G3258 overclocking heat problems

Jackal7618

Reputable
Jul 13, 2015
68
0
4,630
Hey so here's my build

Cpu: Intel G3258 with recently brought and installed Cooler Master Hyper TX3
Ram: 8 gb
Mobo: Gigabyte GA-B85M-HD3
Gpu: Geforce 750ti twin frozr
HDD: WD Blue 1 TB
PSU: 450watt (came with the Ultra XBlaster Mid-Tower V2 Case)
Case Fan: 1 120mm at the back

I'm new to overclocking I followed this guide:
[video="https://www.youtube.com/watch?v=8LazZv1FIwQ"][/video]

So after trying aton of things and experiencing aton of crashes I've decided to seek help so here are the results (even sitting idle with the base clock it heats up around 40-60c):

3.8ghz - instantly spikes to 80c after 5mins in prime95 heats all the way up to 100c!
4.0ghz - Instantly spikes to 85c after 5mins in prime95 heats up to 100c and crashes
4.1ghz - instantly spikes to 90c after 2mins in prime 95 heats up to 100c and crashes
4.2ghz - instantly spikes to 90c after 1 min in prime 95 heats up to 100c and crashes

Please do not not flame my noobness for I am new to overclocking, Thank you please advise.

(sorry for the double post)
 
Solution
Well,
1- do not increase Vcore from default unless you have to and only do so in small increments, don't just turn it up to 1.25V if that is what you did.
2- the temperature rapidly ramping to 90C seems to indicate poor contact between the CPU and heatsink, though that could also be caused by excessive Vcore
3- there is no universal setting for Vcore, every chip is different, you cannot follow someone else's instructions and expect the same outcome. Start from default, bump it up incrementally and if that fails to make a multiplier any more stable, step the multiplier back by one notch and try again. In some rare cases, people have also had success with skipping multipliers but I would have concerns about long-term stability with that...
First of all, prime95 is not a good stress test.
It exercises only a subset of insturctions that generate heat.
I like OCCT which stresses using instructions you might use every day.
It will shut off the test at 85c.

The guide you had space for is blank.

Reset all to default or auto.
Then, gradually increase the multiplier.
Monitor the vcore. When it gets to about 1.25, that is about as far as you want to go 24/7 without damage. 1.30 max.
How high you can go is largely determined by the luck in getting a good chip.
I might expect 3.8 to 4.2.

If temperature gets too hot at 1.25v, a better cooler will help.
The TX3 is not much better than the stock cooler. Check to see that all 4 pushpins are through the motherboard and locked. The cooler should not wiggle if you nudge it. Too much paste can act as an insulator. A small drop in the center is all you need; it spreads under heat and pressure. I would expect that at idle, you would be about 15c. over ambient.
 

InvalidError

Titan
Moderator
Well,
1- do not increase Vcore from default unless you have to and only do so in small increments, don't just turn it up to 1.25V if that is what you did.
2- the temperature rapidly ramping to 90C seems to indicate poor contact between the CPU and heatsink, though that could also be caused by excessive Vcore
3- there is no universal setting for Vcore, every chip is different, you cannot follow someone else's instructions and expect the same outcome. Start from default, bump it up incrementally and if that fails to make a multiplier any more stable, step the multiplier back by one notch and try again. In some rare cases, people have also had success with skipping multipliers but I would have concerns about long-term stability with that sort of quirky OC
4- power supplies bundled with cases usually suck, you might be sitting on a time bomb there
 
Solution
Did you remove the thin clear plastic protector from the bottom of the heatsink?
Did you apply thermal paste to the bottom of the heatsink to fill in the groves beside each heat pipe?
If you did all of the above you may need to turn the heatsink 90 degrees to make each heat pipe contact the cpu cores. The die is a thin strip in the center of the heat spreader. with the cooler mounted one direction the center heat pipe will run along the length of the die and the 2 outer pipes only cool the heat spreader. So turning it 90 degrees will make all three heat pipes cross the cpu die.
 

CompuTronix

Intel Master
Moderator


geofelt,

I'm always careful to be very specific when discussing Prime95, because there's so much confusion concerning high temperatures with regard to the Instruction Sets to which you're referring.

As you probably know, and just to clarify for the benefit of our Members and Tom's readers, Prime95 version 26.6 is well suited to all Core i and Core 2 processors. However, later versions such as 28.5 run AVX code on the processor's Floating Point Unit (FPU) which causes extremely high Core temperatures.

For this reason, I recommend Prime95 version 26.6 Small FFT's for thermal testing in my Intel Temperature Guide - Intel Temperature Guide - http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html

Regardless, the Pentium processors, including the G3258, do not have AVX, AVX2 or FMA3 Intruction Sets, so in this instance, it doesn't matter which version of Prime95 is run, as the core temperatures will be the same.

You and I have been here at Tom's since 2006, and your countless very astute contributions are greatly appreciated! Since 2007, my goal has been to clarify Intel's temperature spec's for our Forum Members and readers through my Temperature Guide, and whenever I can, one thread at a time.

The reasons that most readers find processor temperatures so confusing is not recognizing the vast number of variables in hardware, software, environmental conditions and test methods which influence temperature measurements, and also because of Intel's poorly written Thermal Specifications.

As a result, the topic of processor temperatures is surrounded by a great deal of misinformation, is very confusing and highly misunderstood. Therefore, I always encourage everyone to be as specific as possible in their explanations, because the essence and understanding of the topic is in the details.

Again, thanks for all your hard work here at Tom's!

CT :sol:
 

CompuTronix

Intel Master
Moderator
Jackal7618,

As geofelt has pointed out, the TX3 uses the same push-pins as Intel's stock cooler.

It's very likely that your stock cooler has popped a push-pin loose from the motherboard, which is a common problem. This causes poor contact pressure between the cooler and the CPU, resulting in high temperatures in BIOS, as well as in Windows at idle, and especially at 100% workload.

You can troubleshoot this problem by pushing firmly on each corner of the cooler for about 30 seconds while watching your load temperatures. When you see a significant drop, you've found the loose push-pin.

The push-pins can be deceivingly tricky to get them fully inserted through the motherboard and properly latched.

(1) To re-seat a single loose push-pin, release the latch mechanism to the retracted clockwise position, then reset it to the counterclockwise position, but do NOT push on it yet.

(2) To get the pin fully inserted through the motherboard, push only on the leg, NOT on the top of the latch.

(3) While holding the leg firmly against the motherboard with one hand, you can now push on the top of the latch with your other hand until the latch clicks.

(4) If you're re-seating the entire cooler, then be sure to latch the pins across from one another, rather than next to one another. Use an "X" pattern, so as to apply even pressure during installation.

Intel Stock Cooler Installation Guide - http://www.tomshardware.com/forum/338655-28-intel-stock-cooler-installation-guide

CT :sol:
 

Jackal7618

Reputable
Jul 13, 2015
68
0
4,630


So an update I have upgraded to a Cooler Master Seidon 120m liquid cooling, installed 4 system fans, moved to a better air flow proficient case and brought a 650w bronze 80+ psu I can overclock to 4.2 reaching 60-70 ish while stressing with prime 95 via small FFT's (seemed stable at the time) but after a hour or two I result in a crash. I have tried lowering and raising my vcore to no accord.