Overclocking Help?

[Jdaithesecond]

So I have an i7 7700k currently overclocked to 4.5 ghz and 4.8 ghz boost with 1.235 volts. (using intel xtu)

Cooling will be an issue for any larger overclocks, because I only have an h60i (single radiator AIO)

 

1. Is the voltage safe for my current clock speeds

2. Is this OC safe in regards to temperatures

3. Should this OC be stable (ran 5 minutes of XTU stress testing)

4. Is there a way to stably increase OC slightly without raising temperatures (currently 65-70 C under load)

 

Thanks

 

[knightslugger]

1 minute ago, Jdaithesecond said:

2. Is this OC safe in regards to temperatures

70*C is boarderline. But it is well known that the TIM Intel uses on the 7th gen CPU is absolute garbage. The only way to get better temps is to void your warranty, delid the CPU, and use a gallium based liquid metal replacement, or at least a better paste.

 

1 minute ago, Jdaithesecond said:

1. Is the voltage safe for my current clock speeds

yes.

 

1 minute ago, Jdaithesecond said:

3. Should this OC be stable (ran 5 minutes of XTU stress testing)

hard to say, you've run one test so far for 5 minutes. Did the water temp stabilize?

 

4 minutes ago, Jdaithesecond said:

4. Is there a way to stably increase OC slightly without raising temperatures (currently 65-70 C under load)

not really, not to the point of actually noticing a difference outside of margin of error. You could try bumping clock speed without touching voltage, but that's probably going to crash.

[Archemis.]

16 minutes ago, Jdaithesecond said:

 

3. Should this OC be stable (ran 5 minutes of XTU stress testing)

 

 

run the stress test for a day or do after you feel like you're done overclocking judt to make sure it's stable

[steffeeh]

1. Generally, when you go above 1.3V you enter the greyzone, when you get to 1.35V you're at the recommended maximum among most 24/7 overclockers.

Though I've seen people go even higher, but that's only if you're well experienced. If you're new, settle down for 1.25V-1.3V for the moment until you gain more experience - if your thermals allow it.

 

2. Depends on what test you ran to check the temperature. There is more or less a given standard for testing outlined here, but many consider it overkill so there are many preferred methods.

I personally fire up my most heat delivering realisitc PC usage, let it run for 10 min to warm the PC up, and then monitor the temperature for 10-15 min.

Then look at the calculated average temperature on the cores, while also looking at any potential nasty spikes.

For Intel, 80C is generally the recommended maximum temperature.

 

3. 5 min is too little. What stress test to use is personal preference, but stick to 10 min minimum when checking if an OC increase is stable. Maybe 30 min if you're skeptical.

Then once you've settled for a final OC and want to see if it's long term stable, fire up a torture test like Prime95 (with desired FFT size and software version) for several hours - if that passes, you're most likely good to go.

Or you could do like lazy me and just have 2 week testing period with regular usage - if it passes, then it's probably stable too... buuuut that's not the regular way to do it

 

4. No, not in that manner. All chips react differently, so for instance some extra voltage at a specific level might not do much for heat, but add just a hint more and OH MY GOD THAT'S TOO HOT

[RAM555789]

Honestly 1-2 hours of stress testing should allow you to reach and equilibrium point between the water cooler and the CPU temps.

[Jdaithesecond]

57 minutes ago, knightslugger said:

70*C is boarderline. But it is well known that the TIM Intel uses on the 7th gen CPU is absolute garbage. The only way to get better temps is to void your warranty, delid the CPU, and use a gallium based liquid metal replacement, or at least a better paste.

 

yes.

 

hard to say, you've run one test so far for 5 minutes. Did the water temp stabilize?

 

not really, not to the point of actually noticing a difference outside of margin of error. You could try bumping clock speed without touching voltage, but that's probably going to crash.

 

42 minutes ago, steffeeh said:

1. Generally, when you go above 1.3V you enter the greyzone, when you get to 1.35V you're at the recommended maximum among most 24/7 overclockers.

Though I've seen people go even higher, but that's only if you're well experienced. If you're new, settle down for 1.25V-1.3V for the moment until you gain more experience - if your thermals allow it.

 

2. Depends on what test you ran to check the temperature. There is more or less a given standard for testing outlined here, but many consider it overkill so there are many preferred methods.

