Reboots under load

[Pot Of Jam]

Hello People, This is my first post here, and I hate to say it, but it’s a problem post.

I have had my PC for about a year now and I have been haunted by an issue that I just cannot seem to isolate or solve.

 

The symptoms are: System resets and reboots when being stressed by Prime 95 (after 2-20 minutes) or Games (after 20- 40 minutes). Normal windows use PC is stable.

 

Current PC specs:

Asus Gene VII

Intel i7 4790k (no overclock)

32gb (4x8) Corsair Vengeance pro 2400 (running at 1333)

2x EVGA 980 Ti Superclocked Reference 

Corsair AX860i

Can’t think of anything else that would be critical to the build and would cause error.

 

So when I first built the system It only had one 980 Ti, 16gb RAM and the RM650. The other parts were brought and added in later (after I thought the issue was solved).

 

What I have tried so far:

PSU: Have tried three separate PSU’s, RM650, RM650i and an AX860i

 

Motherboard: Swapped out for another 1150 board, system was stable. No stress testing was ran on it, games however were being ran for 8+ hours with no instance. Motherboard was sent back to supplier and was sent back with no fault found.

 

CPU: After I discovered that Prime 95 caused a reboot and I could successfully replicate the error, I sent the mobo, CPU and RAM back to Scan (supplier) and they confirmed that the CPU was throttling and was replaced, they also stated that the CPU and RAM were fine. After this I brought the 2nd Ti and extra 16gb of RAM, then liquid cooled the system. The issue occurred again after 40 mins of game play on the Division with the new CPU. I ran Prime95 and the system reset again after 20 mins.

 

So I am now running Memtest86 on the sticks of RAM, first I will run all sticks through slot one for 8+ hours, and then if all sticks show no errors, I will run one stick through the other 3 slots.

 

If there is anything I have missed or you have suffered similar problems please let me know, all advice is welcome!

 

Thanks!

[done12many2]

9 minutes ago, Pot Of Jam said:

Intel i7 4790k (no overclock)

I know you aren't overclocking, but what is the voltage setting in BIOS for your CPU?  

 

You had incidents of throttling which indicates that during certain load situations, the CPU is getting too hot and trying to protect itself by lowering operating frequency.  If your CPU is getting extremely hot, it can and often will reboot.

 

Are you monitoring temps and and if so, what are you using to monitor them?  What type of cooling system are you using?

[Pot Of Jam]

18 minutes ago, done12many2 said:

I know you aren't overclocking, but what is the voltage setting in BIOS for your CPU?  

 

You had incidents of throttling which indicates that during certain load situations, the CPU is getting too hot and trying to protect itself by lowering operating frequency.  If your CPU is getting extremely hot, it can and often will reboot.

 

Are you monitoring temps and and if so, what are you using to monitor them?  What type of cooling system are you using?

Hey, yeah temps are on the wamish side for a custom waterloop (and on the H115i I originally had) when playing the Division temps were at 87 while GPUs were at 60ish.

 

I am using after burner then remote server to monitor on phone.

 

Edit, I will get voltage when I get home!

[done12many2]

18 minutes ago, Pot Of Jam said:

Hey, yeah temps are on the wamish side for a custom waterloop (and on the H115i I originally had) when playing the Division temps were at 87 while GPUs were at 60ish.

 

I am using after burner then remote server to monitor on phone.

 

Edit, I will get voltage when I get home!

Just verify that your voltage is good and that your custom loop is doing the job.  Use something like AIDA64 or HWMonitor to monitor temps and voltages during testing.  

 

I'm assuming that you have your 980 Tis incorporated into the loop?

 

When I monitor temps, I keep an eye on a few things.  I start with ambient temp as it impacts all other temps that I monitor.  CPU and GPU temps are a given, but I also watch water loop temp. I actually watch quite a bit more, but we'll keep it basic.  Depending on the thermal displacement capacity of your custom loop, water temps may continually rise over the course of time.  As water temps rise so do CPU and GPU temps.  If temps keep rising, well you get the picture.

 

Good luck and let us know what you come up with.

[Pot Of Jam]

Well, temps never go any higher than i have mentioned in the above posts, even when running heaven.

 

And before I had the custom loop the CPU  was cooled by a H115i and the GPU was aircooled by the reference cooler, CPU temps are the same under both cooling solutiond, but the GPU sat around 78-80 on air. So 2x 980s with a temp of 60 would suggest the loop is working, and that the CPU is getting too hot under both.

[done12many2]

1 hour ago, Pot Of Jam said:

Well, temps never go any higher than i have mentioned in the above posts, even when running heaven.

 

And before I had the custom loop the CPU  was cooled by a H115i and the GPU was aircooled by the reference cooler, CPU temps are the same under both cooling solutiond, but the GPU sat around 78-80 on air. So 2x 980s with a temp of 60 would suggest the loop is working, and that the CPU is getting too hot under both.

