fps though. I don't think it's doable, even with overclocking. On that note:
Overclocking yes or no?

Either way I can't make a partlist until all parts are actually available (one more month for Pascal/Polaris) since the prices would change too much.

Does he really need a completely new build though?

#1223
600$ is more than enough.
Just for TF2 you could get away with the iGPU.
PCPartPicker part list / Price breakdown by merchant

CPU: Intel Core i3-6320 3.9GHz Dual-Core Processor ($154.95 @ Amazon)
Motherboard: Gigabyte GA-B150M-D3H Micro ATX LGA1151 Motherboard ($57.98 @ Newegg)
Memory: Kingston FURY 8GB (2 x 4GB) DDR4-2400 Memory ($32.99 @ Amazon)
Storage: Samsung 850 EVO-Series 250GB 2.5" Solid State Drive ($89.17 @ Amazon)
Storage: Seagate Barracuda 1TB 3.5" 7200RPM Internal Hard Drive ($45.89 @ OutletPC)
Case: Cooler Master N200 MicroATX Mid Tower Case ($34.99 @ Micro Center)
Power Supply: EVGA 500W 80+ Bronze Certified ATX Power Supply ($34.98 @ Newegg)
Total: $450.95
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2016-05-09 11:18 EDT-0400

For other games I'd throw in a discrete GPU as well.
Maxing out the budget the best possible would be a 380 (can do 60 fps in Fallout 4 @1080p Ultra). If you absolutely have to keep it under 600$ drop down to the i3-6300 instead of the 6320.
PCPartPicker part list / Price breakdown by merchant

CPU: Intel Core i3-6300 3.8GHz Dual-Core Processor ($135.99 @ NCIX US)
Motherboard: Gigabyte GA-B150M-D3H Micro ATX LGA1151 Motherboard ($57.98 @ Newegg)
Memory: Kingston FURY 8GB (2 x 4GB) DDR4-2400 Memory ($32.99 @ Amazon)
Storage: Samsung 850 EVO-Series 250GB 2.5" Solid State Drive ($89.17 @ Amazon)
Storage: Seagate Barracuda 1TB 3.5" 7200RPM Internal Hard Drive ($45.89 @ OutletPC)
Video Card: PowerColor Radeon R9 380 2GB PCS+ Video Card ($158.98 @ Newegg)
Case: Cooler Master N200 MicroATX Mid Tower Case ($34.99 @ Micro Center)
Power Supply: EVGA 500W 80+ Bronze Certified ATX Power Supply ($34.98 @ Newegg)
Total: $590.97
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2016-05-09 11:22 EDT-0400

You could wait another month for the new GPUs, if nothing else there'll be some price drops, but it won't change that much.

#1224
1. You can't just assume. You'd drop a few hundred Euro on upgrades and just hope you guessed right? Rip wallet.
2. There are no reviews yet and the GTX 1070 isn't available yet. There is no way to know if it would make sense at all. We don't know if the 480X might not be the better option either, since we know nothing about that one too.
3. If this is just about RAM capacity you could just get another 16GB kit of DDR3. Worst case if you need higher speed as well or if all slots are already full you get a 32GB kit. Way cheaper than a new CPU, a new mobo and 32GB RAM.

So basically you should absolutely not do any of what you just said.
First figure out what's actually holding you back. Like I said if you just need more RAM that's a cheap and easy fix.
I also doubt that it's your GPU. It would be nice to know which editing software you use, but even those that do use GPU acceleration are almost never bottlenecked by the GPU. The reason why you'd want GPU acceleration in the first place is because your CPU is a bottleneck (which is perfectly normal for rendering, really).
If it does turn out to be the CPU I need to know your budget. Because going from a Quadcore to another Quadcore (e.g. i7-6700K, maybe i7-7700K if you'd be willing to wait) that's 20% faster at best and maybe 10% higher clocked (I don't know how much you overclocked the 3770K) won't do a whole lot. At least personally I think dropping 500€ for 30% more performance isn't worth it.
Another 100€ would get you a lower clocked 6 core so that's about 30% on top of that. The question is do you think 600€ for 50-70% better performance is worth it?
The higher clocked Intel 6 cores wouldn't really be worth it, since you're paying 100-200€ for just marginally higher clockrates.
Then there's the 8 and soon 10 cores from Intel for 1000€ and 1500€ respectively which would put you somewhere around 1300/1800€ total. Sure, twice your current performance or even better is nice, but if I had to guess I'd say it's not within your budget.

