Help! Shut off during prime95, 4790k on H100i @ stock clock

[bobbyzzx]

My build:

CPU: i7 4790k @ 4.0 turbo 4.4
CPU Cooler: H100i
Mobo: Asus Maximus VII Gene
Memory: 2x G.Skill Trident X 2400mhz 8GB
GPU: 2x Gigabyte GTX 970 G1 Gaming
PSU: EVGA SuperNOVA 850 G2

This is bugging me.

I initially have all default setting in BIOS except XMP for memory. But then CPU voltage is locked at 1.28v even during idle, so I turned off XMP and manually enter all the setting XMP modifies (very sure, unless there are hidden ones), except that I did not turn off anti-surge support. Now the voltage go down again when idle. So I used that config for a while without checking prime85 cause I tested when I OC before (see below). But then the computer shut off after 30 mins of Crysis 3 one day and I was concerned, so I tried turning off anti-surge support, but then the computer shut off again after I played my heavily modded skyrim for an hour last night. So I decided to check prime95 today (newest version).

According to OpenHardwareMonitor, MSI afterburner or AIDA64, my cpu temp is 27c when idle, 50c when heavy gaming (80c max on gpu when gaming), so the temperature looks fine to me. When I ran prime95, it's ~65 during first phase (weird thing is the H100i coolant temp is 35 at this point). But then after like 5 mins it enters the next phase in prime95 and my cpu got to ~85c! (coolent temp is still only climbing 0.1c every second) It's stock clock and on H100i so it shouldn't be that high, and after 30s my computer shuts off for no reason.

The really weird thing is that on the day I build this computer, I was overclocking my CPU to 4.6ghz on 1.25v, ran prime95 and I got the EXACT SAME temperatures I said above, while I expect it would be higher @4.6ghz, and it did NOT crash for 30 mins. I would test it longer but I did not use 4.6 because I can't get the voltage to drop when idle, so I choose to use stock setting and now I'm getting weird restarts!

I can't think of any reason for it, besides that the electricity might not be stable in my room. Cause I did the OC in the living room and now I'm in my bedroom.

Any help would be appreciated!

 

[gilbadon]

You have a great PSU. The G2, even with dirty electricity is going to provide very stable rails to your computer. No worries there.

It sounds like the pump on your computer is either not running at full speed, or running at all at some points. Do you have it plugged into the motherboard or connected strait into molex?

 

[bobbyzzx]

You have a great PSU. The G2, even with dirty electricity is going to provide very stable rails to your computer. No worries there.

It sounds like the pump on your computer is either not running at full speed, or running at all at some points. Do you have it plugged into the motherboard or connected strait into molex?

In corsair link my pump speed is constantly ~2300 rpm.

My H100i is connected straight to the PSU sata power, and I have the 3 pin connected to the mobo CPU header. The only thing is the sata power is sharing the same cable with an SSD and a hard drive. Should I use another SATA cable and connect H100i to another sata slot on PSU?

 

[bobbyzzx]

Update: Set all BIOS setting to default, even memory profile, along with corsair link setting. Prime95 shut off in 20 seconds.. But I tried AIDA64 stress test, been 30mins and still up, CPU temp around 60C

 

[gilbadon]

Oh yeah, the H100i connects funny. Do you have the USB header installed?

 

[bobbyzzx]

Oh yeah, the H100i connects funny. Do you have the USB header installed?

Yes I do, and I use corsair link to control the fans on H100i. I just want to make sure if it's ok for H100i to share a SATA cable with a SSD and a HDD, or it might not have enough power in one psu SATA slot for three?

 

[gilbadon]

It has enough power. SSD's & HDD's take no power. They barely draw off of the G2's rail. Getting another cable will not do anything. Can you physically hear the pump?

 

[bobbyzzx]

I'm really not sure about the pump sound, this is the first watercooling system I have owned and I'm not sure how it should sound like.

 

[gilbadon]

I am going to nip this. I do not think it is the pump based on the fact that it is not running extremely hot to start. Maybe you just need to update drivers. Also check your bios settings for smart temperature. When it gets passed x degrees C, what does your CPU fan's speed go to? Just an example. Look at that and try to increase fans to 100% for a while.

