[SOLVED] Intel Pentium G3258 Overclock issue

[NerdyComputerGuy]

I've got an overclocked G3258 to 4.2ghz at 1.450v, you may be surprised when you see how much voltage i have on the CPU but nothing seems to work, it's still unstable.

The blue screens occur when I am recording with OBS and playing an intensive game, if i run below 1.4v my PC crashes, i've had to increase the voltage over time to the point where i am at 1.450v now.

Any help is much appreciated.

 

[Lutfij]

You might want to list your full specs like so:
CPU:
Motherboard:
Ram:
SSD/HDD:
GPU:
PSU:
Chassis:
OS:

Also list your cooling onboard that system. Higher voltages to keep an overclock afloat is usually sign that the motherboard is giving out or you've subjected the CPU to too much voltage.

Please mention any other overclocks you've done to it.

 

[NerdyComputerGuy]

You might want to list your full specs like so:
CPU:
Motherboard:
Ram:
SSD/HDD:
GPU:
PSU:
Chassis:
OS:

Also list your cooling onboard that system. Higher voltages to keep an overclock afloat is usually sign that the motherboard is giving out or you've subjected the CPU to too much voltage.

Please mention any other overclocks you've done to it.

CPU: Intel Pentium G3258
Motherboard: Asrock z97 pro 3
Ram: HyperX blu 8gb 2 sticks
SSD/HDD: 120gb ssd, 2x hdd
GPU: Geforce GTX 760
PSU: Corsair CX500
Chassis:
OS: Windows 7 Pro 64bit

Cooling system, like 3 120mm fans, stock cooling fan for cpu.

I've only ever done a 4.2ghz OC to it and i've had it since 2015, i've had to increase the voltage over the years, i used to be able to stream and record with OBS fine but since recently i've had FPS drops on OBS and the PC just blue screens randomly, if i play a game for a long period of time this can cause it to blue screen or if i am recording / streaming for over 1hr this can cause it to blue screen.

 

[NerdyComputerGuy]

I have the CPU Cache voltage to 1.4v and the ram is at 1.65v

Maybe i need to lower the ram voltage?

 

[CompuTronix]

lNerdyComputerGuy,

The maximum recommended long-term Vcore for 22 nanometer processors is 1.300 volts. Your Core voltage is far too high, which has degraded your processor. You're seeing it first hand, and it is not reversible.

Excessive Vcore may result in accelerated "Electromigration" - https://www.google.com/?gws_rd=ssl#q=Electromigration

This prematurely erodes the traces and junctions within the processor's layers and nano-circuits, which will eventually result in blue-screen crashes that become increasingly frequent over time. As a rule, CPU's are more susceptible to Electromigration with each Die-shrink. However, the most notable exception is Intel's 14 nanometer Microarchitecture, where advances in FinFET transistor technology have improved voltage tolerance.

Here's the maximum recommended Core voltages per Microarchitecture from 14 to 65 nanometers since 2006:



When tweaking your processor near it's highest overclock, keep in mind that for an increase of 100 MHz, a corresponding increase of about 50 millivolts (0.050) is needed to maintain stability. If 70 millivolts (0.070) or more is needed for the next stable 100 MHz increase, it means your processor is overclocked beyond it's capability.

The only temporary fix for Electromigration is to further increase Vcore, as you've done, which will only accelerate the degradation process. The better alternative is to reduce Core speed and voltage, which no overclocker wants to hear.

I have a G3258 in my wife's rig which has been OC'd to 4.7GHz at 1.29 volts since I built it in early 2014. I have another in my niece's rig OC'd to 4.6 at 1.28 volts which I built for her in 2015. Both run cool on Hyper 212 EVO's. To get my niece's stable at 4.7 required more than 1.335 volts, which I was not willing to risk. The difference between 4.6 and 4.7GHz is less than 2.2%, which has no noticeable impact on overall system performance. It simply isn’t worth pushing a processor beyond recommended Core voltage and Core temperature limits just to squeeze out another 100 MHz.

A few days ago I saw another forum member with a G3258 OC'd to 5.0. I don't recall the voltage, but since it didn't catch my attention, it couldn't have been exceedingly high. It's the Haswell edition of the Silicon Lottery. He got one in the top few percent, I got a few that were decent, but unfortunately, your sample wasn't so good. Sorry, but that's just how it goes.

Your CPU is dying. If you don't want to be replacing it any time soon, then I recommend that you dial down the Vcore and Cache to no more than 1.300, and live with whatever OC that voltage will give you. This way you can plan for a replacement or upgrade, rather than being taken by surprise one day when it'll no longer boot.

Sorry I don't have better news for you.

CT

 

[NerdyComputerGuy]

lNerdyComputerGuy,

The maximum recommended long-term Vcore for 22 nanometer processors is 1.300 volts. Your Core voltage is far too high, which has degraded your processor. You're seeing it first hand, and it is not reversible.

