Thermal sensor issue i7-7700k?

[CK7]

I have a brand new build; EVERYTHING NEW. i7-7700k is running at stock speeds. I have the RAM set to XMP for DDR4-2666. Motherboard is Asus Maximus IX Hero Z270.

I have found that the i7-7700k reports a momentary (a second or less) temperature spike +25 > 35 degrees Celsius anytime a program is opened, a webpage is opened, a background app runs etc. The temperature blip cascades through the cores in random order; not the same every time. This causes my heatsink fan to constantly cycle up and down. Temperatures otherwise report as steady, normal increases. Peak temperature under Prime95 blend test is 71 degrees Celsius.

Attempted solutions:

I have re-installed my heatsink and thermal paste with no change.

I have tried to manually set my fan speed in the bios. The only setting that avoids this issue is setting the temperature / fan at a constant (and loud) 80-100%. I've tried PWM and DC mode.

I have found a few user reports elsewhere on the web, all reasoning that it's just the way it is. I don't accept that. Opening a folder or browser should not spike temps +30 degrees. Not only is the fan cycling annoying, it puts undue stress on my fan; possibly shortening its lifespan.

What's the answer, if any? RMA?

[idata]

I have same Asus Maximus IX Hero. Just check my video and you will be fine

And that other guy who have Corsair H115i you should be able to handle your prosessor with Quiet mode. At least I can do it with my Corsair H115i and its nice and quiet. Pump water never goes more than 35 deg.

[cmack2]

hi Timoxi,

thanks for your input...

i am extremely concerned about your CPU SVID Support being disabled - i beleive this setting actually screws with correct CPU Package Power readings...

if you could please use HWMonitor & run Prime95 for say 6 minutes (the AIDA benchmarks i have previously mentioned would be quicker & just as accurate) - i would be very interested in the CPU Package Wattage you report with SVID disabled...

i am very cautious before applying settings via UEFI BIOS...

cheers, craig

[idata]

Yes you are right if SVID is disabled it will fuck up cpu power readings. What does that mean? CPU not working with its full pwr or what happens?

First one with SVID off and second one SVID on.

AND wtf this means? Sometimes its 104 and sometimes it shows normal 100??

[idata]

And here is with 4.7 with SVID on. Again Realtemp shows different bus speed?

[idata]

Hi,

I have my CPU at 4.5GHz adaptive with 1.152v.

With Prime95 Small FFTs I'm getting up to 72C, which I guess isn't that bad. What puzzles me is that the Vcore max on HWInfo is showing a max of 1.184. However, the Vcore (current) is pretty steady at 1.136 and the Vcore (Max) doesn't go over 1.152 during the test.

The 1.184 tends to happen when the test is stopped. You can see this in the graph below (each of the Vcore spikes is on stopping the test). Seems like ringing?

On Prime95 large FFTs, Vcore is pretty steady at 1.168 (but again with the odd 1.184 spikes on start/stop).

Should the adaptive setting of 1.152 not prevent Vcore going above 1.152? The spikes could be ringing, but why the 1.168 on large FFTs?

[idata]

TGrable

I did just mess around in the BIOS, nothing did give me any advantage. Spikes and temps still out of control.

First try was SVID disabled, vcore manual at 1.185 -> The temps was higher then before, hitting 83ºC in less than 15 seconds.

Second SVID disabled, vcore manual at 1.150 -> Windows froze 20s after startup

Third SVID on auto, adaptive vcore @ 1.185 with a offset of 0.005 -> It says max vcore was 1,232 V

Screen of HW monitior at first attempt:

TGrable

I'm going to go back in the thread now and see if there's anything I've missed so far. Playing around in BIOS is only frustrating when I'm not sure what I'm actually doing.

[idata]

INTEL these are the numbers I like to see..wink wink..but you can get them only with underwolting...and spikes still stay even tho they are alot lower...

[TGrab1]

Honestly Disabling SVID is pretty much rarely needed.

As for temps being higher that is strange and probably has something to do with other settings. I would have to see more of your bios to know. Is the manual Voltage all that you changed?

[TGrab1]

ra5040 wrote:

Hi,

I have my CPU at 4.5GHz adaptive with 1.152v.

