Support NeoGAF

[Timedog]

Don't be a pussy. Do it, dawg.

But seriously. I was getting 80°C temps at 4.5Ghz and 1.35V. I'll post pics soon but now I'm getting 62°C temps at 1.4V at 4.7Ghz.

Absolutely worth doing but be extremely careful. I also have to say, don't bother if you're not going to use a liquid metal TIM.

oh I'd definitely use liquid metal TIM. Gotta wait for this project to be done though in case I fuck up and am out a computer for awhile.

 

[John Kowalski]

Hmm.

I understand some of those words.

 

[LordOfChaos]

instead of shooting up to 80°C+ right away I'm still sitting at 56°C

Man, Intel IHSs are still pretty horrible.

If it had good paste and no air gap that drop should have been a few degrees at most.


I like that Ryzen IHSs are soldered on, you won't see problems like this there. Intel decided to save a few bucks on their already insane margins (in interest of fairness, they say the solder puts too much pressure on the core when it heats up, but this has rarely seemed to be a problem)

 

[Mineshaft_Gap]

Yeah, but it seems like Ryzen's ability to OC is pretty anemic so the soldered IHS is moot unless NOT soldering the IHS would have resulted in too high temps even at stock clocks and voltages.

 

[Mineshaft_Gap]

I bumped my VCORE up and tossed a 4.7Ghz OC on this pupper and these are my results after an hour in AIDA64, x264 stress test, and Realbench. I had to run the Realbench test twice since it bluescreened on me once. I bumped the VCORE up by a hair (.005V) and it passed. I ran Realbench w/ the GPU test because I wanted to see what would happen to the CPU temps with the GPU dumping heat into the case. As you can see, we still stayed in the mid 60s.

I cannot believe these results. Remember, I was getting 80°C @ 4.5Ghz and 1.335V almost immediately after starting AIDA64 which is not well known for being one of the harder stress tests to pass. Now we're sitting at a max of 66°C @ 4.7 and 1.415V.

So now my next question is... should I validate these settings overnight or should I push the OC a little further? 4.8Ghz wouldn't boot into windows at this voltage so I know I'd need to up the voltage by probably a not-insignificant amount. I have the headroom to do it but at 66°C I think that gives me enough headroom to use these settings year round.

Lowest temps are about 65°F/18°C which must be the ambient temps in the room. In the summer time I expect indoor temps in the 76°F/25°C range. I imagine that could add about about ~8°C to the core temps. I feel safe running in the mid 70°C range. If I push for a higher OC I think I'll need to scale it back during the summer months which seems like a pain in the ass. I dunno.

What do you guys think? Validate and keep these settings year round or push it harder and risk having to monkey around once it gets warmer?

 

[Mr_Moogle]

Noctua make some great fans but damn they need to change their colour scheme. Why on earth would anybody want beige and brown in their system.

 

[Plywood]

Scaling back sounds like a hassle, validate.

 

[Mineshaft_Gap]

Noctua make some great fans but damn they need to change their colour scheme. Why on earth would anybody want beige and brown in their system.

To be fair, you can spring for their industrial fans and get black and faster RPMs (and water/dust resistance). I don't mind because I associate the color scheme with quality. I look at my build and I see a solid, utilitarian power house.

 

[hockeypuck]

I recently delidded both my Skylake chips, one each of i5 and i7, using Rockit Cool's custom vice mold. Very simple and straightforward to use. Used the website's exact instructions. Didn't drop the temps during gaming as much as I expected, but it was a fun little weekend project.

FWIW, I couldn't increase my OC on the i5 after the delid. It doesn't stay stable above 4.5 GHz. The i7 I got to 4.7 GHz at 1.35V, but never tested it before the delid. But I'm very pleased with a 4.7 OC on a Noctua cooler in an mITX box.

 

[Mineshaft_Gap]

I recently delidded both my Skylake chips, one each of i5 and i7, using Rockit Cool's custom vice mold. Very simple and straightforward to use. Used the website's exact instructions. Didn't drop the temps during gaming as much as I expected, but it was a fun little weekend project.

FWIW, I couldn't increase my OC on the i5 after the delid. It doesn't stay stable above 4.5 GHz. The i7 I got to 4.7 GHz at 1.35V, but never tested it before the delid. But I'm very pleased with a 4.7 OC on a Noctua cooler in an mITX box.

What thermal compound did you use on the die after the delid?

 

[hockeypuck]

What thermal compound did you use on the die after the delid?

CoolLaboratory Liquid Ultra

 

[Mineshaft_Gap]

CoolLaboratory Liquid Ultra

Womp womp. Maybe you got decent chips then. It made a huge difference for me.

 

[Q8D3vil]

First, thanks for the thread and opening my eyes to the world of delidding.

Me and my buddy did our first delid a week ago and the results were mind blowing.

We both have 4790k and we were hovering around 80c in stress tests with corsair h110 (running in stock clock).After me and my buddy delidded our cpu's using rockit 88 (extremely easy if you are scared of the razor method + gives you the option to relid)+ CoolLaboratory Liquid Ultra the temp were down to 60c in stress.

I even overclocked my cpu to 4.9 on 1.36v (i probably had a good chip with a very bad intel paste job). The temp were 74c after 8 hours of overnight stress test using p95.
I highly highly recommend delidding.

 

[XiaNaphryz]

I won't use anything like Prime95 as I've noticed that gets dangerously hot regardless of anything you do. Maybe I'm doing it wrong but recent versions of Prime95 seem like literal CPU killers.

You're supposed to use older versions of Prime95 I believe. This is from a year ago, so it might be out of date but was the first article I recall reading about the issue at the time:

http://www.tomshardware.com/answers/id-2985195/4790k-hitting-100-degrees-load-seconds.html#r17568950

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

Core i 2nd through 6th Generation CPU's have AVX (Advanced Vector Extension) instruction sets. Recent versions of Prime95, such as 28.7, run AVX code on the Floating Point Unit (FPU) math coprocessor, which produces unrealistically high temperatures. The FPU test in the utility AIDA64 shows similar results.

Prime95 v26.6 produces temperatures on 3rd through 6th 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 significantly larger Die.

Run only Small FFT's for 10 minutes.

 

[Mineshaft_Gap]

First, thanks for the thread and opening my eyes to the world of delidding.

Me and my buddy did our first delid a week ago and the results were mind blowing.

We both have 4790k and we were hovering around 80c in stress tests with corsair h110 (running in stock clock).After me and my buddy delidded our cpu's using rockit 88 (extremely easy if you are scared of the razor method + gives you the option to relid)+ CoolLaboratory Liquid Ultra the temp were down to 60c in stress.

I even overclocked my cpu to 4.9 on 1.36v (i probably had a good chip with a very bad intel paste job). The temp were 74c after 8 hours of overnight stress test using p95.
I highly highly recommend delidding.

That's a better OC than what I can get with my skylake CPU. Congrats!