Excessive Vcore may result in accelerated "Electromigration" - https://www.google.com/?gws_rd=ssl#q=Electromigration

This prematurely erodes the traces and junctions within the processor's layers and nano-circuits, which will eventually result in blue-screen crashes that become increasingly frequent over time. As a rule, CPU's are more susceptible to Electromigration with each Die-shrink. However, the most notable exception is Intel's 14 nanometer Microarchitecture, where advances in FinFET transistor technology have improved voltage tolerance.

Here's the maximum recommended Core voltages per Microarchitecture from 14 to 65 nanometers since 2006:

When tweaking your processor near it's highest overclock, keep in mind that for an increase of 100 MHz, a corresponding increase of about 50 millivolts (0.050) is needed to maintain stability. If 70 millivolts (0.070) or more is needed for the next stable 100 MHz increase, it means your processor is overclocked beyond it's capability.

The only temporary fix for Electromigration is to further increase Vcore, as you've done, which will only accelerate the degradation process. The better alternative is to reduce Core speed and voltage, which no overclocker wants to hear.

I have a G3258 in my wife's rig which has been OC'd to 4.7GHz at 1.29 volts since I built it in early 2014. I have another in my niece's rig OC'd to 4.6 at 1.28 volts which I built for her in 2015. Both run cool on Hyper 212 EVO's. To get my niece's stable at 4.7 required more than 1.335 volts, which I was not willing to risk. The difference between 4.6 and 4.7GHz is less than 2.2%, which has no noticeable impact on overall system performance. It simply isn’t worth pushing a processor beyond recommended Core voltage and Core temperature limits just to squeeze out another 100 MHz.

A few days ago I saw another forum member with a G3258 OC'd to 5.0. I don't recall the voltage, but since it didn't catch my attention, it couldn't have been exceedingly high. It's the Haswell edition of the Silicon Lottery. He got one in the top few percent, I got a few that were decent, but unfortunately, your sample wasn't so good. Sorry, but that's just how it goes.

Your CPU is dying. If you don't want to be replacing it any time soon, then I recommend that you dial down the Vcore and Cache to no more than 1.300, and live with whatever OC that voltage will give you. This way you can plan for a replacement or upgrade, rather than being taken by surprise one day when it'll no longer boot.

Sorry I don't have better news for you.

CT

Thank you for the reply and since i am new to overclocking (pretty much) this is a breath of fresh air information, currently i have dialled the overclock down to 4.1ghz at core voltage of 1.375 and i think a cache voltage of 1.3v (cache ghz 3.6 ghz i think).

ive ran realbench, intel burn test and prime 95 and they have all passed.

I have also played ETS 2 and recorded / streamed for 2 hours plus without a blue screen so i think i've found a stable overclock and it may have been because i was at 4.2ghz and never tried any lower in fear of losing FPS.

Should i try going lower with the voltage or is 1.375v ok?

About a year or two ago i ran the CPU at 4.2ghz at 1.3v without any bluescreens even under heavy load, but since i was new to overclocking i didn't know to reduce the cache voltage lower than the core voltage.

How does this work then?

The core voltage and cache voltage both are different areas of the motherboard that give out voltage???

Thanks

 

[CompuTronix]

1.375 is still too far above the Vcore degradation curve for 22 nanometer processors. My best advice once again, is to regard 1.300 as the maximum recommended Vcore for long-term sustainability. I hope you're not running the stock cooler.

Concerning Core and Cache voltages, 4th Generation processors have a Fully Integrated Voltage Regulator (FIVR) on the substrate (green circuit board) under the Integrated Heat Spreader (IHS) which creates voltages used within the Die. How the individual voltages are allocated and which settings can be adjusted in BIOS varies between motherboard manufacturers, models and BIOS versions.

Let me put this in perspective for you ... I've been building and overclocking over 25 years, and have OC'd many 3rd and 4th Generation 22 nanometer processors, and have owned several. The only time I have ever exceeded 1.300 volts on a 22nm Ivy Bridge or Haswell processor was to run a few overclock benchmarks at 1.325, and then dial it back down under 1.300 for 24/7 usage. I also ensure that my rigs are cooled well enough so they never exceed 80°C.

Of all the overclocked rigs I've built over the years, those which still haven't been retired or replaced are running perfectly fine every day at the overclock and voltage settings I selected. One such rig I built for a friend is a Q6700 OC'd to 3.6GHz at 1.475 Vcore, which is a very good overclock. If you look at the chart I posted for you, the Q6700 is a 65 nanometer Core 2 processor which shouldn't exceed 1.500 volts. That CPU has been running OC'd in that rig since early 2008 problem free, with just a GPU and an SSD upgrade.

The point, which I can't overemphasize enough, is to keep a healthy perspective on overclocking ... don't get too crazy with it ... just stay within the maximum recommended voltage and temperature limits. The only way your processors will die is of obsolescence.

CT

 

[NerdyComputerGuy]

1.375 is still too far above the Vcore degradation curve for 22 nanometer processors. My best advice once again, is to regard 1.300 as the maximum recommended Vcore for long-term sustainability. I hope you're not running the stock cooler.