There's one option if you're willing to wait though. With AMD's Zen we should be seeing some much more reasonably priced 6 and 8 cores.

#42
You are forgetting that there are two different Polaris GPUs.
Polaris 11 which would be around a 380X in performance (not exactly entry level, more like decent midrange), but would more importantly be used as a high end notebook GPU* and
Polaris 10 which is a whole different beast, literally twice as large and fast so it might/should beat the Fury X or a 980 Ti.

* slightly lower clockrates of course, but that's still around a 380 in performance so you'd get e.g. 60fps on Ultra in Fallout 4 while using <50W which usually means garbage tier mobile GPUs like the GTX 940M or at best 950M which can barely keep up with the GTX 750.

Yes the GTX 1080 will be the best card in June but it's still retarded to buy it in May if it'll drop 100$ in price 2 weeks later.

No, we haven't heard anything about a 1080 Ti yet.
Vega 10 (faster than the 1080) should happen in Q1 2017, Vega 11 and GP102 (1080 Ti or whatever) should be released sometime in summer 2017. If they can even get GP100 working until then because that one takes priority. If GP100 gets terrible yields they might choose not to build a GP102 until later in 2017 since it would be a disaster at consumer prices and instead sell a further cut down version of the GP100, although that would get shit on by Vega 10, let alone Vega 11.

It's time to readjust the hype.

1. PAPER LAUNCH
No GPUs are actually available.
The GTX 1080 won't launch until May 25th and aftermarket designs will only become available at the same time as the GTX 1070, June 10th.
Why did they push the 1080 before Computex? Why is the reference design 100$ more than aftermarket versions, which should have better coolers? The answer is simple, it's a Titan light. They're milking the "fastest GPU" title again.
There's also another reason for it: They can't sell a whole lot of 1080s because they don't have them. There is simply not enough GDDR5X available yet. For the same reason the 1070 uses only GDDR5. Should enough idiots rush for the 1080 you can expect the same supply shortage as with the Fury and similar prices at around 1000$.

2. Founder's Edition aka give us more money.
There is simply no reason not to wait the 2 weeks for aftermarket designs. Why pay 100$ (or 70$ in case of the 1070) more for an inferior cooler? And when aftermarket 1080s and the 1070 are first available Computex will be over and we should have seen what Polaris can do.

3. Polaris
Let's face it, very few people are going to buy the 1080. Most are interested in the 1070.
Now let's take a look at AMD's Polaris 10 which is supposed to launch at Computex, before the 1070 is even available.
Rumour has it that the 480X will beat the 980 Ti as well. Guess what, that's exactly where we expect the 1070 to end up*. But now for the price: 349$. Compare that with the 1070 at 450$ for the reference design and 380$ for aftermarket.
Are you sufficiently hyped yet?
But wait, there's more. The 480 (non-X) should be somewhere between the 390X and Fury X (or 980 and 980 Ti) but at 2xx$, the price bracket of the 380X.
So far nVidia has not announced any GPU that would compete with that.

*We don't exactly know yet how far it'll be cut down and how much of an impact GDDR5 instead of GDDR5X (-> lower bandwidth) will have.