Great looking build!  Very nice.

 

So the Corsair H115i is a great AIO and does a fantastic job of tackling CPU heat efficiently.

 

When you jumped ship and went with custom loop you went from water cooling a CPU only to a cooling a CPU and 2 x GPUs with less surface area per block.  You went from a radiator surface area of 280mm (H115i) dedicated to the CPU only to what appears to be 480mm (2 x 240) to cool all 3 water blocks.  

 

The two GPUs add a great deal more heat to the loop than your CPU and now that you've incorporated them all into one loop, your loop's ability to deal with the CPU's thermal load has technically diminished bit.  If anything, you should be seeing worse CPU temps under full GPU/CPU load when comparing the effective CPU cooling capacity of the H115i and custom loop with the same given air movement through the radiators.  If you did improve air flow through the radiators when you upgraded to a custom loop then that would lower the delta between the old setup and new with regards to CPU temps, but it would take a significant amount of airflow to get the new loop to the same CPU cooling capacity as the old.

 

With GPU temps in the 60's with the new loop, I have no doubt that your water temps are getting high which means that the thermal dissipation capacity could be improved.  For the sake of comparison, I have 2 x 980 Tis in my loop as well and under full load on both CPU and GPUs, my GPUs never leave the 30's.  They were getting into the low 40's when I had 960mm of surface area (2 x 360mm and 1 x 240mm) to cool them and the CPU, which wasn't bad, but I have since added radiators resulting in a dramatic improvement.

 

60's for your GPUs is well within their acceptable operating temp and there are no issues there, but the problem is that if your GPUs are hitting 60c, that means that your water temps are fairly warm.  In the end, your CPU is being cooled buy much warmer water than it was when it was on the H115i alone.

 

A 4790k that is not overclocked should never be in the 80's on water and as previously mentioned, you should confirm your CPU voltage in BIOS and continue to monitor it while running some stress tests.  Watch CPU temps and voltage closely.  Be fair to yourself in your assessment of CPU temps.  If they were in the high 80's before, the individual cores may be spiking really high from time to time, which can result in throttling and reboots.

 

If you determine that heat is a possible cause of the throttling/reboots, you're going to have to increase your loops thermal dissipation capacity.  This can be done by increasing radiator surface area or by increasing air flow over the radiators that your currently have.  Judging by the picture of that great build of yours, you don't have room for anymore radiators, so you may need to focus on airflow through the rads.  That's easier, but comes at the cost of noise, which I despise!

 

Low CPU voltage will cause reboots as well, but I don't suspect that due to your mention of throttling and high CPU temps at stock CPU speeds.

 

Man that was a lot.  Hope it helps.

[done12many2]

Additionally, get yourself a water temp sensor if you can.  Being able to compare water temp to ambient room temp will tell you a lot about what's going on with your loop.  Your motherboard has a temperature sensor header available so all you need is the sensor.

 

[Pot Of Jam]

Thanks for the info!

 

Temp sensor,: it is on my shopping list! Nice monsoon one to go with the fittings.

 

Well my worry is this; when the PC was on Air/AIO the CPU was still hitting the same temps as it is now with the full loop. I understand what you are saying, but if the water in the loop is enough to cool a 250 tdp (GPU)to 60 then surely a 88 tdp should be near enough that in games, after all temps are the same while Idle (from 24 with fans at 100% to 43 with fans on 30%). But I don't have the expertise in this area so I am assuming.

 

Fans are at the moment not the best. On the front rad I have two Phobya Nano 2g 1500's and on the top Rad/Predator I have the Vardar fans that came with it. Due to the inconsistency with the RPM's (Vardar being 2200 and Nanos Being 1500) I can't really set a good enough curve. So for now, while benching, stressing and gaming fans go on full tilt and I wear my headset.

 

This doesn't bother me, as you have pointed out, I have very little radiator space/surface. I knew this from the offset and the main job of the loop is to keep the temps of the GPUs under 83 (throttle point), so to accomplish this, If the fans spin at 75% and make more noise than silent, but less noise than two reference coolers at 100% I will live with it (2x 980 ti's at 100% is about the same noise as a 747 at take off - fact).

 

 

 

 

[done12many2]

Great post.  

 

I have to mow the lawn, but I'll think about your setup while I'm at it.

 

 

 

 

[Pot Of Jam]

Some more info for you, ran like this with one stick for 30mins no reboot.

 

Temps stayed at 60, they did raise to 68, but dipped back down.

 

 

[done12many2]

4 minutes ago, Pot Of Jam said:

Some more info for you, ran like this with one stick for 30mins no reboot.

 

Temps stayed at 60, they did raise to 68, but dipped back down.

 

 

Delayed mowing due to dead battery!  It's charging up.

 

Temps look good on the CPU for a CPU only load.

