Considerations about my OC and whether to include water cooling
Considerations about my OC and whether to include water cooling
I recently assembled a new PC last weekend, though I didn’t plan any overclocking when I bought everything. I’m quite new to this kind of work, having only experimented briefly a decade ago and not since. Still, it seems like little has changed, which is a plus!
My components are:
- i7 7700k
- Asus Maximus IX Hero
- Corsair Vengence LED 32GB DDR4 3200
- EVGA GeForce GTX 1070 FTW
- Thermaltake NiC C5 CPU Cooler
- 120mm (likely) Case Fan x 5
- 200mm (likely) Case Fan x 1
I was hoping to get a 5Ghz performance, but so far I’ve managed only 4.9 Ghz with a -1 offset (running at 4.8 under full load) and a voltage of 1.32 volts. I use XMP to push the RAM to 3200 MHz, but haven’t overclocked it yet. The ambient temps were around 77-80°F, which is warm but not excessive.
During a RealBench stress test, temperatures reached a maximum of 91°C and averaged between 83–87°C. That’s significantly higher than ideal, so I’ll need to lower the overclocks to around 4.8 GHz with a 1.2 or 1.25 voltage. It’s disappointing seeing others reach 5 GHz at those levels.
My questions:
1. Is this an expected overclock for this chip without water cooling? It feels like I might have missed out on the best performance, but maybe it’s normal and many with 5Ghz opt for water or high-end air coolers.
2. Could water cooling really lower the temps enough to allow stable 5 GHz operation? From what I see, I’d need about 1.4–1.45 volts to run reliably at 5 GHz, and I’m not sure if even water cooling will suffice at those voltages.
3. My benchmark scores improved by around 30% when running at 4.9 GHz with the -1 offset, which reduced the clock speed to 4.8 during heavy load compared to running at the same voltage without the offset. That’s quite a difference—should I investigate further to find the optimal setting?
4. Am I right that these temperatures are too high? Should I aim for a lower overclock, keeping my max below 90°C and average temps between 75–85°C?
Is this a typical overclock for this chip without water cooling? It seems like I might have missed the chip lottery on this one, but perhaps it’s just normal and many users with 5 GHz overclocks are using water coolers or powerful air coolers.
I have vcore at 1.344 for 4.9. It might be that your chip isn’t a 5.0 model.
Would water cooling really make a noticeable difference to bring the temperatures down enough for stable 5 GHz operation? It appears I’d need around 1.4 to 1.45 volts to run stably at 5 GHz, and I’m not sure if even water cooling will suffice at those levels.
A custom loop (I doubt an AIO will be enough) could cool sufficiently, but it probably isn’t worth the effort. Check out the chart here for others’ setups and settings: http://www.overclock.net/t/1621347/kaby-...statistics
3. My benchmark scores improved by about 30% when running at 4.9 GHz with a -1 offset, which reduces clocks to 4.8 during heavy load compared to running at the same voltage without the offset. Is this typical? If yes, I’ll need to study further to find the optimal setting.
Memory and CPU issues at that frequency? Possible instability? I’m not sure.
4. Am I right in thinking these temperatures are too high and I should lower the clock speed so that my maximum stays below 90°C and average temps stay between 75-85°C?
Yes, until a better cooling solution is found. I currently have the NH-D15S, but ambient temperatures must be extremely high for my chip to reach those levels while running realbench. At 72°F, my hottest core reaches 73°C. Prime95 and IBT give different results!
Biglizard:
1. This seems like an unusual OC for this chip without water cooling. It looks like I might have missed out on the chip lottery, but perhaps it’s just typical for those with 5 GHz overclocks—many are using water coolers or powerful air coolers. I’ve set my vcore to 1.344 for 4.9 and maybe this chip isn’t a 5.0 model.
2. Would adding water cooling really lower the temperatures enough to allow stable 5 GHz operation? It seems I’d need around 1.4 to 1.45 volts to run reliably at that frequency, and I’m not sure if even water cooling would suffice at those levels. A custom loop might work, but it probably isn’t worth the effort. Check out the chart here for others’ experiences and settings: http://www.overclock.net/t/1621347/kaby-...statistics
3. My benchmark scores improved by about 30% when I ran at 4.9 GHz with a -1 offset, which reduced clocks to 4.8 during heavy load compared to running at the same voltage without the offset. That’s not unusual. If it does, I’ll need to study further to find the optimal setting.
