Interpretation of OCCT Test: CPU Temperature, Performance Issues, Arctic MX-2 Paste & ATX 500W PSU

I have a problem with CPU temperature and performance. At no load it is around 50 degrees, when playing it reaches 84 and rather does not rise higher (the lowest CS: GO settings; measured HWMonitor). During the OCCT test, the temperature exceeds 85 degrees and the test stops. BOX cooling, thermal paste Arctic MX-2. I have the impression that the processor (or maybe something else) is not stable, in the game you can feel performance drops, some cutoffs. How can you solve the problem? Replacement of cooling or maybe it's the power supply's fault? Below are the results from OCCT, could someone interpret them?
I can say only about the power supply: BOX ATX 500W SFAN CE + PFC 12cm / FAN. Such information is on the computer case

Hello, cooling is box, but from a different processor. It is too weak for this processor, its maximum allowable temperature Tcase is 67 degrees (so on the cores it will be maybe about 75 degrees). If the housing does not provide adequate ventilation, it will also be warmer in it.

Information on the power supply can be found on the power supply itself.

A Power Supply test would be more useful than a CPU - the PS test will also heat up the graphics card.

Temperatures are good or too hot (I don't want to count). Generally it is Xeon on a modified disc. HWmonitor is a moron ancient program that should be removed from the network but dinosaurs still love it.

Install HWiNFO and run in "sensor only" mode. At the core temperature, right click and set Tjmax to 85 degrees. The core temperature readings in this cpu on this board are just speculation. As already be guided by the CPU temperature reading from the motherboard.
With HWiNFO in the background, run 10 minutes of the power supply test. In the OCCT settings turn off the test stop after exceeding the specified temperature or set the core for 95 degrees. After the test, give screen from HWiNFO.

PS - follow the department regulations and list the components as you please.

In order for the results to be reasonably real, you need to set the correct TjMax in the temperature monitoring program. e.g. for E54XX it is 85 ° C

PS. Everything is OK.
PS2. Although on this BOX ATX 500W SFAN CE + PFC 12cm / FAN - no matter what set would be powered, I would be afraid.

Temperatures are not tragic. You can think of a bit more efficient cooling. However, the first thing I would do was replace the power supply. Have you ever checked the voltage on it ??

310artur wrote:

HWmonitor is a moron ancient program that should be removed from the network but dinosaurs still love it.

OCCT is also an ancient program that also uses HWMonitor for monitoring by default. Well, but I understand this is not the main application of this program. But in the Powersupply test compared to IntelBurnTest (set 8/12 threads - to prevent GPU reduction) + Furmark: GPU shows similar power demand and CPU?
OCCT: powersupply - maximum approx. 185W, and IBT + Furmark approx. 230W. At least in my situation, the Powersupply test does not load the components (including the power supply) to the maximum extent, so it does not fulfill its task.

Testing in extreme situations that NEVER occur in normal computer use is also pointless

But we are sure that such things as restart due to a too weak power supply will never occur, unless the power supply will wear out.

sylweksylwina wrote:

But we are sure that such things as restart due to a too weak power supply will never occur, unless the power supply will wear out.

With such testing and with the author's topic of the pseudo-power supply it will probably lead to its damage, and by the way the rest / some component!

And this is another matter, besides, we do not know the full configuration of the computer.

With such a weak power supply, I wouldn't risk too extreme tests. Damage to the power supply may result in other components.

sylweksylwina wrote:

OCCT is also an ancient program that also uses HWMonitor for monitoring by default.

That's right What is worse, I have never managed to force this program to use everesta, although it has such options. Therefore, when I propose the OCCT test, I always ask for a screen from HWiNFO which collects background readings during the test. However, the algorithms for loading are all the same - you can clearly see the furmark and GPU in OCCT are very similar tests and CPU tests are based on calculations in iterations of the linkpack and its author's variation. This is not all new, but it does the job, so why bother.

As for the load - the test power supply loads individual components less than even separately tested CPU and GPU in the OCCT itself, but this is its task. This is a simulation of a very hard game. And so the total load is greater than it will ever appear in normal operation. The game, even if it forces such a load, is not continuous and uninterrupted for several minutes. Well, unless someone likes to play on the edge of playability in the eternal lag, or uses a PC to work with some applications, e.g. video processing that can effectively use the computing power of both CPU and GPU.