Z97-A Asus Mobo, i4-4690k CPU - Overheated Suddenly on Test 4 W/ Prime95

Hellrott · Oct 29, 2015

Hey guys,

I have a problem here. I upgraded a LOT on my computer in order to support crossfired R 380s including switching to a better mobo and upgrading to the above mentioned i5-4690k processor.

I clearly had a much better affinity for AMD overclocking with MSI because I am having nothing issue after issue with this one. I used the "Wizard" tool on BIOS and it determined that 4488 MHz was the appropriate OC setting. Fine, great in fact. So I start my test, which I'll include a picture of below, and it's all going great. 60 degrees, tests ramp up and it goes between 70-75. Well within where I'd like it to be during a stress test.

Then, suddenly, I'm seeing 100 degrees c on my cores so I immediately shut down the test.

http://imgur.com/zFZFPu2

Here is the aftermath.

Hopefully no damage was done, I cancelled it pretty quickly but I'm confused on what the issue is. Why the sudden jump? Prior to this I actually very slightly loosened my Evo 212 screws because I thought it may have been too tight and it seemed to have helped then suddenly this jump happened.

Anyone have any idea what happened?

Darkbreeze · Oct 29, 2015

First of all, 60°C is not a good temp at idle, but I'm not seeing that on your HWmonitor screen. I see about 36°C, which is a little bit high for idle, but not bad.

Two, I don't recommend using HWmonitor. It's notoriously inaccurate for some people. Try using HWinfo or CoreTemp.

Three, what version of Prime95 are you running? If you are running ANY other version than 26.6, uninstall it, and install version 26.6 here:

http://windows-downloads-center.blogspot.com/2011/04/prime95-266.html

Newer versions of Prime run AVX instructions which does not provide a steady state load and does not provide a realistic thermal example. Unrealistically high temps will be observed when running AVX instructions.

Yes, regardless of architecture. P95 v26.6 works equally well across all platforms. Steady-state is the key. How can anyone extrapolate accurate Core temperatures from workloads that fluctuate like a bad day on the Stock Market?

I'm aware of 5 utilities with steady-state workloads. In order of load level they are:

(1) P95 v26.6 - Small FFT's
(2) HeavyLoad - Stress CPU
(3) FurMark - CPU Burner
(4) Intel Processor Diagnostic Tool - CPU Load
(5) AIDA64 - Tools - System Stability Test - Stress CPU

AIDA64's Stress CPU fails to load any overcloked / ovevolted CPU to get anywhere TDP, and is therefore useless, except for giving naive users a sense of false security because their temps are so low.

HeavyLoad is the closest alternative. Temps and watts are within 3% of Small FFT's.

Even with all that, I'd say there's a good chance if your temps shoot to 100°C instantly under any of the testing, you may have used too much, too little (Thermal paste), or do not have the cooler correctly mounted and mounting pressure is not right.

Also, be sure to ONLY run Small FFT when thermal testing. Do not run Large FFT or Blend.

As to your cooler, take a look at this. It's possible the center screw has caught or is not centered or fully tightended:

http://www.tomshardware.com/faq/id-2520482/solving-temperature-issues-hyper-212-evo.html

Hellrott · Oct 29, 2015

Idle temps are closer to 30 degrees, sorry if I was misleading with my text. Let me attempt a test with the instructions you've given and I'll screenshot the results

Hellrott · Oct 30, 2015

http://imgur.com/oCN1WDz

Here is an image from the most recent test with your programs. Seems MUCH more stable now.

Darkbreeze · Oct 30, 2015

Are you using a utility or automatic feature of the bios, like an EZ overclock, or are you manually configuring the overclock?

Hellrott · Oct 30, 2015

EZ Overclock, I'd tried other methods and failed with stability and so far this seems to be working

Darkbreeze · Oct 30, 2015

Manually overclocking is far more efficient and stable. But if you're happy with using one of the automatic methods, you'll just have to settle for what you can get that way. The main problem with the automatic overclocking methods is that it generally over compensates with voltage to ensure stability, at the cost of increased heat, so you end up having to limit the extent of your overclock to remain within your thermal tolerance.

What are your full system specs? BTW, I'd retighten your heatsink base. Just loosen it up a bit on each fastener, then retighten it until the bottom bracket standoffs just begin to turn along with the screws. The instruction in the tutorial linked to in my tutorial are pretty specific, so it should be easy enough to get right but takes a bit of practice to get the hang of.

Hellrott · Oct 30, 2015

First of all, love your signature "Resistance is futile (If <1 ohm", that's great.

Second, I followed a step by step guide last night and I *think* I set my max voltage to either 1.25 or 1.3 and successfully overclocked to 4.5GHz.

