CPU has scary temps while idling after GPU Upgrade

TheDemonsSanctum · Mar 26, 2016

Hi guys, this is my first post on Tom's Hardware. I have a problem that I seem to not to be able to solve. I have an AMD A8-5600K Trinity Quad Core and had an EVGA GeForce GTX 750TI.

When I was running with my 750TI, my CPU temps were perfectly normal (Idling around 36*C with DeepCool IceBlade 200M CPU Heatsink). My CPU is not overclocked although it can be overclocked to 4.1GHz stable. I bough an Asus GeForce GTX 770 OC DirectCU II with 2GB GDDR5. I completely pulled my Computer apart to dust and clean the inside properly. I got my computer back together successfully and got into windows. Did a clean install as JayzTwoCents told me to do, after that, i restarted, and opened up American Truck Simulator playing on all high settings. A few minutes in, my screen goes black and I hear my PSU fan ramp up (Thanks to EVGA putting in what sounds like a server grade 40MM fan in the PSU *sarcasm*). Well I installed CPUID HWMonitor and it says my CPU Core Temps are above 50*C nearing 80*C. At first i just thought that i shook the Heatsink around or the Thermal Paste could be Dried out or created a air bubble in there that is prevented good heat transfer and dissipation (again pulled my computer apart to get into the small crevice to pull the Heatsink off). Booted back up after new thermal paste has been applied (Cleaned both CPU dye and Heatsink pipes for those wondering) and HWMonitor was still giving me a high reading. I look on the internet and see that HWMonitor has been giving incorrect CPU readouts. So i installed Core Temp, same thing.

I need your guys help because most of you are probably Computer Scientists or Highly Respected Enthusiasts. Is this a bottleneck that i have here. And what is the cheapest resolution to this, or could my CPU be going out the window.

I will be upgrading to a i5-4690K and a custom watercooling loop, but that will take a few months so i kinda need this PC up as this is my main rig for editing

Thank you ahead for any replies and for this lengthy article, i really need help to get this fixed

Darkbreeze · Mar 26, 2016

Your first mistake was listening to anything JayTwoCents has to say.

What is your cooler model?

How much thermal interface material (Paste) did you apply and what kind. What method did you use?

Sounds like the heatsink isn't firmly seated, which is pretty common actually. Hard to say without knowing your full specs including case model, number of fans, fan orientation, cooler model, etc.

GhislainG · Mar 26, 2016

HWMonitor is not useful with an AMD CPU. What are the CPU thermal margins according to AMD Overdrive? They should be fine (> 10°C) if the cooler is installed correctly. Is your PSU powerful enough to power the new GPU? Presuming the GTX 770 is used, did you test it in another PC?

Darkbreeze · Mar 26, 2016

How would testing the GPU in another machine be helpful in determining why the CPU is getting hot? Not sure where you're going with this?

Also, CoreTemp has the option to set the thermal display to thermal margin in the advanced settings, and it's just as reliable as Overdrive. Readings are taken from the same sensors and it uses the same tables to calculate margins.

GhislainG · Mar 26, 2016

Darkbreeze :

A screen that goes black is not caused by a CPU that overheats; it would throttle and make the system slow (even an AMD processor can do that). It most likely gets black because the PSU isn't up to the task or the GPU has an issue (the reason to test it in another system).

TheDemonsSanctum · Mar 26, 2016

Darkbreeze :

The Cooler Model of the CPU or the GPU?

I used Thermaltake TX-7 and used about a pea sized dot of thermal paste.

My case is the EVGA Hadron Air ITX
I have two fans as intake at the top, one of the two fans on the heatsink in push orientation. The heatsink is firmly seated as i had checked a few times before installing the motherboard.

TheDemonsSanctum · Mar 26, 2016

GhislainG :

The screen didn't only go black, my whole computer had restarted. After it restarted, it had no error messages, and the BIOS startup screen didn't have any errors popping up.

TheDemonsSanctum · Mar 26, 2016

GhislainG :

I also looked up the minimum power for the 770 and Asus's website said minimum 500W, Recommended 550W, the proprietary PSU that i have is from EVGA and only fits in the EVGA Hadron Air/ Hadron Hydro.

GhislainG · Mar 26, 2016

What PSU do you have? Personally I'm not convinced you have a CPU overheating issue. It really is up to you, but if it was my system, I'd monitor the thermal margins with AOD while stress testing the CPU; it shouldn't throttle or cause a system restart. If the CPU is fine, then the issue has to be with the PSU (most likely) or the GPU.

