Firstly - great post. And cheers for your input so far. I have based my cooler purchase of the h110 off your previous posts. Ok so clearly I am missing a couple of things then. I know I have definitely used to much paste according to those images you posted but no more than I what I have seen other "guides" on. You have recommended the least amount of paste so far compared to what others have had to say about.
The h110 is priced a bit high, but honestly I think it's worth it since the ~$30 cheaper x60, which is the exact same cooler, is from NZXT, who has a very spotty service record, especially at the moment. Long story short and a 7 months and a post on every single overclock forum about still not receing my rebate from nzxt, the choice between the identical Asetek 280mm units is NZXT is the worst customer service and RMA in the business, and Corsair is the best customer service and RMA in the business. After a rather nasty fight with NZXT, I went with Corsair.
Most guides are awful about paste recommendations, because they go by footprint. I think it was Benchmark Reviews who's the worst, but if you got enough paste to cover the entire IHS then you are going to see increased temps from using too much.
If you find paste recommendation guides that go by benchmarked, actual temp results, you'll find what I am saying is consistent with what they say. There is one site out there, I recall, that tested Pea vs Dot vs Ricegrain vs Spread, and they basically found that there was zero difference in paste applications so much as you avoid using too much. They pretty much said you couldn't use too little, but once you went too much, boom, +5C or more. They also said spread was just slightly worse if you just spread a rice grain amount, but the problem most people do with spread is they try to cover it all and end up using way more than a rice grain.
If you look even deeper, I don't think you'll find a single temp-benched thermal paste comparison where they ever say "X is too little paste".
(now dont laugh I've tested and it's fully stable) 2x Gskill 2gb pc12600 1600mhz 9 9 9 24 + 2x Gskill 2gb pc10800 1333mhz 9 9 9 24 dimms running at 1333 config
There isn't anything wrong with using 2 different sets of RAM. Now, there is sort of the blanket, 'omg dont use 2 different sets of RAM' answer that is regurgitated by people who don't know better, but on a good board you can even set different timings for each channel just fine.
It's very likely that your Gskill 1333mhz RAM, might even be the exact same RAM as your 1600. You dont tell us what model number or exact ram you are using (there are 50 different gskill ripjaws X, i know, it's confusing), but basically there are only like 4-6 different types of RAM ICs (with a couple sub-types).
Also, nothing stopping you from just doing 1.75v and cranking up the RAM speeds, since DDR3 runs very, very cool. But based on the timings you said, you probably have some really low-end spektek or qimondas that won't do anything. But given Haswell's awesome IMC, you should really get better RAM when you can.
Here's an amazing kit:
They are 2400CL11 but easily capable of over 3ghz, for $63. If you buy something else on newegg first, you'll get a 10% off Memory coupon (if you dont have one already), which will drop it to $56.
Huge difference in 3ghz and 1333, will definitely be a big performance boost. I mean it's not necessary you get it asap, w/e, but you could sell your current ram for ~$50 so it's basically a free, noticeably large performance boost.
At the moment of typing this I have successfully booted fine into windows and web browsing etc at 45x @ 1.27v. If I put any stress on the cpu she'll blue screen & restart (no hardlocks thankfully) The one thing I havent attempted yet is the VRIN settings. NOOB question time - what setting is the VRIN called on the asus mobo's and how much should you recommend I raise it to to help stabilize the system.
CPU Input Voltage. Max is ~2v.
Nothing wrong with hardlocks, or having to do cmos reset, it doesn't do anything bad. You should just set your VRIN to +.4-55 of vcore. How much you want to fine tune is up to you, but if you use the optimal VRIN then you can actually use less vcore, ~.02v less.
What I did was dialed in around what voltage was needed using +.4VRIN over vcore, and then played around. So I would test 1.4 + .4VRIN, then .45, then .5, then .55, and then i'd step up vcore a bit and do it all over again. It's up to you how fine tune you want it.Edited by Belial - 7/27/13 at 5:30pm