Today I will be (hopefully) be completing my build. If you've been following, the CPU is the 6950X and I'm pairing it with the Swiftech H320 X2 Prestige (with 3x Corsair ML120 fans on top - pull) in a DG-87 case (with 3x Corsair ML140 fans on the front as intake)
 
I HAVE Arctic Silver 5 on-hand. Would I be better off waiting a day or so and buying the Gelid GC Extreme TIM? Or am I alright with what I have until it needs to be re-applied? I've read a bunch of articles and comparisons and the GC Extreme seems to be around ~5C cooler than the AS5 on certain benchmarks but other than that, it looks like most TIMs are within 1-3C of each other. I don't expect my temps to be high anyways with my current cooling setup but I am planning on an OC of ~4.2Ghz and part of me just wants this thing built already
 
I really do appreciate all of the feedback given to me thus far. The friendliness and advice really means a lot to me a first-time builder and I'm trying to make this go as "ideal" as possible.
 
Also, the debate between the "line" and the "rice/pea" method. Intel's specs say the "line" method for these processors but personally I'm more comfortable doing the rice/pea for the first time to lower the risk of excess bleeding off the edge (especially since it's semi-capacitive and don't want to run the risk of shorting anything). Again, all tests show the methods only vary 1-2C between each other. Should I just do what's comfortable?

When you're all said and done, some finished images of it all would be nice. 

GTXJackBauer
When you're all said and done, some finished images of it all would be nice. 

Will do! I started an album on facebook but it's just a matter of moving everything here lol.
 
 

XrayMan
 
Getting the better ones is always good. Let us know how it goes.

SethH

 
Parts pic
The fans on the left are the Corsair ML120 Pros and on the right they're the ML140 Pros.

Nice!        

Too many Seths. There can be only one! 
 
On my 6900 I used the EK Ektotherm that came with my EK Supremacy EVO block. On my 1080 SC I had used the last of my Prolimatech PK3. Honestly it's not going to matter all that much which one you use if you look at the benchmark results and pick one in the top 5 or 8 or so; they'll all be within 1-2 C of each other. I used EK's recommended cross pattern. The Ektotherm is less viscous and will undoubtedly spread easier, but they'll spread out just fine under the pressure of the water block if you tighten the hold-down screws properly.
 
Unless you bought a $1700 lemon, 4.2 GHz should be very easy for you to achieve, and without resorting to extreme voltages. My VCORE as reported in HWMonitor is currently just over 1.25V with my 6900 OCed to 4.2 GHz. For 4.3 GHz I had to use just over 1.3 V, but other than that it was also extremely easy to reach. I didn't try 4.4 GHz, but I suspect it wouldn't have been too hard to run, but I just didn't see any point. For all of my uses the 4.2 GHz gets me 95% of the bang for like 70% of the buck. I also raised my ring ratio (which will OC your cpu cache) to 35 from its default (which was either 30 or 32), and slightly nudged up the ring voltage to support that. The temperatures are still great (down in the 30s at idle, around 63 C or so under folding at home.

I'm glad to see that you went with the Swiftech unit. I was cooling my overclocked Xeon 5680 and an overclocked EVGA 1080 with the baby brother of your unit - a Swiftech H220-x. You should be fine with that unit. One thing you can easily do to improve your cooling is buy two more fans and stick them onto the bottom of the radiator so that 2/3 of it runs as push/pull. Get the same PWM fans you're using for the top and there's still plenty of room on the Swiftech PWM controller harness that controls them and the pump. This will improve your cooling significantly, and is a trivially easy mod. There's no problem running 2/3 push/pull and 1/3 push only or pull only. Unfortunately the Swiftech pump/reservoir unit will block the third radiator slot from an additional fan, but oh well.
 
I'm glad you went Swiftech because if this is your first foray into water cooling you will find it trivially easy to cut into your existing lines and add in new components to the loop, such as a gpu block, another radiator if you feel the need, etc. I'm still using my H220-x for the pump unit and it's 2x120mm radiator with 1 push and 1 push/pull fan on it, but have added a 2x180mm radiator to the loop (four crappy fans in push/pull), and am currently cooling my 1080 and 6900 with it, and will very soon be cooling the motherboard chipset and VRM as well, and in the future another 1080. This is the flexibility you get with the Swiftech - even as it's an AIO unit, it's totally expandable. Someday I'll replace the H220-x with a dedicated pump and reservoir, but have not yet felt the need to do it. The Swiftech pump is a well-regarded pump that is easily capable of supporting an expanded loop.