4. Reviews & potential problems
Always wait for reviews before you buy.
As it turns out I overestimated nVidia. They never updated their roadmaps. "Full" DX12 support was planned for Volta and it stayed that way. Because of that they were able to deliver Pascal at around the same time of Polaris but it ends up being basically Maxwell on 16nm FF. Again, none of this is confirmed so wait for reviews, but Pascal might not have asynchronous compute either. In other words Pascal cards that are as fast or slightly faster than their AMD counterparts in DX11 might get massively shit on in DX12 again, just like we've seen with Maxwell.

There's also this https://semiaccurate.com/2016/05/03/thing-go-bumpgate-in-the-night-for-nvidias-gp100-pascal-gpu/
If this affects the GP104 too, oh boy, nVidia would be in some deep shit.
Links to the articles of the last time this has happened:
http://www.theinquirer.net/inquirer/news/1004378/why-nvidia-chips-defective
http://www.theinquirer.net/inquirer/news/1013947/why-nvidia-duff-chips-shoddy-engineering
http://www.theinquirer.net/inquirer/news/1036374/what-nvidia

I'm not overly concerned since I won't be buying any Pascal GPUs anyway if they don't have ACEs. Anyone in a country with proper (or any) customer protection laws probably won't have a problem either. E.g. in the EU should the GPUs fail within 6 months nVidia has to replace them, unless they can prove that manufacturing defects are somehow the customers fault. In the US however you'd be fucked, so I'd wait until it's confirmed that GP104 isn't affected by this.

#21
I'd be more hyped for the 480(X) since they should be cheaper.

#22
Unless you want to throw away 300$ more on the reference designs wait until June 10th. For watercooling you have to wait for full blocks anyway so I don't see the point of ordering as soon as possible.

#29
Neither the 1080 nor the 1070 are actually available yet.
If AMD launches the 480 and 480X at Computex as expected then they would be available sooner than the 1070 or aftermarket 1080s.

#30
Wrong, the 480X should compete with the 1070.
I'm also not sure how 6 months are a year now.

#34
Better GPUs don't help at all in TF2.

#1217
No, don't get a GT 740.
Like I said it's barely faster than the iGPU,
So try the iGPU first, on low settings you should definitely get more than 100 fps in TF2.

Get a proper GPU later, if you need one.
E.g. the 250X which is cheaper and faster than the GT 740.
http://pcpartpicker.com/part/xfx-video-card-r7250xcgf4

Or if you can afford it something like a 370 or 750 Ti, since they are significantly faster than the again, but only slightly more expensive.

Get the SSD first. You want the OS on your SSD, so unless you like reinstalling windows getting the HDD first would just mean unnecessary hassle.

tl;dr
GT cards are shit.
Try the iGPU first, only get a discrete GPU if you really need it.
If so get a proper GPU (not GT).

SSD first.

#1218
Can't verify that.
Skylake also uses a different iGPU than Haswell so I'm not really concerned.

#1215
You would mostly wreck your room when that PSU bursts into flames.

Also you are getting ripped off on that GPU. 90$ for a GT 730 is ridiculous.
For less money (~70$) you could get
-a GT 740 which is not quite as terrible, but only barely (50% faster)
-a 250X which is about 2.5 times as fast.

For slightly more money (~100$) you could get:
-a 370
-a 750 Ti
both of which are 3.5-4 times faster.

Or for no money at all you could use the i3's iGPU which is actually faster than the GT 730.
http://gpu.userbenchmark.com/Compare/Nvidia-GeForce-GT-730-vs-Intel-HD-530-Desktop-Skylake/m12582vsm33102

Budget would be nice to know.
I don't like the case either, but I won't stop you.