I personally fire up my most heat delivering realisitc PC usage, let it run for 10 min to warm the PC up, and then monitor the temperature for 10-15 min.

Then look at the calculated average temperature on the cores, while also looking at any potential nasty spikes.

For Intel, 80C is generally the recommended maximum temperature.

 

3. 5 min is too little. What stress test to use is personal preference, but stick to 10 min minimum when checking if an OC increase is stable. Maybe 30 min if you're skeptical.

Then once you've settled for a final OC and want to see if it's long term stable, fire up a torture test like Prime95 (with desired FFT size and software version) for several hours - if that passes, you're most likely good to go.

Or you could do like lazy me and just have 2 week testing period with regular usage - if it passes, then it's probably stable too... buuuut that's not the regular way to do it

 

4. No, not in that manner. All chips react differently, so for instance some extra voltage at a specific level might not do much for heat, but add just a hint more and OH MY GOD THAT'S TOO HOT

Thanks guys, I'll run stress testing for like an hour soon 

[knightslugger]

1 minute ago, Jdaithesecond said:

 

Thanks guys, I'll run stress testing for like an hour soon 

other stress testing programs that i like include ROG's RealBench.

[Brett_Bst]

I would keep it as is and stress test it to see what the temps hit after a few hours, the 7700k is a hot chip and the more you push the hotter it will go, I had the h60i for a while but upgraded to the H115i right now I'm at 68c max under prime95 load for 2 hours @5.0Ghz @1.330 volts

[Jdaithesecond]

 .

[Jdaithesecond]

7 hours ago, steffeeh said:

1. Generally, when you go above 1.3V you enter the greyzone, when you get to 1.35V you're at the recommended maximum among most 24/7 overclockers.

Though I've seen people go even higher, but that's only if you're well experienced. If you're new, settle down for 1.25V-1.3V for the moment until you gain more experience - if your thermals allow it.

 

2. Depends on what test you ran to check the temperature. There is more or less a given standard for testing outlined here, but many consider it overkill so there are many preferred methods.

I personally fire up my most heat delivering realisitc PC usage, let it run for 10 min to warm the PC up, and then monitor the temperature for 10-15 min.

Then look at the calculated average temperature on the cores, while also looking at any potential nasty spikes.

For Intel, 80C is generally the recommended maximum temperature.

 

3. 5 min is too little. What stress test to use is personal preference, but stick to 10 min minimum when checking if an OC increase is stable. Maybe 30 min if you're skeptical.

Then once you've settled for a final OC and want to see if it's long term stable, fire up a torture test like Prime95 (with desired FFT size and software version) for several hours - if that passes, you're most likely good to go.

Or you could do like lazy me and just have 2 week testing period with regular usage - if it passes, then it's probably stable too... buuuut that's not the regular way to do it

 

4. No, not in that manner. All chips react differently, so for instance some extra voltage at a specific level might not do much for heat, but add just a hint more and OH MY GOD THAT'S TOO HOT

So I ran a stress test for 10 minutes and it crashed around minute 8 (which theoretically shouldn’t be happening (?)) because I only had it overvolted to 1.235 and running at 4.5 ghz with 4.8 ghz turbo boost.

(temperatures were at like 60-75 until I left to go get something and when I got back blue screen ☹)

[steffeeh]

All chips are different in how far they can overclock. What stress test did you use? You could try and up the voltage to 1.245V and see how it goes for thermals. Do you overclock in the BIOS and with fixed voltage?

Out of curiosity, what motherboard do you have?

[knightslugger]

5 hours ago, steffeeh said:

Do you overclock in the BIOS and with fixed voltage?

no he uses the Intel software, which honestly is NOT the way to OC.

[Jdaithesecond]

15 hours ago, steffeeh said:

All chips are different in how far they can overclock. What stress test did you use? You could try and up the voltage to 1.245V and see how it goes for thermals. Do you overclock in the BIOS and with fixed voltage?

Out of curiosity, what motherboard do you have?

I'm on an ASUS Maximus IX Hero. I used the stress test built into Intel XTU. I will try and oc with the bios.

9 hours ago, knightslugger said:

no he uses the Intel software, which honestly is NOT the way to OC.

so you're suggesting BIOS overclocking right?

[knightslugger]

Yes I am.