I took Nateman's VRM and PCH block apart and looked inside, and apart from the right angle turns they're pretty wide open and I don't anticipate them being very restrictive. Adding anything to the loop will drop the pressure a little of course, but I don't see these as dropping it all that much. I realize they're not strictly necessary, but oh well, not everything we do on our machines is because it needs to be done. 
 
ETA: I believe the Swiftech units use a slightly less powerful version of the DDC as their pump unit. They're way better than typical AIO pumps, so I'm not too worried about whether it'll be enough to support my gpu + cpu +vrm +PCH. I read a thread somewhere where a guy asked a Swiftech representative whether the pump would be strong enough to support a setup similar to mine and he said it'd be fine. I guess I'll find out. I do plan on going to a dedicated pump/reservoir combo in a future expansion, possibly when I add a second 1080, but we'll see. 

I have finished! I'll post pics tomorrow when I get home from work as they just added a new Firewall filter here and every photo upload site is blocked.

Sorry for the delay, I'm still trying to do Windows updates. I left "Checking for updates" on overnight while I'm at work because it takes FOREVER for some reason.
 
Also, I've already overclocked it to 4.2Ghz @ 1.252 Volts. When running the newest version of Prime95, after just a few minutes I started getting temps in the 80s with the highest of 88 before I stopped it. At first I thought it was my cooler/waterblock install but then...
 
Using Aida64, the max I got after 30 minutes or so was 67C (this was just a peak), it averaged around 63-64 with fans on PWM (which I have to change btw because they love to go to full RPM and it's annoying).
 
Should I just not run Prime95? A lot of people seem to have the same issue with temps.

Just ran Prime95 Version 26.6 using the Small FFT for almost 15 minutes, max temps @ 4.2Ghz & 1.252 Volts was 70C.
The VCore reported on OpenHardwareMonitor is actually the VRM voltage I believe, NOT the CPU. CPU-Z shows the correct 1.252V for my CPU.
https://puu.sh/rs7IA/4d4d759965.png
 
 
My CPU-Z results/validation. I upped the BCLK in BIOS to get it over 4.2GHZ (upped it to 101 instead of 100 because it was downclocked slightly). Voltage says 1.248 even though when idle it's 1.252 and override is set to 1.252 soooo lol.
http://valid.x86.fr/7dtdcg

Nereus
 
Bit late to the party, but here's my method:
 
I have been using MX-4 the past couple of years. I used to always use Shin-Etsu, and there's no doubt it is a good tim, but I found it a little viscous and sticky and harder to work with and spread evenly, whereas MX-4 seems just about perfect consistency - spreads nicely without being near liquid.
 
My application method is a bit involved.
 
I first treat and clean the surfaces of the CPU and the cooler using Arctic surface cleaner and purifier (this stuff).
 
Next I wrap some saran/glad wrap around the end of my finger (make sure there is a smooth surface of saran on the pad of the finger), then take a very small dollop of tim and smear it on both surfaces - the CPU and the cooler plate.. and I mean only a tiny amount, enough to basically take the shine off the surface but very little beyond that.
 
After that, I use the credit card method (ideally a thinner credit card with some flexibility, not a standard bank credit card) and place enough of tim on the CPU surface to spread it evenly over the entire surface in a very thin even layer. This is where the MX-4 was significantly easier to work with than the Shin-Etsu, because it is not so viscous and sticky, and allows you to make a nice thin even layer with relative ease.
 
Slowly and gently place the cooler down, and I give it a very slight bit of pressure and give it a little rotational twist - just a few degrees is all, then hold it in place and carefully secure it down, and make sure you don't just fully tighten down one corner, then the next, then the next etc., but instead just screw down all four corners until there's just the slightest resistance, and then work your way around doing maybe a half turn on one screw, then a half turn on the diagonally opposite screw, then a half turn on one of the remaining screws, then a half turn on the opposite side to that, and rinse and repeat until it's completely screwed down. I get a nice even mount every time using this method, even though it takes a while.
 
Once mounted, I give it maybe a half hour run with Prime 95 just to 'cure' the tim and help 'melt' it into a nice clean application, although most tims don't need that now, but I feel better about it anyway, lol.
 
