4790K Temps too high?

[smithg97]

My 4790k (clocked to 4.5ghz) temps have gone up since swapping my GPU (did have fury x so a lot of the heat was being thrown straight out of the case by the liquid cooling of the card) to Nvidia 1080Ti.

Currently my idle temps are between 42C to 46C and in games my temps are exceeding 90C after a relatively short amount of time.

I have a fractal design R4 case with 2 Noctua NF-A14 fans for inlet at the front and 2 more at the top of the case as exhaust. Also have a Corsair H80i AIO with 2 Noctua NF-F12 fans mounted to the radiator, all fans are definitely facing in the right direction

Have tried reapplying thermal paste and reseating the cooler which hasn't really show much any improvement.

My system fans are set to smart mode in BIOS so that at 60C the fans run at 50% / 70C fans run at 60% / 80C fans run at 75% / 90C fans run at 80% I also have a similar fan profile for my CPU fans.

Any suggestions would be greatly appreciated

 

[Flyfisherman]

Hi,
The Tcase temperature of the i7-4790 is 72°C. The processor supports up to 100 Celsius degrees for a small period of time.
90C i way to high. It will start throttling and lower its cpu core frequencies.
I have also a 4790K OC @ 4,6GHz and rarely it exceeds 56C at full load with my Noctua NH C-14 and Antec P193 V3 chassis.

It sounds to me that your cooler is not properly mounted, meaning that the cooler plate do not have a firm contact to the cpu top plate.

Best regards from Sweden

 

[smithg97]

Thanks for your response Flyfisherman.

Am i correct in thinking that if the mount is bad that the face of the cooling block will feel cool even after the CPU temp remains at 90C+ for sometime?

 

[Flyfisherman]

Perhaps, the cooling block would be somewhat warmer due to the heat transfer and not feel cold at all, that suggests that it has a poor contact to the cpu top plate?

90 degrees C is way out of the question, it is to hot.

Best regards from Sweden

 

[smithg97]

I've just reseated the cooling block and run fire strike ultra, I can see that the CPU temp reached 87C when at 92% usage (per MSI afterburner). After running the benchmark i removed the side panel and can feel that the AIO water pipes, rad and water block all feel hot to the touch, if the mount is bad surely the pipes wouldn't get so hot as they wouldn't be transferring the heat away from the CPU?

 

[Flyfisherman]

Well MSI afterburner could indeed raise the numbers, but it don't seems right.
How is the temp when running a game or other heavy cpu task at full power for a couple of minutes?

Close MSI afterburner and restart Windows.

To find out if it has some strange temperatures or voltage: download and install HWMonitor (free) http://www.cpuid.com/softwares/hwmonitor.html
To the left download the free version.
Obs! During installation make sure that you untick any third part s/w. I always recommend this procedure no matter what on any free s/w. Ok.

Then, in the HWmonitor window there are three columns:
Value= shows the actual readings from all the sensors including temperatures and voltage
Min and max speaks for itself.

Now, Start the most cpu demanding program you have and work with it ie. start a heavy game for a couple of minutes.
Then quit these program and take a reading from HWmonitor, which of course you have been running in the background:
In the column Max: what does it says?

Next step is to reset Your CPU overclocking to its default values in BIOS and check again.

If the temps is still way off, then something is wrong, perhaps with the cooler block, the pump or else?

You have a Water cooler Corsair H80i, right?
Make sure the pump is working correctly and that the power are connected properly.
If this is the correct one, check out the installation guide under FAQ and Support http://www.corsair.com/en-gb/hydro-series-h80i-high-performance-liquid-cpu-cooler

Otherwise perhaps you need to do a RMA of that cooling system, because something is not working correct?
Best regards from Sweden

 

[smithg97]

Well i installed HWmonitor, ran witcher 3 on max settings @ 1440p for about 10 minutes then checked HWmonitor and saw that the max temp was 80C, i then restarted, loaded BIOS and reset the CPU overclock back to stock, loaded HWmonitor and once again played Witcher 3 for about 10 minutes, HWmonitor now says 79C max CPU temp.