With Prime95 Small FFTs I'm getting up to 72C, which I guess isn't that bad. What puzzles me is that the Vcore max on HWInfo is showing a max of 1.184. However, the Vcore (current) is pretty steady at 1.136 and the Vcore (Max) doesn't go over 1.152 during the test.

The 1.184 tends to happen when the test is stopped. You can see this in the graph below (each of the Vcore spikes is on stopping the test). Seems like ringing?

On Prime95 large FFTs, Vcore is pretty steady at 1.168 (but again with the odd 1.184 spikes on start/stop).

Should the adaptive setting of 1.152 not prevent Vcore going above 1.152? The spikes could be ringing, but why the 1.168 on large FFTs?

The difference is the type of load and how Vdroop affects it.

Also setting your Adaptive at a level doesn't always cap it there as the motherboards is trying to provide that amount of voltage, but with the way it calculates Vdroop it can go higher or lower depending on the load. I am guessing your LLC is on level 1 which is fine, but just means that it adds a little voltage to compensate for Vdroop at full load.

As for voltage. Large FFT do not fit inside the cpu's cache so it has to store some of that information in ram. So it limits how fast the cpu can get that information and thus reduces the temps and volt draw on the cpu

Small FFT are all able to fix in the cpu cache meaning there is no bottleneck waiting on ram. So it produces much higher temps and voltage draw.

So you are just seeing the effects of Vdroop.

[idata]

The only thing so far that I've seen dropping the temperatures and spikes a little is if i use the AVX Offset, but then i also get a lower clock speed.

With the AVX Offest set to "5" i get 3.9ghz on the P95 small FFT test, was stable below 70ºC for 5 minutes.

AVX Offset "2" temps on picture below

[TGrab1]

Also I wanted to throw this out there.

Before I delidded my cpu

I did a nice little test to see where I was before and after. The after isn't what we are focusing on, but it allows me to show you where I was at stock clocks.

At about 22.5C ambient before delid I ran a 30 min stress test

My water temp went up to 27C and stabilized

Core 1 -78c

Core 2 -68c

Core 3 -71c

Core 4 -66c

So as you see my temps were around the 70 average mark. You can also see the huge difference between cores. This what lead me to the conclusion it was a TIM/silicon Consistency issue.

My point is that at stock clocks it is unfortunately par for the course to see 70+ C on P95 which stressed the chip harder than any of the others tests.

Now just for giggles sake I will post my after delid results.

For this test my ambient has risen a little due to time of day >.<

Ambient was 23C

Water temp went up to 27.8C

Core 1 - 53C 25C drop

Core 2 - 53C 15C drop

Core 3 - 51C 20C drop

Core 4 - 55C 11C drop

This was the same 30 min stress test and at a slightly higher ambient! Also this was with fresh TIM between the IHS and Water block which actually cures for better thermal performance. I would probably see a 2-3C improvement if I tested it now.

Another interesting bit of information. I can hit 5.0Ghz now at temps LOWER than my pre-delid stock temps. I think My temps equalize around 66-70C at 5ghz currently.

So in the end my point is that 70C on stock clocks with these chips seems to be par for the course.

I will also throw out that I originally had a h240X when I got this cpu, was not happy with temps and figured it was just not enough cooling power at first. So I added another 240mm radiator. I gained maybe 1C. Knowing I would be putting my GPU under water soon I figured wth and added a 3rd 360mm Rad to the loop with a second pump to really push some water. I actually didnt gain anything and now had 840mm of Rad space.

Now I have a very nice custom loop to work with, but decided the issue MUST be with the CPU. I was as shocked as most people when I first realized the chip was performing that poorly at transferring heat to my water block.

Anyways I have rambled on. The point is if you have bad temps RMA it or risk a Delid, but there are very few other options for getting these things under 65C at even stock voltage and clocks on p95.

[idata]

Thanks, that's helping me decide what to do next.

I'm actually thinking of returning it and ask for a new one. If the new one isn't showing better performance I'm gonna delid it.

How would you advise me to stresstest the CPU?

Right now I'm running the P95 Small FFT stress test, and im not sure how long or even if its the correct test to actually test the CPU with.

Ambient temp now is 23,5C and max temp after 10 minutes is 82ºC.

And is it normal for the clock to be at 4.2GHz during the test?