GhislainG · Mar 26, 2016

The EVGA Hadron Air/ Hadron Hydro include a 500W PSU that provides 480W on the +12V rail; unless the PSU has an issue, that's plenty for your system.

Darkbreeze · Mar 26, 2016

No, that's not actually correct. It's a U1 form factor, and while not widely used, it's certainly not proprietary either. There aren't many of them out there, but there are replacements available from other sources than just EVGA. None of them are extremely high powered though and you have to be careful that they have the necessary power cabling because many of them have only a single 6 pin PCI power plug and the GTX 770 uses several different configurations depending on model, but mostly a single 6 pin and single 8 pin, and a 575w unit is recommended.

http://www.realhardtechx.com/index_archivos/Page362.htm

GhislainG · Mar 26, 2016

Is this the case? https://www.guru3d.com/articles-pages/evga-hadron-air-review,2.html They tested it witha GTX 780.

Darkbreeze · Mar 27, 2016

I wasn't indicating the PSU wasn't capable enough. I was responding to the statement that it was proprietary.

GhislainG · Mar 27, 2016

Darkbreeze :

I understood that and it obviously is valuable info; I only want to make sure the case really is the one that I linked.

TheDemonsSanctum · Mar 27, 2016

Darkbreeze :

I had thought that when i purchased it, i was told that the PSU was a proprietary piece of hardware, not a Server PSU

TheDemonsSanctum · Mar 27, 2016

GhislainG :

That is indeed the case. I just don't know why my CPU is getting so hot and not cooling down. Could my power supply be going bad and therefore giving it too much juice?

GhislainG · Mar 27, 2016

Can you post the thermal margins at full load? If it gets too hot, then the CPU should throttle. AOD is very nice because it displays a nice graph for each core, etc. Throttling usually causes loss of performance, not a system reboot. That would happen only if the cooler is not mounted properly and throttling the CPU down to 1.4 GHz still isn't enough to keep thermal margins above 0°C.

Darkbreeze · Mar 27, 2016

Power supply/CPU relationships don't work that way. Your PSU could never "give it too much juice" unless there was a short on the motherboard or CPU, and that wouldn't last more than a few seconds before the entire deal was over, and you needed new hardware.

TheDemonsSanctum · Mar 27, 2016

GhislainG :

I will tomorrow after i get home from work, however, my motherboard has thermal protection on which is set to power off if and when the cpu gets too hot.

GhislainG · Mar 27, 2016

You may want to temporarily disable that setting to check the thermal margins.

Darkbreeze · Mar 27, 2016

HWmonitor and Open hardware monitor are notoriously inaccurate on a variety of chipsets. I never, ever, recommend using them. Either use HWinfo, CoreTemp or AMD overdrive, for your configuration.

Idle temps on AMD chipsets are meaningless. They are not accurate at the low end of the spectrum. Upper temps are pretty accurate, but are more accurately reflected as thermal margin, so using CoreTemp with thermal margin readings enabled in the advanced settings, or using Overdrive, is highly recommended. Using HWinfo is still a much better option than HWmonitor though. I see twenty threads a week here where temperatures are misreported in HWmonitor and switching to another utility immediately shows that thermal compliance is fine, despite showing differently in HWmonitor or Open Hardware monitor.

TheDemonsSanctum · Mar 27, 2016

*UPDATE* So, I left my computer on for the entire time i was a work, and CoreTemp and CPUID HWMonitor are still giving those same readings, however in the BIOS, it says the temperature is around 30*. I have done some researching and found that HWMonitor and CoreTemp read the temperature of the AMD Socket, not the actual dye, and after leaving it on the entire time, i touched my CPU Heatsink, and my GPU Heatsink/Backplate was very cool to the touch. I think that CoreTemp and HWMonitor was just reading the socket temperature, not where the heat gets dissipated from the Heatsink. If this is correct then please inform me that it is, if it is not, please reply with the correct answer, and how i may fix it as well as others.

GhislainG · Mar 27, 2016

Only AOD is reliable. If your thermal margins are too low, then you have an issue.

Darkbreeze · Mar 28, 2016

Clearly he doesn't want to listen to what we're recommending. Guess he doesn't really need help. I'm out.

GhislainG · Mar 28, 2016

I'll continue to help only if he provides the AOD results.

CPU has scary temps while idling after GPU Upgrade

Reputable

Retired Mod

Titan

Retired Mod

Titan

Reputable

Reputable

Reputable

Titan

Titan

Retired Mod

Titan

Retired Mod

Titan

Reputable

Reputable

Titan

Retired Mod

Reputable

Titan

Retired Mod

Reputable

Titan

Retired Mod

Titan

Share this page