My rig is:
Evo 212 Cooler
Z97-a mobo
Cross fired Sapphire R 380 and Gigabyte 380
i5-4860k
850w EVGA 220-G2-0850-XR 80 PLUS GOLD Crossfire Support
150SSD and 500GB HD
x4 1600 Corsair 4GB RAM

Hellrott · Oct 30, 2015

And by successfully I mean I stress tested it for about 40 minutes and had no issues, still plan to run a longer test to ensure stability.

Darkbreeze · Oct 30, 2015

There's a big difference between 1.25 and 1.3. Like night and day when it comes to heat and stability. What I'd recommend doing if you're going to overclock is that you disable Intel speed step, set your Processor power management setting in the advanced section of the control panel power applet to 100% for both min and max settings, or just change the entire power profile to performance and then change a few bios settings as well. RAM voltage should probably be changed from auto to whatever is specified on it's spec sheet unless you've had problems with it and had to bump the voltage up for stability, in which case, leave it. The CPU voltage should be adjusted incrementally.

Stability testing should be done over a 15-20 HOUR period, not 40 minutes. 40 minutes is more than enough to determine thermal limits during each step of the tuning process because really that only takes about ten minutes on small FFT, but stability testing can't be determined without much longer testing as I've seen many systems stop workers or begin to error out at points all the way up to 15hrs and beyond. 15-24 hrs is acceptable variance though. Any system that doesn't error out or stop workers by 24 hrs is unlikely to allow those small errors to creep into the system and corrupt data which is what can happen with a slightly unstable overclock. A system doesn't have to "crash" for there to be a problem happening. Slightly unstable overclocks can introduce small errors into the system over time that will eventually entirely corrupt the operating system, files, folders, settings and any other data that's relevant or has been processed at any point by the system until it becomes unusable and unsalvageable.

This is easily avoided by determining stability with a nice long 15-24hr run on Small FFTs using version 26.6.

Here is the best general overclocking guide I've come across:

http://www.overclock.net/t/91/ultimate-overclocking-guide

And this one applies specifically to your configuration:

http://www.overclock.net/t/1411077/haswell-overclocking-guide-with-statistics

Also, I think you typo'd your CPU model. There is no 4860k.

Darkbreeze · Oct 30, 2015

Additional things I'd seriously consider.

1. Get rid of the 4 x4gb RAM and get 2 x8gb. The more modules you install, the more voltage that's required (1.5v x2 =3v. 1.5v x4=6v) and the more stress that's placed on the memory controller, which is IN the CPU and will affect all manner of functionality including stability while overclocking.

2. If you do change the memory, consider jumping up to 1866 or 2133mhz since your configuration will handle it and the prices have become fairly similar anyhow. You can always sell your 4 1600mhz modules to somebody else or use them in another system.

3. What is your case model, how many case fans are installed and what direction is each fan oriented in it's configuration, as in, blowing out (exhaust), sucking in (intake)? With an overclocked system, a relatively low end aftermarket cooler and crossfired graphics cards, the amount of airflow through the case becomes absolutely critical.

Hellrott · Oct 30, 2015

http://imgur.com/9KXsgAI

Here's the current rig setup from before changing mobos/CPUs. Just imagine there is another 380 under that one and you'll have an idea of what I'm working with.

I know the MAX voltage I want is 1.3 and that's pushing it, so I am pretty certain I had it set to 1.25. At work so I can't check right now.

Definitely going to take advantage of your advice with the RAM though, I need everything optimized for a stable OC'ed CPU.

And thank you for all this, you're a great help

Hellrott · Oct 30, 2015

Red arrows indicate which way the air is flowing, by the way. Once I get the 2 stick set-up I'll be able to push air through the evo cooler instead of pulling it which I think will help a good amount.

If I do that, where should I move that top fan? Or should I just leave it where it is?

Darkbreeze · Oct 30, 2015

If it were ME, I'd just mount the cooler higher on the heatsink, which you CAN do, to clear the RAM. But for now it might be ok. You might not gain that advantage even with two sticks, as I believe the population rules for that board dictate that the two slots closest to the cooler should be used for two sticks anyhow. I'll double check that though.

Honestly, I'd consider a different cooler, for certain. The EVO is pretty low end as aftermarket coolers go. You can get a much better cooler, with ZERO RAM clearance issues, for about 46 bucks.

http://www.newegg.com/Product/Product.aspx?Item=9SIA4UF30R0134

This would be even better, with maybe 15% better cooling performance than the H5, but it's also more expensive:

http://www.outletpc.com/ur7144-noctua-140mm-sso2-bearing-cpu-heatsink.html?utm_source=ur7144-noctua-140mm-sso2-bearing-cpu-heatsink&utm_medium=shopping%2Bengine&utm_campaign=pcpartpicker&utm_content=Noctua%2B-%2BFans%20%28Cooling%29%20%3E%20CPU%20Fans%20%26%20Heatsinks%20%3E%20Intel%20Processor%20Coolers

Darkbreeze · Oct 30, 2015

You definitely need to majorly change things on the case cooling, probably with a different case. Until then, I'd switch that top fan to exhaust. Top and rear should always be exhaust and front and bottom should always be intake. That fan blowing down will interfere with the flow of cool air from the front to the CPU cooler, instead redirecting warm air that has risen to the top of the case into the CPU cooler. This is what all tower systems that do not have a top mounted power supply should look like.

