Upgrade the i7 7700K using Phanteks TC12DX driver.
Upgrade the i7 7700K using Phanteks TC12DX driver.
I recently purchased an i7 7700K built on an MSI Z270 Gaming M5 board, which is cooled by a Phanteks PH-TC12DX. I’m curious about the potential for overclocking this setup. Is it feasible to extract performance gains? If yes, what are the best safe settings? Also, do you suggest using the MSI app for overclocking? Thank you in advance.
Manual overclocking via the BIOS is generally the favored method, whereas auto or preset overclocking settings on the motherboard or software often push core voltage too high. These presets usually estimate values that are more likely to be stable rather than optimal, and each CPU behaves differently. Some units might run smoothly at 1.2v vcore, while others could need 1.32v to hit the same overclock level. If your CPU consistently runs stable at 1.22v, the extra voltage mainly contributes to increased heat generation.
You cannot boost a CPU using "the MSI app." You must access BIOS and typically adjust just the multiplier and voltage if necessary. A safe setting is around 80°C temperatures, and it's important to perform detailed tests after each change, such as running Prime95 for about 30 minutes to check stability.
You're not entirely right. MSI offers two distinct Windows tools that support overclocking:
1) MSI X-Boost
2) Intel Extreme Tuning Utility
Overclocking can also be achieved through the BIOS settings.
I agree I wouldn't let the CPU exceed 80 degrees Celsius often, as it would reduce its lifespan.
You can find numerous overclocking guides online; however, I prefer BIOS adjustments for optimization, since they tend to be more power-efficient.
Raising the multipliers and voltage is one method. On my i7-3770K, I only increase the multipliers until instability occurs. I didn't adjust the voltage because it significantly raised the temperature (voltage has a greater impact on temperature than frequency; if you can stabilize by boosting frequency without increasing voltage, then voltage shouldn't be changed). Still, occasionally a small voltage boost is necessary, provided the temperature remains stable under load.
PRIME95 works for testing but is more demanding than realistic conditions, so I'd consider a 80-degree Celsius worst-case scenario using tools like HANDBRAKE (must run near 100% for at least 5 minutes).
Otherwise, some CPU temperatures are actually from sensors adjacent to the CPU, so verify the correct readings. I personally use CORE TEMP, though alternatives exist. MSI likely has suitable solutions on their motherboard page.
My goal is around 4.6GHz maximum under full load (all four cores near 100%). The cooler is fine, but I suspect cooling capacity is limited, making frequency adjustment and avoiding voltage changes the best approach.
Manual overclocking through the bios is usually the preferred route, ez/auto overclocking presets via the motherboard or software tend to be aggressive on the core voltage. They guesstimate what will more than likely be stable rather than what's efficient and each cpu is different. Some may be stable at a given speed using 1.2v vcore, others may take 1.32v vcore to achieve the same overclock. If your actual specific cpu is stable at 1.22v then the additional vcore is just adding to the heat it produces.
p95 v26.6 is usually the preferred version of prime 95 to run on haswell and newer intel cpu's while running the small fft's test. It's a static load which makes getting temp readings a lot easier than a cyclic or fluctuating test. It is generally more stressful than regular programs which is the point, to provide a worst case scenario. If it were a light load test and the cpu looks ok only reaching 74c it doesn't do much good if by chance you run some real world program more demanding and come to find out it's actually running at 89c under heavy loads.
Prime is more of a measure for thermals rather than system stability. There are better solutions out there like rog realbench which run a whole suite of stress tests and are more likely to reveal any weaknesses since it stresses the cpu, cpu+ram, gpu, many of the subsystems the way they'd interact in real world programs. It's possible to have a cpu run p95 stable for 30min then run realbench or ibt and have it error out within 5min.
How far you can oc on that cooler depends on a number of variables. You'll just have to test and slowly increase the oc, slowly increase the vcore in small increments if the oc isn't stable and watch temps under stress tests. Your ambient room temps will play a role, if your pc room is 30c it will be more difficult to keep the cpu cool than a room that's 22c. Case cooling will have an impact, if you only have a single exhaust fan it will be more difficult to keep the cpu cool than if you had several intake and exhaust fans with good case circulation.
It will depend on your actual cpu, if it's a more 'golden' chip it will oc while using lower core voltage. If you didn't win the silicon lottery it could take a fair amount of core voltage to just get 100-200mhz above stock speeds, or anywhere in between. If your particular cpu has an efficient amount of factory thermal compound between the cores and the ihs (metal cover on the cpu), it will be easier to cool. If it's one that got an extra thick or uneven layer of factory thermal compound under that ihs lid then it may prove difficult to cool or might be a candidate for delidding.
Each and every one of those variables comes into play determining how well that cooler will work and what sort of oc you can expect to achieve. There's no hard rule for 'use this cooler and you'll absolutely get this overclock in this temp range'. Overclocking can be a roll of the dice and you just won't know until you try and see what results you get.