Concerning Core and Cache voltages, 4th Generation processors have a Fully Integrated Voltage Regulator (FIVR) on the substrate (green circuit board) under the Integrated Heat Spreader (IHS) which creates voltages used within the Die. How the individual voltages are allocated and which settings can be adjusted in BIOS varies between motherboard manufacturers, models and BIOS versions.

Let me put this in perspective for you ... I've been building and overclocking over 25 years, and have OC'd many 3rd and 4th Generation 22 nanometer processors, and have owned several. The only time I have ever exceeded 1.300 volts on a 22nm Ivy Bridge or Haswell processor was to run a few overclock benchmarks at 1.325, and then dial it back down under 1.300 for 24/7 usage. I also ensure that my rigs are cooled well enough so they never exceed 80°C.

Of all the overclocked rigs I've built over the years, those which still haven't been retired or replaced are running perfectly fine every day at the overclock and voltage settings I selected. One such rig I built for a friend is a Q6700 OC'd to 3.6GHz at 1.475 Vcore, which is a very good overclock. If you look at the chart I posted for you, the Q6700 is a 65 nanometer Core 2 processor which shouldn't exceed 1.500 volts. That CPU has been running OC'd in that rig since early 2008 problem free, with just a GPU and an SSD upgrade.

The point, which I can't overemphasize enough, is to keep a healthy perspective on overclocking ... don't get too crazy with it ... just stay within the maximum recommended voltage and temperature limits. The only way your processors will die is of obsolescence.

CT

Thanks for your in-depth answers, they are very informative and i am thankful that you can provide this much information to me.

Regarding cooling, i'm using the stock cooler yes and when running stress tests i've seen it go around 90c.

Overclock:

The overclock is now at 4.1ghz with a Vcore of 1.300v and a cache voltage of 1.250v.

I've ran Intel Burn Test on standard testings for 10 passes and it passed all of them, i'm currently running realbench right now for 15 minutes and will see if there are any blue screens.

After that i will try prime95 although i want to play my games so i'll probably only do it for about 20 mins.

 

[NerdyComputerGuy]

OK, so realbench passed and i've run prime 95 for over an hour and had no blue screens...

Should i try going lower with the voltage or is 1.3v fine for long term use.

I've had the CPU since late 2015 and had always ran it overclocked at 4.2ghz around there going from 1.3v to 1.450v, i don't know how i've got it stable all of a sudden, it's like magic and at such a low voltage.

 

[CompuTronix]

As I previously stated and as shown in the graph in my first post, 1.300 volts is the maximum recommended Vcore for 22 nanometer processors for long-term usage. However, as Core voltage drives Core temperatures, lower is always better.

The consensus among well informed and highly experienced system builders, reviewers and overclockers, is that cooler is better for ultimate stability, performance and longevity. So the objective is to achieve the maximum stable overclock without exceeding the maximum recommended Core voltage and Core temperature limits.

... i'm using the stock cooler yes and when running stress tests i've seen it go around 90c ... i will try prime95 ... for about 20 mins.

90°C is too hot. Nonetheless, Prime95 is the most ideal utility for testing thermal performance, and Small FFT's is the correct test to run, because it's a steady-state 100% workload with steady (not fluctuating) Core temperatures.

“Stress” tests vary widely and can be characterized into two categories; stability tests which are fluctuating workloads, and thermal tests which are steady workloads. Intel tests their processors at a steady 100% TDP (Thermal Design Power) workload to validate Thermal Specifications.

Utilities that don't overload or underload your processor will give you a valid thermal baseline. Here’s a comparison of utilities grouped as thermal and stability tests according to % of TDP workload.



All tests will show 100% CPU Utilization in Windows Task Manager, which indicates processor resource activity, not % TDP workload. Core temperatures respond directly to Power dissipation (Watts), which is driven by workload. Prime95 Small FFT’s provides a true and steady 100% workload, so if Core temperatures are below 85°C, then your processor should run the most demanding real-world workloads without overheating.

Here's the nominal operating range for Core temperature:

Core temperatures above 85°C are not recommended.

Core temperatures below 80°C are ideal.



Core temperatures increase and decrease with Ambient (room) temperature. "Standard or 'normal' is 22°C or 72°F.

Highest Core temperatures occur during stress tests or heavy rendering and transcoding, but are lower during less processor intensive applications. Gaming generally averages around 55°C, yet can range from 40 to 70°C, depending on how a particular game title allocates CPU / GPU workloads, as well as differences in overall cooling performance and Ambient temperature.

Also, many processors have a voltage "sweet spot" when overclocking, where too much (as well as too little) Vcore will result in instabilities. Once Vcore is edged up above the sweet spot, no amount of extra Vcore can make it stable; it will only continue to drive Core temperatures even higher, which in turn adversely affects stability.

You might want to read this: Intel Temperature Guide - http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html

CT