Hellrott · Oct 30, 2015

Huh, I'd only ever heard great things about the evo, didn't realize it wasn't as good as the hype made it out to be.

You're the expert here, I will do whatever you think I should do to optimize the computer. Any other recommendations? I plan to pick up the parts and get right to it once I'm off work

Hellrott · Oct 30, 2015

Do you think where I have ziptied the other front case fan is doing what I hope it is where i have it? I felt like there was a lot of resistance putting in front of the hard drive slots and that a more direct path to the GPU would be best

Darkbreeze · Oct 30, 2015

Updated above thread.

Darkbreeze · Oct 30, 2015

What is your case model? And that looks like an awfully small fan up front, and probably NOT in the correct place. Usually you need to remove the front panel or the drive cages, while installing it, then put them back, unless that is a crappy case that wasn't intended to have fans in that location.

Hellrott · Oct 30, 2015

http://www.corsair.com/en-us/carbide-series-spec-01-red-led-mid-tower-gaming-case

Hellrott · Oct 30, 2015

No, it's a 120MM fan that was mounted by taking the front face plate off. That's where I original had the fan that is now ziptied to the harddrive mounter.

Darkbreeze · Oct 30, 2015

Ok, so that case supports 2 front 120mm fans, and that's what I'd use, in the location where they belong. NOT zip tied behind the framework. Mounted in FRONT of the framework, they will suck in ONLY cool outside air. Mounted where it is, it is likely to draw at least some of it's air from inside the case, which is already heated air.

That case should easily support two top fans as well, so I don't know why it looks like the EVO is interfering with the top rear fan location in your picture, but if you can get a fan in there, I'd move your front top fan there as an exhaust. If not, then I'd flip the top front fan to exhaust and leave it where it is.

Optimally, I'd highly recommend moving that front 120mm to the top, so you'd have two top 120mm fans in exhaust mode, and get two additional 140mm fans to mount up front as intake. The rear fan of course is exhaust. That would probably drop your temps by another 10°C, at least. You may not even need a different cooler if you reconfigure the case cooling as I've outlined, but it wouldn't hurt to change that as well. I'd seriously reconfigure the case cooling before worrying about the CPU cooler though as the EVO can do the job, just not as well as some other models. What it CAN'T do, is do it's job if the airflow through the case is out of whack.

Darkbreeze · Oct 30, 2015

Also, having good fans is just as important as having fans at all. I'll be glad to work with you on that, but it's unlikely you're going to find GOOD fans locally, unless you live in a large city that has a lot of tech vendors and a wide selection. Since your board supports 4 pin PWM fans, you can use those OR 3 pin voltage regulated fans. PWM is better, but more expensive. When it comes to fans, brand DOES matter in most cases. Antec, Cooler Master and Corsair should be, for the most part, avoided. They have some good models, but are too proud of them. You can get better fans for less than comparable quality models by those brands.

And, your motherboard has four 4 pin chassis fan headers and two cpu fan headers. I'd suggest connecting your rear case fan to the CPU_OPT fan header, so it will speed up and slow down in direct proportion to your CPU temps, just as the CPU cooler fan does, rather than in response to the motherboard thermal sensor which reacts much more slowly. This will be somewhat louder to some degree, but will offer much better cooling performance. You can take your pick on that as to whether cooling performance or noise levels are the more important issue.

Hellrott · Oct 30, 2015

That all makes sense. I'll definitely make those modifications and pick up a few fans on the way home. Do you think sticking to the 4x4GB set-up for RAM is ok for now or should I make that a priority to change that too?

Hellrott · Oct 30, 2015

I live in Phoenix so I have access to any store I need haha. What would you recommend for a fan that's not too outrageously priced? I've pretty heavily invested in this computer already and want to mitigate spending more as much as possible. That said, I've spent a lot of money already and want it to work properly

Edit: My computer already sounds like it's trying to take flight when things start to heat up so I'm more worried about effective cooling

Z97-A Asus Mobo, i4-4690k CPU - Overheated Suddenly on Test 4 W/ Prime95

Honorable

Retired Mod

Honorable

Honorable

Retired Mod

Honorable

Retired Mod

Honorable

Honorable

Retired Mod

Retired Mod

Honorable

Honorable

Retired Mod

Retired Mod

Honorable

Honorable

Retired Mod

Retired Mod

Honorable

Honorable

Retired Mod

Retired Mod

Honorable

Honorable

Share this page