FYI the ambient temperature in my room is about 25C and the max temp shown in HWmonitor for my GPU is 69C

When i had the fury x installed my CPU temp ceiling was about 70C (and that reading came from MSI afterburner which as i mentioned earlier is now showing 90C+ with the 1080TI installed)

I'm pretty confident that all of the power connections are fine, when i look in Corsair link software it states that H80i pump is operating at 2218RPM if that makes sense :-S

 

[smithg97]

Just to add to my previous comment, the H80i i have is about 3 years old.

 

[Flyfisherman]

the H80i i have is about 3 years old.

Maybe, its system has lack of cooling media, just a wild guess here?
I know, it sounds ridiculous but what else could be the problem, since You seems to have tried almost everything?

If the cooling plate is uneven that will cause this behavior to, because it will not have a firm contact over the cpu top plate on all of its surface.
I still bet that there are a problem with your cooling system, one way or another.

Perhaps other member will respond to this thread because I'm all out for tips for now.
Best regards from Sweden

[Edit]
Perhaps there is a problem with the 4790K CPU itself?
Inside the package there are a certaian thermal compound on the die itself (the processor dies) attached to the processors top cooling plate from within.
If this somehow doesn't work, neither will the external cooling do.

 

[amtseung]

Pump speed looks acceptable. They usually function at around 2400rpm or thereabouts.

You have an H80i, which is on the lower end of cooling capabilities for AIO's in general. My H100i couldn't keep my 4790k at 4.6ghz 1.35V cool under full load, and would thermal throttle within 5 minutes. I had to dial VCORE wayyyyyy back to 1.15V and run it at 4.4 ghz to see acceptable load temps of 73C after an hour of P95 small fft's. I'm seriously contemplating delidding it now.

My H100i is one of the older variety, with the square plastic shell, and the steel bracket that sticks to the pump/block with magnets. I discovered one day that the bracket had warped over time, so that even cranking down on the 4 thumb screws all the way yielded horrible contact with the CPU, and sometimes, no contact with the CPU at all. Had to shim the bracket with popsicle sticks to make good contact. If your H80i is 3 years old, you may be facing the same problem I did with my 3 year old H100i.

Another thing is to listen for gargling noises from the loop when you boot up. If there are, there's enough air in the loop to prevent the thing from functioning properly. It's time to retire it or RMA it. 3 years is about the average lifespan of these first gen HXXXi AIO's anyway, or so it seems.

 

[smithg97]

Unfortunately my case and mobo combination means that i can't install any of the NZXT 240mm radiators in the top of the case otherwise i would upgrade to the Kraken series of AIO.

Any thoughts on the best AIO with 120mm rad?

I'm torn between upgrading now or waiting to see whether AMD threadripper or the new I9 CPUs are worth the purchase (for primarily gaming use) as i wouldn't want to fork out for a new cooler now then need to buy a new one because of different socket.

 

[smithg97]

So, at the weekend i replaced the thermal paste on the AIO and reseated, the temps remained the same as they were all of yesterday whilst gaming (approx 90C) looking again today my idle temps are around 31/32C and gaming in Witcher 3 for about the last hour and my CPU hasn't exceeded 72C so a massive improvement. My guess is that the newly applied paste has cured properly and doing the job

 

[smithg97]

Temps have gone back to 42ish at idle and mid-80's at load.

I'm beginning to think it is due to having the OC Genie option activated in my BIOS settings and is causing the CPU to constantly run at 4.4ghz and probably at an unnecessarily high voltage.

I have tried to change this in BIOS by reverting back to a manual overclock but the CPU multiplier option is greyed out and won't let me change it (with or without the OC Genie option activated).

 

[amtseung]

Clear CMOS and try again. I can't tell you how to do it for your particular motherboard, but there are typically two ways to accomplish this: 1.) short the jumpers of the JBAT1 pins with something, or
2.) yank out the CMOS battery for about 5 minutes, then stick it back in.

Clearing CMOS is forcibly resetting the bios ROMs.

