Overclocking a i7 4790k

[Rnazer]

 

Hi,

I have a i7 4790K with an Asus z97 A motherboard, EVGA 650W Gold PS, 2 x 8GB 2400 mhz DDR3 RAM, Noctua D15S cooler and 2 intake and 2 exhaust fans.

I was able to achieve a stable (using RealBench) overclock of 4.6GHZ at 1.215 CPU Core voltage. My temperatures get up to mid 70s under stress tests and high 60s in games.
I have tried going to 4.7 and i keep getting BSOD regardless of how high I up the CPU voltage (went as high as 1.35).
It seems strange that i would be stable at a relatively low voltage at 4.6 but not be able to reach 4.7 at all.  Perhaps there is another setting that needs to be changed. 

Is there any other setting I need to change in the bios? I currently have set it to manual with everything on default except the multiplier and the CPU Core voltage.

Thanks !

[JoostinOnline]

The amount of power needed doesn't increase on a linear scale. It's also down to the silicon lottery. 4.6GHz was really high when the 4790k came out. It's a significant overclock.

 

You could try lowering your memory frequency to 1600MHz. That's the only way I received a stable 4.7GHz.

[SnowyMus]

You can try your luck by pushing 1.4 V. This might be enough to get your CPU to 4.7 GHz, though you might find a big jump in load temperatures.

 

You can also try setting LLC (load-line calibration) if you haven't already. This may counter voltage droop when your CPU is under load.

[Jurrunio]

Mind the CPU input voltage, when overclocking it's usually set to 2V, no more no less.

[1van]

3 hours ago, Rnazer said:

a i7 4790K

it uses Intel's patented mayonnaise under the heat spreader, it will require a delid to replace the paste with a liquid metal to push the CPU beyond 4.6GHz.

Look it up, guides are available on youtube and elsewhere.

 

Another way to make an LGA1150 system live longer is a CrystalWell laptop CPU with an L4 cache, which is adapted to LGA1150 by handy people in China.

[JoostinOnline]

23 minutes ago, 1van said:

it uses Intel's patented mayonnaise under the heat spreader, it will require a delid to replace the paste with a liquid metal to push the CPU beyond 4.6GHz.

Look it up, guides are available on youtube and elsewhere.

Please check my signature. And no, delidding it won't suddenly allow him to get a stable overclock. He's not thermal throttling, he's blue screening.

[1van]

12 minutes ago, JoostinOnline said:

Please check my signature.

This is misleading. Most people who delid replace the Intel's thermal paste not with another thermal paste, but with a liquid metal, which has a thermal conductivity of 73 W/(mK) vs 12.5 W/(mK) for a high-end thermal paste, which is 6 times difference in thermal conductivity, and yes, the thickness of the layer is smaller, too.

 

I agree that a delid may not help in his case. I sold my 4790K long time ago and don't remember what voltages it required to run to get over 4.6 GHz which is a typical non-delid OC for these chips.

 

 

[Rnazer]

Thanks for the suggestions. I realize that you often hit a wall when overclocking but I thought because I was able to get 4.6 stable at relatively low voltage, I should be able to go higher. 

I will try increasing input vintage to 2.0. 

What is load line calibration and how do I set it?

 

Thanks again

 

[Herman Mcpootis]

7 minutes ago, Rnazer said:

 

I will try increasing input vintage to 2.0. 

What is load line calibration and how do I set it?

setting the voltage to 2v is only gonna cook your CPU and motherboard.

[GoldenLag]

15 minutes ago, Rnazer said:

I will try increasing input vintage to 2.0.

what is low voltage if 2.0 is fine?

 

 

2.0v is  in the region of instadeath for even 32nm CPUs and before. quite far back. 

 

edit: sorry, just noticed its not talking about Vcore. my bad. 

[Mick Naughty]

Got mine to 4.8. Think it was at 1.35 or so, can’t really remember. Was pretty hot too but didn’t wonna delid as it was gonna get sold as it couldn’t perform anymore. 

