I7 4790K gets hot with Prime 95

The point is for worst case... That is why small fft was asked/what we normally run as it tests all cpu instruction sets.

When I run a Prime95 or the IntelBurnTest RealTemp says the temperatures are in the mid 90s. When I look at the Gigabyte Hardware monitor it says the temperature is in the high 70s low 80s. I'm running at 4.4 ghz. Could someone tell me what temperature is more accurate? Thanks.

The gigabyte is likely reading the socket temperature and Realtemp is reading the cores.

I have mine @ 4.7Ghz/1.25v/H100 cooler with two fans on med... LinX = 71 degrees max, Prime95 = 82 degrees max.

I have a 220x with a second 240 rad and prime still tops 80c. I just delided this morning.


Using CLU between die/IHS My load temps dropped 21c. Now prime [email protected] 66c
Intel's thermal material is crap. The only way to get great temps is deliding

Quest For Speed said:

I have a 220x with a second 240 rad and prime still tops 80c. I just delided this morning.
Using CLU between die/IHS My load temps dropped 21c. Now prime [email protected] 66c
Intel's thermal material is crap. The only way to get great temps is deliding

Click to expand...

Wow... that's an amazing drop in temps. Well done.

While I had the chip out I also lapped the IHS.

I was surprised how concave it was.

Quest For Speed said:

While I had the chip out I also lapped the IHS.
View attachment 154320
I was surprised how concave it was.

Click to expand...

Why not... I guess the wty is gone anyway, and that's a huge win.

4790k's are basically factory overclocked 4770k. And like previously mentioned, prime 95 28.5 small ffts temps will hit 100C even on custom water, even with vcore 1.25 range. It is normal temps for intels paste tim1 solution given dissipating 130+W (via stock settings running prime 28.5 small ffts), ie that much dissipated power usually requires solder tim given power density of intels modern cpus.

And like above, if you want normal temps, improving cooling wont help much, using better and thinner bondline tim1, ie liquid metal via delidding will solve the problem, will give similar temps to solder tim.

With my stock vid 1.19v (on gigabyte board anyways). But mine at stock on GB board (since will hit 44 on all cores) will set "stock vid" 1.19 and I was at high temps even on water at stock running prime 28.5 small ffts or linx. After delidding, like most, my temps dropped 22C, no more problems running prime 28.5.

4790k isnt that hot of a cpu, it is just intels tim1 paste solution was designed for tdp 90 watts and below. And for 4770k at stock it is fine. For an overclocked 4770k (ie 4790k) tdp is too high when running prime 28.5 small ffts, and needs solder or do it yourself delidding/liquid metal tim1 to run high tdp.

If I recall my stock VID was 1.12 and at stock the temps were hot but reasonable so I agree Intel's paste can handle 88W. The problem is Intel marketed the Devils Canyon processor to Enthusiasts as a overclocking chip but yet used a slightly improved paste instead of going with solder like the 5960. It seems that one department at intel is pinching pennies vs another wanting to produce what the end users wanted so we get this compromise.

I think they just want us to delid. They know that 30% will try razor method, scratch the pcb and screw up their memory controller and only be able to run in single channel. Intel believes they will most likely buy another one. Its sad to think of it this way but I believe that the people at Intel know this and are riding the delidding train right now.

Seriously?

Quite an assumption.
Yet somehow, I can see them doing it.

I wonder if Intel has money invested in Coollaboratory? They helped sales skyrocket for CLP and CLU. Seriously It's all about $$$. Making shareholders money.

De-lid it.

Lol, you people and your conspiracy theories...

When people say the PC is dead, they mean PC growth is stagnant, and lack of growth is death for a publicly traded company (at least its stock price). While desktop pc is still going to be around for a while, intel had to switch gears and design for efficiency for portables or become obsolete. In the interview with Devils canyon product manager, intel said they quit using solder on Haswell since (99.99% of use) doesnt require solder (low stock tdp), and they would not financially/practically do things to a few dies to use solder (referring to need to metalize the die with gold/other metal in fab process so indium solder will stick to silicone die) hence no chance of soldering DC. Although intel did within set cost contraints make some changes, which in end just weeded out haswells that couldnt meet DC specs. And they added since enthusiast platform is not designed for low tdp, they stated that platform will still have solder (requires solder with its high tdp).

But you dont need any conspiracy theories, intel was pretty frank. Intel deemed it too expensive to change fab process (metalize die and use solder) for very small percentage, though they were willing to change less expensive things. So if you want to rail on intel, just use the facts, they clearly explained it was cost.

And even if 1000 people bought new cpus from botched delids, that wouldnt register a blip on intels profit.

But using vice only method, Im pretty sure I could delid 10 in a row without any issue, pretty easy.

EarthDog said:

Lol, you people and your conspiracy theories...

Click to expand...

IKR

rge summed it up pretty well in post #57.

I will also add, this is what happens when there is lack of competition.