OC Genie is never a good thing in my book. Manual OC or don't OC at all IMO. Since you say OC Genie, which is typically an MSI thing, then I assume you have an MSI motherboard. My MSI Z97 gaming 5 likes to auto voltage my poor 4790k at 1.4V, which is WAAAAYYYYYYYYYYYYYYYYY more than it needs to run at any speed. My chip does 4.4ghz at 1.15V, 4.5ghz at 1.2V, 4.6ghz at 1.25V, and 4.7 at 1.35V, where I then hit a VCORE wall and would probably need 1.5V to reach 4.8ghz. Actually, I've never hit 4.8 stable, so I wouldn't know. Of course, obligatory YMMV because silicon lottery. My personal rig info is in my sig. I believe it's up to date.

Try something: while you're running something reasonably CPU-heavy, try push down on the pump/block pretty firmly. If doing so makes your temps drop, then you've got a contact issue, and you may need to find a way to shim it to regain proper mounting pressure, or buy new mounting brackets/hardware from Corsair. If it doesn't change your temps, then something else is wrong.

Oh, and idle temps aren't all that useful of a measurement when water cooling. Load temps are though.

 

[smithg97]

I haven't yet tried to clear CMOS but i did remove the OC Genie option and manually overclocked CPU to 4.5ghz at 1.2V and playing Mass Effect andromeda at 1440p with everything maxed, the CPU seems to sit at around 78C with very occasional and brief spikes up to mid 80C.

I'll happily take that as i do have all my fans running at fairly low RPM in order to run quiet so i guess i can't have my cake and eat it too! Plus the CPU is fairly old and may soon get swapped out for an i9 or Thread ripper depending on game performance.

 

[smithg97]

So to give an update, i tried clearing CMOS which didn't have any effect on temps, unmounted my aio and cleaned out the rad, wiped down water block and the cpu and reapplied the thermal paste sparingly, temps still remained around 45C at idle but when i ran the cpu test in cinebench my cpu temp in hwmonitor almost immediately rose to 100C (and that is with the side panel off). I tried pushing down on the water block whilst running cinebench which reduced temp by only about 5/6C.

I've decided to buy the h80i v2 and will install at the weekend, if that doesn't work would it be safe to say that my cpu is on it's way out? Or is it possible that it could be a mobo problem?

Also might be worth noting that when i checked my temps in corsair link at the time my cpu was at about 45% utilisation, my cpu temp reading was 75C but the h80i temp was only about 50C, the water pipes and block felt hot to the touch as did the back of the case where the radiator and exhaust fans are located

 

[SinxarKnights]

So what are your temps like at stock?

If you push on the block and temps drop, you have a mounting/contact issue. Not saying you did it wrong, but the mounting kit could be damaged in some way.

 

[smithg97]

At stock i would still get on the bad side of 90C whilst gaming.

I can't see any signs of physical damage on the mounting bracket or the water block, it all looks like everything is in the right place.

I'm going to try the h80i v2 this weekend and if that doesn't improve the temps then either the cpu or mobo has a problem so hopefully still under warranty (bought about 2 years ago)

 

[amtseung]

I'd skip the thicker 120mm rads and go straight for the 240mm h100i v2 (25mm thick) or gtx (30mm thick) or the 280mm h115. Way better thermal overhead for not that much more cost, especially if you can source a factory refurb unit.

After fixing my own mounting problems by shimming the top-side bracket with shreds of old business card paper stock, I have idle temps of 22-26C (EIST and Cstates are on, so about 0.1Vcore at idle, locked multiplier of 48 though) and load temps at 1.36Vcore of 80C~ish. That 48 multi was just for benching/torture testing, I'm dailying it at 4.7ghz at 1.3V, which maxes out load temps at 73C. Of course, this is with my frankenstein H100i-based loop and gentle typhoons instead of those garbage corsair fans, front mounted to an NZXT S340 with the front panel tossed in the bin long ago. Picture below.

 

[SinxarKnights]

At stock i would still get on the bad side of 90C whilst gaming.

I can't see any signs of physical damage on the mounting bracket or the water block, it all looks like everything is in the right place.