 

Had a fancy asus board with oc’s Paradise on it. Was quite easy though. 

 

 

[Rnazer]

23 hours ago, Jurrunio said:

Mind the CPU input voltage, when overclocking it's usually set to 2V, no more no less.

I tried setting it to 2.0 and still cant get 4.7 stable. 

Seems crazy that 4.6 is stable at 1.215 but 4.7 is not at at 1.35

[Fasauceome]

1 minute ago, Rnazer said:

that 4.6 is stable at 1.215 but 4.7 is not at all the way up to 3.5. 

You didn't set your CPU voltage to 3.5 did you?

[Rnazer]

9 hours ago, Fasauceome said:

You didn't set your CPU voltage to 3.5 did you?

No.  Fixed the typo.

[Fasauceome]

2 hours ago, Rnazer said:

No.  Fixed the typo.

thank goodness

[Rnazer]

On 10/30/2019 at 10:13 PM, Kavawuvi said:

You can try your luck by pushing 1.4 V. This might be enough to get your CPU to 4.7 GHz, though you might find a big jump in load temperatures.

 

You can also try setting LLC (load-line calibration) if you haven't already. This may counter voltage droop when your CPU is under load.

Do you know where this is done on the Asus Z97A?  I do not see an option for LLC.

Also even with slight voltage drops I would think 4.7 Ghz would be stable at 1.3 when 4.6 is stable at 1.215, no?

[k24_member101]

i got mine at a semi stable 4.8ghz itll pass stresstests and all and game smooth but every not so often itll just crash bsod or not and restart. games around 60c and stesstests push it to 73ish on highs. 

msi z97 gaming 5 mobo with evga clc. 

[SnowyMus]

3 hours ago, Rnazer said:

Do you know where this is done on the Asus Z97A?  I do not see an option for LLC.

I took a quick look at your motherboard manual. Here's the information for it.

 

 

 

Quote

Also even with slight voltage drops I would think 4.7 Ghz would be stable at 1.3 when 4.6 is stable at 1.215, no?

If you said it isn't stable at 1.35 V, it probably won't be stable at 1.3 V. However, if you use LLC, it might be stable at 1.3 V.

 

Note that load-line is there as a protection against overvoltage with voltage spikes when your CPU is under load. As a result, your voltage decreases when your CPU is under load, and this can result in instability when your CPU is under load. You can set LLC so you can get higher load voltages, but obviously you don't want to set it too high. For this particular board, if you need more than level 5 LLC to be stable, then it's probably not worth it.

[Rnazer]

3 hours ago, Kavawuvi said:

I took a quick look at your motherboard manual. Here's the information for it.

 

 

 

If you said it isn't stable at 1.35 V, it probably won't be stable at 1.3 V. However, if you use LLC, it might be stable at 1.3 V.

 

Note that load-line is there as a protection against overvoltage with voltage spikes when your CPU is under load. As a result, your voltage decreases when your CPU is under load, and this can result in instability when your CPU is under load. You can set LLC so you can get higher load voltages, but obviously you don't want to set it too high. For this particular board, if you need more than level 5 LLC to be stable, then it's probably not worth it.

Sorry I meant 1.35.

I tried applying level 9 LLC and setting the other options to extreme and still could not get stable 4.7 at 1.35.  I guess my cpu can only do 4.6 even if it is at low voltage.  thanks anyways.

[SnowyMus]

6 hours ago, Rnazer said:

Sorry I meant 1.35.

I tried applying level 9 LLC and setting the other options to extreme and still could not get stable 4.7 at 1.35.  I guess my cpu can only do 4.6 even if it is at low voltage.  thanks anyways.

As I said before, you could try pushing it to 1.4 V, but, yeah, it sounds to me that 4.7 GHz might be out of reach for your particular CPU, at least with that motherboard.

 

What constitutes a "stable" voltage can be really weird. On one CPU, a 6700K, I managed to get 4.5 GHz with stock voltage, but 4.6 GHz required somewhere around 1.38 V - not worth the extra heat if it's just another 100 MHz.