Deciding on a Skylake CPU--i5 vs i7, k vs. non-k, etc., also RAM


@Tman , what do you mean by “registering your account with Microsoft”? You mean just activating Windows or whatever, or something else?

Speaking of Windows, is there a way to grab the Windows 10 installer from a legit source, install it on an SSD, do whatever hardware testing is required, then activating the Windows 10 with a retail Windows 7 key that you don’t plan to use anymore (I’ve heard that works–although just now the chat person on the MS store said no go)?

Or should I just get some kind of Linux “distro on a stick” that will let me do the testing with memtest86+, Prime95 etc.?


Yes, make sure you’ve activated your windows installation - here is the MSFT support article on it.

@Papageno I kept my windows 10 install, but I stumbled upon this article when I was preparing - perhaps it might help you?


BTW, I’d like to report that my motherboard (Z170-A) CPU (i7 6700 with stock HSF) and RAM (Corsair Vengeance LPX running at 3000 MHz) posted just fine tonight. CPU temp in the BIOS hovering around 33/34 C, Motherboard at about 23, slightly over room temp. During the rest of this week I’ll be installing an SSD (the OS drive, a 500 GB 850 EVO)*, Windows 10 and then testing it with memtest86+ and Prime95 over the weekend.

Question for those with this motherboard: displayed prominently over one of the sets of two SATA 6 GB connectors is a red sticker or something that reads “OS drive.” Can I presume they’re referring to the top one? Why is that one so special? I don’t think the manual says anything about it.


Maybe that’s the default HD that is selected in BIOS for boot sector stuff when the machine figures out which HD to boot from?

Don’t worry, Windows will find a way to fuck that up anyways. :P


Fwiw, I actually have NFP-12 fans on both sides of the Noctua U12 heat sink. That rig is almost 5 years old now. I did have one of those two fans go bad on me And had to replace it in September, but have been very happy with it.

@Papageno nice!


I think this is the case, but it can be configured in the BIOS.


I remember there was a supplemental flier with the manual that discussed this.


Indeed there was, thanks.

Now I’m faced with this: I installed Windows 10 Pro (yet to be activated–I’m planning to use the retail Windows 7 Ultimate key that I’m using on this, my old machine), then downloaded something to monitor temps and Prime95’s latest version. Fired up the latter and started the “blend” test, and noticed with some alarm that the CPU (which at idle seems to sit at around 30C) was climbing to 79-80, so I stopped the test (don’t know if it was maxxing out there or if it was still climbing). As a comparison, my i5 2500k never got above 60C around a year and a month ago running the then-current version of the test. Now I know that the Sandy Bridge only has 4 logical cores and this Skylake chip has 8 (don’t know if that alone would account for some of the difference). I also BTW had some trouble seating the stock HSF the first time and had to momentarily lift it off before getting it back on right (with the same orientation, so presumably the paste stuck to the CPU and heatsink was not too disturbed).

Over on ArsTechnica there was a thread where someone thought that his i7 4770k was running too hot (about 80C) and the guys there who should know said don’t sweat it. Of course that’s a different chip as well. Anyway, I don’t suppose that if it maxxes out at around 80 there will be a problem.


Just downloaded that here - on my 6700K it seems to have settled at 62 C. About 5mins in so far.

Edit… Changing now as it does different stuff… Highest 71 lowest 38 ish. Dunno how long these tests run… forever?

Edit2: Gave it 15 mins for you and that behavior stayed consistent - low 60s plateau with occasional oscillations.


Hmm, what cooler are you using? I’m using the stock Intel one that came with the non-k version (mine is a non-k i7 6700). That might explain the difference. Over on Ars a frequent poster who goes by Hat Monster says that 80 is probably about right with the stock cooler, and won’t damage the chip. I might switch to something else later though, because that seems pretty high to me as an ongoing thing over the life of the computer.

And yeah, the tests do run forever. The software is meant to find new prime numbers, so…


One of these. Nothing too special but non-stock.


Yeah, I bet that’s going to run quite a bit cooler (and quieter) than mine with the stock HSF. It fit OK in your case, and didn’t interfere with other stuff?
Also, did it require a backplate under the motherboard?


No, no backplate. It’s an easy fit in a Fractal Nano case. I actually got it because I thought it would be easy to fit compared to the disaster it replaced. Which it pretty much was.


Do you have 6700K overclocked, or at least the turbo enabled so it runs at 4.3 GHz? I ran Prime95 on mine with the Cryorig H7 cooler and during the first phase of the test I staying around 60 degrees, but when the next stage started I quickly jumped up to around 77 and slowly climbed to 80, at which point I stopped it. The H7 is a pretty beefy cooler so I wasn’t expecting to get so hot. I think it only ran for less than 10 minutes.


Use LinX, not Prime95. There’s some kerfuffle that Prime95 isn’t safe on newer intel CPUs. That’s probably BS, but it does get the CPU dramatically hotter than any other burn-in tool when it uses small FFTs.


Thanks for the advice, stusser. I’m sad to report that on the second try the CPU temp just kept climbing using Prime95’s “blend” test. I stopped it at 89C because I didn’t want to fry the 300-dollar part.
I hope that the other tool you mention puts the RAM through its paces, because that’s what the blend test does, and that’s what finally caught the problem with the first RAM kit I bought for my old machine 5+ years ago (the kit that kept passing memtest86+ with nary an issue). I dealt with impossible-to-diagnose BSOD’s for 4 effing years due to that BS.


At least from task manager it looked like it was hitting 4.16 GHz.


Ugh, so I thought I’d reseat the heatsink and I really shouldn’t have. After doing so my temps were higher (creeping up toward 50C at idle). Maybe I didn’t clean the old thermal paste off well enough, but I’m pretty sure I also didn’t apply the new stuff properly. I’d always read/seen that you don’t want to use a lot, and I didn’t, but maybe I didn’t use enough, or maybe it didn’t get spread out enough on the processor (I used the “small pea sized amt. in the center of the CPU” method). I think the next time I’m going to try to spread it around a bit.

Also will rubbing alcohol make a good enough solvent to get the old stuff off properly? That’s what a link at PC Gamer recommends.

EDIT: In my research it turns out a paper coffee filter makes a good lint free cloth, so I grabbed one, cleaned off the old stuff really well, applied new stuff to the heatsink surface, and used one of those square plastic doohickies from a bread bag to spread it around nice and thin. Reseated, and…SUCCESS. Idle temps at around 30C.


Application method doesn’t really matter, just don’t use too little.

IMO cleaning off the old stuff is the hard part. I’ll have to remember the coffee filter tip.