[idata]

That makes sense ... thanks!

[idata]

The one thing that seems to make quite a difference to my setup is having the voltage at Manual/1.155, Clock at 4.5GHz. With Prime95 Small FFT at high prority, AIDA84 stress test, memtest86 Pro running multiple tests, and a video playing I'm getting a solid 64C max and 53C average with only a 3C max difference between the cores. Vcore is solid at 1.136 with very occasional spikes to 1.152.

I'm using a Kraken X62 water cooler and the water temperature is 32C (ambient about 25C). At that temperature there is essentially no fan sound.

I'm not really interested in overclocking because even if I pushed it to 5GHz all I would be getting is an extra 10%, which I probably wouldn't notice.

Adaptive just doesn't seem so good because of the effect of VDroop, as TGrable explained. The question is ... is it OK to have the CPU running on a minimum 1.155 constantly???

Cheers

Robert

[idata]

I'm struggling to understand what you're even trying to achieve. Honestly, it seems like people are complaining over nothing in most cases. If your temps aren't always in the 80's, who cares? You clearly have more voltage room to use if you need it, if you don't and your PC is stable, leave it.

[ICitr]

Ritchie, it would be understandable if you haven't read through the entire 20 page thread. I think you're missing the main point. The problem is with sudden, very short and very high temperature spikes for doing stuff as mundane as opening Notepad. The immediate effect is that a system's fans are constantly cycling up and down.

Because of Intel's almost complete silence, people have been speculating as to what's wrong and have been hobbyshopping solutions. As such, the thread kind of turned into an overclocking symposium when perhaps a new thread ought to have been started. But, the real purpose of the thread is the temperature spiking.

[idata]

Well, what I'm trying to achieve is a stable system running at a reasonable temperature. At default settings it's stable but the temperatures go well up into the high 70s and 80s running Prime95, for example, and with lots of spikes when running normal applications.

This is why I've been experimenting ... and as I mentioned in my last post, the one setting that seems good is the Manual setting at a low voltage. I'm happy with that .... providing it's not bad to have the CPU running on a constant voltage. I don't know the answer ... although I would imagine that it isn't the best in the long-term. As I would like to get 6+ years out of my CPU that's a concern to me.

Robert

[idata]

I've read the entire thread, I made my own before, too.

Make a custom fan curve, if its a problem - but my point is, I am not getting ANY spikes past 45 really.

It's not a surprise that every one who I've recommended a 7700K + Z270 MSI Pro Gaming Carbon has no issues. It's 90% motherboard related.

[idata]

ra5040 - what version of prime95 are you running first of all?

Secondly, people need to stop thinking prime95 is an actual acceptable indicator (especially the newer versions) of stress testing temperatures on a PC.

As for the voltage scenario. I can only go from my experience and every PC I've built for my friends.

My Q6600 has been at a constant 1.45v for 9-10 years, clocked at 3.6 GHz.

My 2500K has been at a constant 1.38v for 6 years, clocked at 4.3-4.5GHz (and on almost everyday, and overnight for the entire duration, except vacation). This CPU also ran at 100 degrees for over a month due to my fan being broken, but I'm yolo.

My 7700K is at a constant 1.328v clocked at 5GHz.

My friends 2500K has been at a constant 1.42v for 6 years, under the same conditions as mine, clocked at 5GHz, and also ran temporarily (over a month) at high temperatures without him knowing until I fixed it.

My grandmothers E8400 has been at a constant whopping 1.48V (past it's safe limit I believe) for 10 years clocked at 4.4GHz.

See my point?

Look, I get it, it's obviously a concern after spending $xx amount of money. I'm not telling you to disregard it. I'm telling you that you're worrying wayyyyy too much.

I also recommend to anyone who is going to purchase a 7700K to use the MSI Z270 Pro Gaming Carbon.

Ritchie

[TGrab1]

I have to disagree that it is board related.

Some boards were having issues with auto voltage being too high, but bios updates fixed most of that.

If you go with a solid name manufacturer though then the board should all be very similar minus certain feature sets etc.

I personally use an ASUS board and have had no issues with it. So I will say though that if you are running a board on auto.. make sure you get the newest bios updates for it.

[idata]

Disagree all you want, anyone with a Z270 Gaming Pro Carbon has no issues.