Here's what I'd do:
PCPartPicker part list / Price breakdown by merchant

CPU: Intel Core i3-6100 3.7GHz Dual-Core Processor ($111.99 @ SuperBiiz)
Motherboard: Gigabyte GA-B150M-D3H Micro ATX LGA1151 Motherboard ($76.97 @ Newegg)
Memory: G.Skill NT Series 8GB (2 x 4GB) DDR4-2400 Memory ($30.69 @ Newegg)
Storage: Samsung 850 EVO-Series 250GB 2.5" Solid State Drive ($84.88 @ OutletPC)
Storage: Seagate Barracuda 1TB 3.5" 7200RPM Internal Hard Drive ($45.89 @ OutletPC)
Case: Fractal Design Core 1000 USB 3.0 MicroATX Mid Tower Case ($34.99 @ Micro Center)
Power Supply: Corsair CX 430W 80+ Bronze Certified Semi-Modular ATX Power Supply ($28.98 @ Newegg)
Total: $414.39
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2016-04-26 13:44 EDT-0400

Drop the HDD if you don't need more than 250GB.
Get the 120GB version of the SSD if you do need more than 250GB but absolutely need to stay below 400$.
You could also get the i3-6300 or even 6320 instead if your budget allows it, but it wouldn't make much of a difference.
If you want to you can get a GPU for other games later.
PSU alternative in case the deal for the CX430M ends: EVGA 500B
CX430M, CX500M, CX600M, CX430, CX500, CX600, 500B, 600B are all possible option, get whichever is cheapest.

#1200
You can record mumble with either, but that's not what you're trying to ask, right?

#1201/1203
When does he want to build it?
Does he know about AMD Raptr/plays.tv?
Does he want to overclock?
Resolution/settings for Fallout 4?

#1204-1212
Y'all need to chill.

1. Fuck Newegg only partlists. Also fuck their return policy.

2. 212 Evo only makes sense if it's on sale, you can't afford a better cooler and can't afford to wait until you can afford a better cooler. The budget would allow for a good cooler but he doesn't need an aftermarket cooler for a locked CPU at all.

3. Z mobos for locked CPUs smh.

4. An i5 won't get you noticeably better performance in TF2 than an i3. Single digit percentages at best, worst case 3 cores turbo at the same time then even the i5-6600 gets dragged down to 3.6GHz and beaten by the i3-6320 at 3.9GHz. Even in Fallout 4 it's more about clockrate than physical cores once you reach 4 logical cores (i3). I don't know about CS:GO though.

5. 16GB RAM. Seriously? Yes, there are games that will use it. There are also programs that will use 64GB RAM. And programs that can use 8 CPU cores. But as long as he isn't running those he won't need either. Going by your logic he should get a 980 Ti now since PCIe GPUs will stay the standard for a long time and he might need it.
RAM is the same as GPUs, it only gets cheaper. Get as much as you need, but not more, and the option to upgrade later. 8GB and a mobo with 4 DIMM slots will suffice because Fallout 4 won't use 8GB.
http://www.gamersnexus.net/game-bench/2177-fallout-4-pc-video-card-fps-benchmark-all-resolutions

And TF2 and CS:GO are still 32bit, they absolutely can't use more than 4GB.

6. What the actual fuck are those mobo comparisions? What are you trying to prove by comparing the most expensive and cheapest mATX and ATX boards? You could prove so much bullshit with that logic.
Compare similar boards, like the ASRock x70 Pro4(S) and x70M Pro4(S). You'll notice that they are virtually identical except for superficial differences due to the physical size (e.g. one more PCIe x1 slot, 110mm instead of 80mm maximum M.2 card size). The full ATX versions are 5-10$ more expensive simply due to material cost. For the same reason full ATX cases have to be more expensive than microATX cases (ceteris paribus).

7. Name at least 2 PCIe cards (excluding GPUs) that you need for a gaming PC. Otherwise I don't see mATX limiting upgradability.
Unless you were talking about the "13.92" max GPU length", but I don't see that becoming a problem any time soon.

8. RAM quality over quantity. I've tested it for TF2 and Fallout 4 is the same.
http://www.techspot.com/review/1089-fallout-4-benchmarks/page6.html

#1197
Yeah, just post again here 1 or 2 weeks before you want to build it.