I'm going to try the h80i v2 this weekend and if that doesn't improve the temps then either the cpu or mobo has a problem so hopefully still under warranty (bought about 2 years ago)

Well not necessarily. In fact I would go back to the stock cooler and reset the cpu to stock clocks and voltages to rule out the CPU being at fault. It seems like the cooling solution is broken, not the CPU/MB.

I'd skip the thicker 120mm rads and go straight for the 240mm h100i v2 (25mm thick) or gtx (30mm thick) or the 280mm h115. Way better thermal overhead for not that much more cost, especially if you can source a factory refurb unit.

After fixing my own mounting problems by shimming the top-side bracket with shreds of old business card paper stock, I have idle temps of 22-26C (EIST and Cstates are on, so about 0.1Vcore at idle, locked multiplier of 48 though) and load temps at 1.36Vcore of 80C~ish. That 48 multi was just for benching/torture testing, I'm dailying it at 4.7ghz at 1.3V, which maxes out load temps at 73C. Of course, this is with my frankenstein H100i-based loop and gentle typhoons instead of those garbage corsair fans, front mounted to an NZXT S340 with the front panel tossed in the bin long ago.]

I get similar temps with with my h115i. Maxed it out at 4.95ghz @ 1.35. I keep it on quiet pump with the included fans at low RPM. With my cooler, i've found that increasing the pump RPM and fan RPM does not make much of a difference.

forgot to add, i use a 4790k with an Asus z97-e.

 

[amtseung]

I get similar temps with with my h115i. Maxed it out at 4.95ghz @ 1.35. I keep it on quiet pump with the included fans at low RPM. With my cooler, i've found that increasing the pump RPM and fan RPM does not make much of a difference.

forgot to add, i use a 4790k with an Asus z97-e.

Lucky you and the silicon lottery. My chip won't run at 4.8ghz for more than 10 minutes without BSOD-ing, regardless of voltage or ram loosening. 4.7 is the highest I can get it completely stable (overnight P95 small fft). Push the bclk up one notch over 100, and the whole system just instantly fails, gotta clear CMOS to bring it back.

Glad to get some verification that my own temps are back to normal.

Sorry for stealing the thread. Carry on...

 

[SinxarKnights]

I get similar temps with with my h115i. Maxed it out at 4.95ghz @ 1.35. I keep it on quiet pump with the included fans at low RPM. With my cooler, i've found that increasing the pump RPM and fan RPM does not make much of a difference.

forgot to add, i use a 4790k with an Asus z97-e.

Lucky you and the silicon lottery. My chip won't run at 4.8ghz for more than 10 minutes without BSOD-ing, regardless of voltage or ram loosening. 4.7 is the highest I can get it completely stable (overnight P95 small fft). Push the bclk up one notch over 100, and the whole system just instantly fails, gotta clear CMOS to bring it back.

Glad to get some verification that my own temps are back to normal.

Sorry for stealing the thread. Carry on...

Yeah I know right. For a long time I thought I got a bad chip until I started asking here and apparently 4.7 is uncommon and 4.8 is almost unheard of. I thought it was bad because I couldn't gain that last 0.05. I eventually did and was able to benchmark it but that was about it.

 

[smithg97]

The only problem with getting a 240mm setup is i would have to remove the HDD cage and mount the rad on the front of the case which i'm not to keen on.

The new AIO is being delivered today so fingers crossed it does the trick.

Will report back once installed

 

[smithg97]

So i've installed the new AIO which was a very stressful experience lol however it seems to have paid off! With the side panels on i just ran a quick CPU test in cinebench and the HWMonitor reading for the CPU package peaked at 61C with the cores sitting between 56C and 60C so a massive improvement!

Will test on some games tomorrow and report back the temps but so far so good

 

[Flyfisherman]

So i've installed the new AIO which was a very stressful experience lol however it seems to have paid off! With the side panels on i just ran a quick CPU test in cinebench and the HWMonitor reading for the CPU package peaked at 61C with the cores sitting between 56C and 60C so a massive improvement!

Will test on some games tomorrow and report back the temps but so far so good

Haha, Well done sir. That was indeed good news and not fake news... ehhh..
Best regards from Sweden