CPU Heating up to 100C and Crashing Every Couple of Hours

JazzJazz said:

Apologies on my lack of understanding here, I'm very new to this. I'm trying to find the clover file to copy it - would that be something on the Mac OS USB Install drive or would I have to go through the Clover Configurator app?

I am only seeing softwares used for the install on the stick (see attached).

Click to expand...

Mount the EFI partition on the drive that you boot from. Your EFI folder should be there.

JazzJazz said:

Hi all,

I am a video editor and I run Davinci Studio 17 on my Hackintosh to edit. When I playback anything or perform anything complex on the software, the CPU temperature rises to about 70-100C. That is ridiculously hot. My regular web-browsing temperature remains about 30-40C pretty consistently. I even re-seated the CPU and made sure all the fans were working the way they should be and they are. I am not sure what the problem is here. I understand that I am running an i9 at max 5 Ghz, but should it be generating this much heat?

I'm sure this is a separate issue.. but aside from that, even if the machine doesn't get hot, it crashes consistently on a regular basis. When this happens, the monitors usually projects solid colors randomly (all 4 monitors).

Here is my gear:

Mac OS	Catalina, 10.15.7
CPU	3.6 GHz 10-Core Intel Core i9
RAM	Corsair Vengeance RGB Pro 64GB (3200MHz DDR4)
GPU	Radeon Vega Frontier Edition 16GB
Motherboard	GIGABYTE Z490

I am thinking it's possibly a BIOS setting or bad coolant system, but if you have any ideas please let me know! Happy to provide screenshots and data as well. Thank you kindly.

Click to expand...

Have you enabled your memory's XMP Profile in your BIOS settings? I've heard it can cause problems on some Z490 boards with Comet Lake chips. I've had to disable mine in order to run it stable.

10th Gen Core CPUs can draw 225-250 W under AVX load, with is most likely the case with DaVinci Studio. You may consider a 360 mm AIO or a Noctua NH-D15 (possibly with a second fan) to evacuate that.
Incidentally, your GPU is also a 250-300W monster. Is the power supply up to the task?

I'm running an ASUS over-clocked 10900K + XMP 3600 and Radeon W5700 under full load (Re Cinebench and transcoding) with CPU temps in high-80s and stable over days.

This said, overocking is a notorious PITA.

If you are going past CPU 95C, and crashing on long haul. look at case cooling.

More thoughts:

BIOS overclocking features will let you push past overload no matter what cooler you use, and in last couple years factory overclocking has become science, so there's no headroom — well maybe 3% or 5% if you're lucky. The rule is that Intel and mainboard have already tuned the parts to edge and also given you options to go over the edge.

What you have to grok is that as frequency increases, thermals magnify, so the closer you get to edge the easier it is to go over. A cooler cannot solve this problem, per se. And it can tip the edge, making it even easier to go over, especially with improper installation. Intel parts are rated for steady 100C so that's not dangerous, but at 250W the sheer heat load is a challenge and its likely to build a box that slowly heats up then crashes due to heat affecting timing past bin tolerances, including other parts such as RAM.

(Non-ECC RAM is a bugaboo, which should be kept in back of mind when tuning. Overall, modern gaming kit is tailored to ride edge of peak clock rates where the users don't care, they just restart. This is a very different view than a Mac Pro, which is offered with more care for reliability. Yet lately the i9 OC engineering is so dependable even Apple uses uses in iMac.)

For most hack kit, the simplest way to improve reliability is to put all BIOS OC features at default. There is a cauldren of timing black magic in modern boards, which is very difficult to grok. So respect rated RAM speed, esp XMP there is almost nothing left on table. Ensure cooler is properly mounted, and ensure sufficient airflow / cooling of case, esp around RAM, but also NVMe. Get a temp monitor and watch various units under load to identify any cooling shortfall. Maybe add a fan to cool RAM.

If CPU blows past 90C, back off voltage / timings a little in BIOS. If temps 90 or less back RAM timing off a bit. Penalty for RAM latency increase is minor.

Run memtest86 overnight. If RAM glitches, solve that.

It comes down to reputation+grace of parts suppliers, assemble your kit properly, avoid BIOS tweaks, monitor to not push over edge, and troubleshoot failures as zones, esp. thermal.

Best luck

I would suggest you change the orientation of your Radiator fans, so they are pushing cool external air through the radiator from the front of the case, using the side vents.

If I am correct about your current setup you are pulling hot air from the interior of the case through the radiator, which is not optimal for cooling your CPU.

If you are overclocking your CPU, set it back to stock speed. See what that does for your CPU heat reduction and crashing of the Graphics applications.

When I test for stability, I run Prime95 for a few hours with the following settings:

If you are feeling adventurous, the many i9's are able to tolerate some under-volting. it can lower temperatures and even gain you some performance since it will need to throttle less.