Overclocking would get you better performance in TF2 but since it costs time and money it would just be a waste of money if you won't do it anyway. I'm not sure if I could've made it happen on that budget anyway but I live for challenges like this.

#5
You don't know how AyyMD memes work.
If your PSU goes up in flames because of 30W more I wouldn't trust it with any CPU.
All in One liquid coolers are useless and overpriced.

Actual advice in the pc build thread.

EDIT: Chill out, the rest is correct.

#1195
When are you going to build it? New GPUs and more importantly CPUs this summer.

Are you willing/able to overclock?

The AMD FX-x3x0 (6300, 8320, etc) are really not good for TF2 for two reasons:
1. The concept behind them is to trade single threaded performance for more cores. TF2 loves single single threaded perfomance and can't do anything with more than 3 cores. So the whole concept of 2x the cores, 0.8x the single threaded performance just gets you cores you won't use and 0.8x the performance.
2. They are 4 years old. Don't buy a 4 years old CPU.

What you should be looking for is an i5-6600 (or 6500 since it's slightly cheaper), or an i5-6600K if you want to overclock, or wait for their respective successors.

#14
CPU for TF2.
GPU for The Division.

I like how I reiterated the same question as #4, yet #10 still manages to ignore it.

How does he expect us to figure out what he needs? Sure I could play the guessing game, but I shouldn't have to so I won't.

I'm actually everywhere.

Bumping this to check if OP's problem has been solved.

#1
Do you want/need more performance? If no, don't upgrade.

If yes, what program/game do you want to perform better?

See #4

#6
Same as #1.

Since I got asked about it here's some updates on what I think is going to happen regarding the new GPUs.

I can't give you any exact performance numbers, since that depends on clock speeds (which might be tweaked to make the performance fit the segment they're going for) and the actual number of execution units, which I can only guess at this point. I definitely won't talk about how the salvage chips stack up to each (e.g. will the cut down version of "big Polaris" beat the cut down version of "mid Pascal") since that depends on those and on top of that also how many cut down versions there will be (2 or 3?) and how far cut down they are (which again might be decided by market placement, rather than technical reasons).

So here's what I think is most likely (no guarantee that it'll actually happen exactly like that):
AMD (code names are a mess):
"small Polaris"/Polaris 11/Baffin: Performance around 370 / 380 (AMD) or 950 / 760 / 960 (nVidia), 3-4GB GDDR5, vastly more efficient but not all that interesting on desktop. Important for AMD since they need efficient "low end" GPUs for mobile where they've been losing ground to nVidia. This is the one AMD demoed back in January, so it should definitely be ready for computex, launching it sooner doesn't make much sense. Availability might be bad since most of them will be going into mobile, unless you care about power consumption immensely AMD already got a very strong lineup in that performance segment.

"big Polaris"/former "mid" Polaris/Polaris 10/Ellesmere: Performance around 390X / 980, 4-8GB most likely GDDR5, again vastly more efficient, we're talking just above 100W. These are the cards that I think will be the most interesting. Think 390 except with better performance and half the power draw. Since there's no HBM and it's a fairly small chip compared to the old high end chips we could see some pretty good prices here. I'm expecting it to be ready slightly after small Polaris so they're probably trying to have it in time for Computex as well. If they have to delay it for some reason or want GDDR5X at all cost we might only see it later in Q3, in that case GDDR5X seems possible/likely. They'd probably do a "on paper launch" at Computex with working GPUs, but not have enough volume to ship them until late Q3.

Vega/"small Vega"/former "big Polaris"/Vega 10/Greenland: Performance around 20-30% faster than Fury X / 980 Ti, 8GB HBM2, same deal with the efficiency (>Titan X performance at 980 power draw (<200W)). Probably around Q4 2016 at best, Q1 2017 seems most likely. This is mostly because of HBM2 availability, since they probably won't want to run out of stock like with the Fury.