 

+1 MX-4
 
I like a razor-blade --> you need to finish with One continuous wipe across the TIM

sethleigh
Yeah I think you're right. cpu-z is showing 1.300V right now for my cpu, which is running at 4.2ghz with 3.5ghz cache not quite fully loaded, but with 6 of 8 cores folding, and the other 2 cores servicing my two gpus which are also folding, and my web browers and whatever else. Meanwhile VCore in HWMonitor is showing 1.256V.
 
I'm going to change my offset and see if it still runs stably. I'd had it a little over 1.3V when I was running at 4.3ghz, and it crashed when I tried changing it down to 1.25. I thought I'd dropped it when I went to 4.2ghz, but it's at 1.3, so I think I may have some room to drop it more. Btw, with this particular load I'm seeing around 60-62C cpu package temperature in my custom water loop that's also cooling my 1080 at 100% load, with the cores fluctuating continuously around that same level. I may try Prime95 for a few minutes to see what it does to the temperatures.

 
Those are pretty good temps. I don't know how much the 2 extra cores of mine fully affect temperatures but both of us are in the "green" temp-wise which is a huge plus. Using Aida64 my temps are only 63-64C so Prime95 is only 5C more than that (69-70C) when some people are high 70s and low 80s so I'm definitely happy with this. I guess I put the cooler & paste on right for my first time
 
IF you are going to use Prime95, use version 26.6 Small FFT test. I had originally used the newest (like 28.9?) and my temps were through the roof hot! Like 80s with the highest being 88C at which point I stopped it immediately and had a "what the heck" moment (it was only 5 minutes of testing and only about 30 seconds of the 80s so no harm done). Version 26.6 is MUCH better.
 

sethleigh
I noticed that too about my bclck and it not really being 4.2ghz, but rather something like 4.197 or whatever. It didn't bother me though. I grew up with analog watches and clocks and fully accepted that "quarter past 1" could really be 1:13-1:17 or so and it just didn't matter. I hear ya though.

 
Honestly, it's more of an OCD thing. I don't think I'm going to push it further because a 2200 score in Cinebench compared to like 700 with my old CPU (AMD fx-9370 @ 4.72Ghz) is good enough for me haha.


 
Next step is working this damn air bubble to the top of my reservoir. At high RPM my pump makes a clicking sound unless I tap the reservoir/pump, at which point the sound lessens so I know it's not a hardware issue.

GTXJackBauer
There's a new version of Prime that many don't use because it gets CPUs really crispy as many resort to using a version prior to that update.  I can't recall exactly which version that was.  You will never get that kind of of loads in real life operations so you shouldn't have to worry and I wouldn't push the CPU more than 80c at the most.  They all have thermal shutdown for protection purposes but that also doesn't mean you've degraded your CPU at those high temps.  Obviously you'll need to let it sit at high temps to really do some damage overtime but for a few seconds you should be ok.  Being a very expensive high end CPU. I think 4.2-4.4 Ghz is more than enough.  
 
Ok.  After a little research, I found this for you which is exactly what we need.



 

Do NOT run any versions of Prime95 later than 26.6. Here's why:
 

Core i 2nd, 3rd and 4th Generation CPU's have AVX (Advanced Vector Extension) instruction sets. Recent versions of Prime95, such as 28.5, run AVX code on the Floating Point Unit (FPU) math coprocessor, which produces extremely high temperatures. The FPU test in the stability testing utility AIDA64 shows the same results.

Prime95 v26.6 produces temperatures on 3rd and 4th Generation processors more consistent with 2nd Generation, which also have AVX instructions, but do not suffer from thermal extremes due to having a soldered Integrated Heat Spreader and a 35% larger Die.

Please download Prime95 version 26.6 - http://windows-downloads-center.blogspot.com/2011/04/pr...

Run only Small FFT’s for 10 minutes.

Use only Real Temp to measure your Core temperatures, as it was designed specifically for Intel processors: Real Temp -http://www.techpowerup.com/downloads/2089/real-temp-3-7...

Your Core temperatures will test 10 to 20C lower with v26.6 than with v28.5.

Please read this Tom’s Sticky: Intel Temperature Guide - http://www.tomshardware.com/forum/id-1800828/intel-temp...

Thanks,
 

 
 
THANK YOU!! This ended up solving my issue and I am very thankful for that.