If you do want to check if the pump is running...

Try and disable every fan in your computer (and maybe hard drive) so the only mechanical noise would be your pump. Turn it on. If you hear a motor running or water moving, then the pump is working. I know it is a lot of work, but pumps often go bad.

 

[CompuTronix]

bobbyzzx,

I see that you're using the latest version of Prime95, which is 28.5. Under certain hardware configurations, 28.5 will crash due to inadequate Vcore.

Guys,

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

Core i 2nd, 3rd and 4th Generation CPU's have AVX (Advanced Vector Extension) instruction sets. Recent versions of Prime95 run AVX code on the Floating Point Unit (FPU) math coprocessor, which produces unrealistically high temperatures. The FPU test in the software utility AIDA64 shows the same results.

Prime95 v26.6 produces temperatures on 3rd and 4th 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 35% larger Die.

Please download Prime95 version 26.6 - http://windows-downloads-center.blogspot.com/2011/04/prime95-266.htm

Please run only Small FFT’s for 10 minutes.

Your Core temperatures will test 10 to 20C lower.

Please read this Tom’s Sticky: Intel Temperature Guide - http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html

CT

 

[DubbleClick]

It's not like temperatures are the problem here. I'd rather suspect a non definite voltage setting here that puts so much voltage to the cpu it simply suts down.

Fix your voltage manually to like 1.2v and test again. For stability testing, I'd recommend intel burn test, though. Prime may run fine for 12 hours straight but IBT makes it crash after a minute. I have not seen the other way around.

 

[bobbyzzx]

CompuTronix,

Thank you for pointing that out, because the temp seems unrealistic with watercooling.

DubbleClick,

Thank you so much for pointing me to the right direction!

Guys, I figure this out, I think. After I fiddle with all the voltage settings, it seems like turning off VRM Spread Spectrum helps! I ran prime95 again and did not crash for 20 mins. It's just kinda strange because enabling this setting is recommanded in the BIOS tooltip. Hopefully there's not a big draw back for disabling this!

 

[bobbyzzx]

It's not like temperatures are the problem here. I'd rather suspect a non definite voltage setting here that puts so much voltage to the cpu it simply suts down.

Fix your voltage manually to like 1.2v and test again. For stability testing, I'd recommend intel burn test, though. Prime may run fine for 12 hours straight but IBT makes it crash after a minute. I have not seen the other way around.

Is there a way to set the upper voltage limit to 1.2 but then still let it drop when idle?

 

[DubbleClick]

It should do that on it's own, but if it doesn't, where's the problem? You should then try to decrease voltage to the lowest stable level, which is probably going to be around 1.1v-1.15v, either way, even 1.2 isn't even dangerous at all.

 

[bobbyzzx]

It should do that on it's own, but if it doesn't, where's the problem? You should then try to decrease voltage to the lowest stable level, which is probably going to be around 1.1v-1.15v, either way, even 1.2 isn't even dangerous at all.

I think I got a good chip, I can probably do 1.15 or lower, but it just doesn't drop voltage when idle...I know it might not be important but I really want it to do that. Could I be missing some settings? Because when I manually set the voltage I did not change any other setting at all.

 

[DubbleClick]

Might manually enable C states. But there's really nothing wrong with a static voltage, if it varies it will push a lot vcore into the chip after load to keep it stable.

 

[bobbyzzx]

Might manually enable C states. But there's really nothing wrong with a static voltage, if it varies it will push a lot vcore into the chip after load to keep it stable.

I tried enabling c state and tried all the C1 settings, still not luck. Only the frequency drops but not the voltage.

I just ran tests on 1.1v seems stable for now. so you would choose constant 1.1v over 0.7v idle 1.23v load right?

 

[bobbyzzx]

Well seems like I've won the silicon lottery here. 4.4ghz @ 1.09v passed intel test, running aida64 stress right now and still strong! max temp @50c (sounds like I should overclock this...)

Thank you so much guys my main problem is solved! Although I did not get voltage drop to work, I got low voltage now!