4. Are my temperatures too high? Should I lower them to keep max below 90°C and average temps between 75-85°C? Yes, until I find a better cooling solution. My current setup has the NH-D15S, and ambient temps are quite high—around 73°C at 72°F. Prime95 and IBT show different results.
5. It’s interesting you’re running 4.9 GHz at 1.344 while I’m at 1.32 but with lower temps. What cooler are you using? The temperature in my home is a bit high due to a minor heat spike, but it’s only about 78-80°F. That shouldn’t significantly affect things, though I’m not sure. Could be a mistake on my part.
6. I double-checked and the peak temperature from a one-hour RealBench run was 91°C—not the 93°C I mentioned earlier. That’s still well above what I consider acceptable and far beyond the 73°C I’m reporting.
Biglizard confirms the issue persists until a more effective cooling method is found. The current setup with NH-D15S shows extremely high temperatures during realbench testing, reaching 73°C at 72°F. Prime95 and IBT report different results—72°F equates to 22°C, which is quite low. Ambient temps above 30°C are manageable, but anything over 35°C becomes more challenging.
Interesting, you're at 4.9 with a temperature of 1.344 while I'm at 4.9 with 1.32, but your temps are lower. What cooler model are you using?
Noctua NH-D15S, http://noctua.at/en/nh-d15s.html.
Nonsense :
Biglizard :
Yes, until a better cooling method comes along. I’m using the NH-D15S and the ambient temperatures need to be extremely high for my chip to reach that level while running realbench. At 72f, my hottest core hits 73c. Prime95 says otherwise—22C. That’s an incredibly low ambient temperature. Above 30C it gets pretty hot, and above 35C things start to get interesting. I’m sorry if your indoor temps are between 30 to 35c.
Do you usually disable HyperThreading? I just did and I think my jaw dropped, along with my temperature.
Before, at 4.8 GHz @ 1.25v I was reaching a max of 83°C with an average of 75-80°C. Now that I've found the right setting, it's time to aim for 5 again!
After some benchmarking, I noticed the scores without HT were as low as the temperatures, so I thought a lower clock speed would help maintain HT performance. That being said, further testing revealed a few points.
1) The best benchmarks on a stable system with manageable temps occurred at 4.9 GHz with a -2 offset and 1.31v, achieving a max temperature of 86°C.
2) RealBench results were highly inconsistent, showing scores between 110k and 145k after just one test after another. It seems this benchmark isn’t very reliable.
3) Cinebench gave more consistent outcomes, but only improved by about 9% compared to stock settings.
Conclusion: Overclocking this chip is not worthwhile due to the poor performance gain and potential wear. I’d stick with stock settings. FML.
I’m currently using PC Mark 8 and will check if it also shows a 9% improvement. If not, I might remove the CPU cooler, verify paste coverage, and try reapplying paste. If coverage was sufficient, I’d leave it at stock settings.
you are so funny - in a good way.
stock 7700K clock under load is 4.2Ghz.
you got your self 4.7Ghz under load.
500MHz is like 12% increase in clock speed. so here your 9% improvement which is perfectly fine.
I personally don't see a reason to bother with 100Mhz.
IRL, you can't tell the difference between 4.8 and 5.0 overclock.
You're quite funny in the right way.
The 7700K clock runs at 4.2Ghz under load, while you manage 4.7Ghz.
A 500MHz boost is about a 12% jump, which means a 9% gain—definitely acceptable.
I don’t see the point in targeting 100Mhz.
In practice, you can barely tell the difference between 4.8 and 5.0 overclock.
Since the chip includes a Turbo mode reaching 4.5GHz, I’d say it’s a 200MHz improvement over stock—only a 4.4% increase.
I guess a 9% boost on benchmarks is fine for a modest 4.4% overclock, but at most you’ll notice a 200MHz difference.
That’s why I think this POS chip isn’t worth pushing it too hard.
I’d expect around a 20% performance gain overall, not just a small number.
I’m not sure I’ll ever notice a 9% improvement in everyday use or gaming—probably just a tiny delay in apps or a few frames less in games. Not worth it.