 

I'd recommend loading those GPUs a bit with something like Heaven or Valley looping in windowed mode in the background while running your CPU stress at the same time.  This will thoroughly load your water cooling, which is what can happen at certain parts of a game.  It won't happen too often, but you have experienced reboots during gaming.

 

Also I recommend using RealTemp to monitor your temps during your testing.  I've found it and AIDA64 to be my most dependable sources for temp monitoring.

[done12many2]

Also, it doesn't look like your processor is hitting its max stock turbo speed?  Have you changed any setting in BIOS for anything?

 

Nevermind.

[Pot Of Jam]

Ah, batteries suck on all things then!

 

I tried using AIDA64, but kinda confused me a little so stuck to what I know!

 

Real Temp is doable though.

 

And will now run with heaven too.

 

Will keep you posted

[done12many2]

3 minutes ago, Pot Of Jam said:

Ah, batteries suck on all things then!

 

I tried using AIDA64, but kinda confused me a little so stuck to what I know!

 

Real Temp is doable though.

 

And will now run with heaven too.

 

Will keep you posted

Please do.  I want to see this get fixed.  I hate seeing a setup like yours in a crippled status!  

 

Yeah, every year the battery on the mower dies, yet I still turn the key hoping that it didn't.  I need to keep it on a charger during the winter, but it just never happens.  I end up buying a new one about every 3rd year.

[Pot Of Jam]

Hmmm... cant get the GPU's to 100% load on heaven with Prime 95 running.

[done12many2]

Just now, Pot Of Jam said:

Hmmm... cant get the GPU's to 100% load on heaven with Prime 95 running.

That's fine as long as they are loading up pretty good.

 

Make sure when you talk about CPU temp, you use core temps.

[Pot Of Jam]

Well the GPUs are not even hitting 35% usage, temps are sitting at 40 on them.

 

Real Temp is showing: 61 64 64 62 on the cores.

[done12many2]

If you are still using Prime95 to load that CPU, you'll want to wait a bit because they will shoot up as it progresses.  As it stands right now, everything seems good.

[done12many2]

Another thing that will more easily heat up both at the same time is RealBench.  I use it, AIDA64 and OCCT to stress my computer and once all three have passed, I call it good.  I stopped using Prime95 a long time ago.

 

You can download here.  The option to download is near the top and centered.

http://rog.asus.com/rog-pro/realbench-v2-leaderboard/  

[done12many2]

I'm off to test that battery.  I'll check in on your progress in a bit.

[Pot Of Jam]

Oh... hello. I am back on the login screen.

 

Damn it. Whlie I was outside too.

 

And as I thought, the recording corrupted.

[done12many2]

4 minutes ago, Pot Of Jam said:

Oh... hello. I am back on the login screen.

 

Damn it. Whlie I was outside too.

 

And as I thought, the recording corrupted.

Yeah bud, we need to get this fixed.  Sorry to say, but I'm on the mower now. It'll be a bit. 

[Pot Of Jam]

No worries man! I appreciate the help!

 

Although, I think I am done for the night, its 9pm here, I have set this stick of ram off on Memtest86 for the night.

 

After the system failing with this stick in, either it was bad timing, or, it is the cause. I am interested in seeing the results tomorrow.

 

When you say "on the Mower", is it a ride on? because that is pretty awesome

 

Edit: Also After the reboot I set it all up again, because when this happened last time (when I had the original kit of RAM in) I could replicate the error, but after 20 mins of holding my phone pointing at the screen, I got tied and bored.

 

Also switched to Furmark for the GPU stress, as it was Actually stressing the GPU's out. Temps on the CPU were rising to around 73-75 across the cores, and 55-59 on both GPU's when I finally stopped the tests.

[done12many2]

2 hours ago, Pot Of Jam said:

No worries man! I appreciate the help!

 

Although, I think I am done for the night, its 9pm here, I have set this stick of ram off on Memtest86 for the night.

 

After the system failing with this stick in, either it was bad timing, or, it is the cause. I am interested in seeing the results tomorrow.

 

When you say "on the Mower", is it a ride on? because that is pretty awesome

 

Edit: Also After the reboot I set it all up again, because when this happened last time (when I had the original kit of RAM in) I could replicate the error, but after 20 mins of holding my phone pointing at the screen, I got tied and bored.

 

Also switched to Furmark for the GPU stress, as it was Actually stressing the GPU's out. Temps on the CPU were rising to around 73-75 across the cores, and 55-59 on both GPU's when I finally stopped the tests.

Do they make a mower without a seat and steering wheel? 

 

We can pick this back up whenever you have some time.

[Pot Of Jam]

Hah! Yeah well over here in the UK most of us don't have lawns that warrent a sit on mower! But we do have petrol ones, they will remove a foot if needed too...

 

Well after 4 passes in Memtest86 there were no errors on that stick of RAM that was being used when the PC rebooted.

 

So... meh.