After that we'll most likely see "big Vega"/Vega 11/ maybe Vega 20 (fuck these code names) sometime around Q2 2017, significantly faster than the Fury X / 980 Ti, ballpark figure would be +70%, probably 16GB HBM2, but this is all pure speculation. At that point it's a matter of how far you want to go since bigger chip size does have a significant negative impact. Going bigger doesn't only affect those GPUs, it also means your next GPUs with a possibly more area efficient architecture need to be bigger to be faster by large enough of a margin to sell well. So you want to hold back as much as possible without giving up market share/sales/profit to the competition or even just due to people not buying and waiting for the next gen. On 14/16nm neither AMD nor nVidia can or want to go balls to the wall with 600mm² chips for the sake of the performance crown in the consumer segment since it wouldn't be profitable. Both will put a lot of thought into how big their new high end chips will be so I really can't predict that yet.

Now nVidia:
GP107/"very small/entry level Pascal": Very similar to small Polaris/Polaris 11, maybe slightly slower. Will probably appear in Q1 2017 so not worth talking about yet.

GP106/"small Pascal": Performance around the 380X / 970, so between small (11) and big (10) Polaris, probably 4GB GDDR5X, launching probably Q4 2016.

GP104/"mid Pascal": Performance slightly faster than Fury X / 980 Ti and therefore somewhere between big Polaris and small Vega, 8GB GDDR5X, possibly GDDR5. Now this is where it gets complicated. I'm fairly certain that this is the chip that nVidia was just about to get in January, which means that it won't be ready for computex. There's also the fact that quite a few people are fairly certain that it will use GDDR5X which they won't get enough volume until then either. On the other hand they probably don't want to let AMD just grab the juicy 300-500$ market, so a paper launch at computex, which will indeed look impressive on paper since it should beat big Polaris significantly, seems likely. At the same time I'm not all that fussed about it since it will beat the Titan X. That's right, performance is a downside here. They don't want to massively devalue all of their cards so they actually have to charge a massive premium for them. Not that nVidia will mind that or wouldn't have done so anyway. Since they still have to compete somehow we could see some very nice price drops on the 980 and 970 though. Again nVidia won't mind since they got terrific yields on those. In light of the DX12 issues of Maxwell, possibly another round of driver induced performance gimping and the fact that you really don't want a monopoly that probably won't change my recommendation. Unless AMD fucks up "big Polaris" is where it's at.

GP102/"big Pascal": see "big Vega", about 70% (plus minus 20%, it's that uncertain) faster than Fury X / 980 Ti, 16GB HBM2, Q2 2017, pure speculation.

tl;dr
Releases:
Computex (June):
Polaris 11
Polaris 10
GP104 (on paper)

Q3:
GP104 (actual availability)

Q4:
GP106
possibly Vega 10 (on paper)

early Q1 2017:
Vega 10 (actual launch)

sometime Q1 2017:
Gp107

Q2 or later 2017:
Vega 11 or 20, whatever they'll call it.
GP102

Performance of the full chip, cut down versions will be lower* (lowest to highest):
GP107
for reference 380 / 960 are about here
Polaris 11
GP106
for reference 390 / 970 are about here
Polaris 10
for reference 390X / 980 are about here
for reference Fury X / 980 Ti are about here
GP104
Vega 10
GP102
Vega 11

*There will be overlap, e.g. cut down Polaris 11 between cut down GP107 and full GP107 seems likely

still tl;dr
Computex (June) 2016: POLARIS 10 GET HYPE

#321
I don't think I understand the question, what do you mean with "CAS Latency > Access time/Actual latency"?

Yes, the demo used to be static, thank Valve for breaking demos.

See #3
In Steam go to Settings -> In-Game
The last option is a drop down menu "In-Game server browser: Max pings / minute".
Set it to the lowest (250) then work your way up until your Internet crashes again and use the highes working value (or one below that to be safe).