Yes, restricting CPU speed can slow down performance and cause inefficiencies, especially under heavy workloads.
Yes, restricting CPU speed can slow down performance and cause inefficiencies, especially under heavy workloads.
I entered a discussion with a friend who believes manually setting a CPU to high frequency isn't wise, especially overclocking it in stable conditions. He argues that since his computer often sits idle, it's similar to revving an engine without moving it. However, my view differs. From what I've seen, keeping the CPU at a high frequency and voltage only slightly reduces power consumption during idle compared to its default state. Regarding clock cycles, there are benefits in having the CPU primed for peak performance when needed. The main drawbacks seem to be increased power usage and heat generation. According to Wikipedia on CPU clocks, locking frequency can indeed have an impact, but it's not a clear advantage. From my experience with the TR3960x, using locked voltages and clocks resulted in a noticeable boost—about double the frequency at 1.8GHz versus 3.6GHz all cores—while consuming only a small extra 10W during idle. This suggests that for high-performance scenarios, the trade-off is worthwhile, though it depends on your specific CPU model and workload. From an engineering and electrical standpoint, this approach can improve efficiency in active phases but may not be ideal for sustained low-power operation.
Other than stability checks or A/B cooler evaluations, there are no benefits. Switching clock states happens in nanoseconds due to Intel Speed Shift, and usually you might be sacrificing single-core performance since most contemporary CPUs run faster during one-core tasks.
It's not from 1990 anymore, keep your CPU at its default settings and avoid risky overclocking tools.
It comes with numerous drawbacks: high heat output, reduced lifespan, stability and performance concerns, and clock stretching effects.
I refer to the software as somewhat sketchy, but overclocking is typically handled through the BIOS (assuming supported chipset). I think this is likely the situation since I don’t mention many CPU-focused programs, and those that do usually require deeper integration into firmware (which should come from a trustworthy provider).
Sure it can, though most advantages are limited to comparison settings. Benchmarking alone rarely matters much in practical use since things like clock stretching add extra context.
You're lowering your CPU's clock speed to 4.0 GHz while keeping all cores active in the BIOS settings. This keeps it from reaching higher boost levels, maintaining low power usage and avoiding excessive heat or noise when cores aren't being utilized.
Modern processors offer numerous power reduction techniques, but lowering clock speeds is just one option. Without matching workloads, you won’t notice significant changes in power consumption. An idle chip might reduce internal bus rates, turn off individual cores or groups (CCX on AMD), aiming to conserve energy. The advantages lie in controlling the clock for performance goals—like optimizing scores, testing silicon quality, or managing timing with other devices. The downsides include increased heat, added system strain, minimal power savings under load, potential long-term instability, and possible hardware wear. You also risk damaging motherboard components and RAM if voltage settings are altered. Data reliability suffers too, as errors accumulate even with error correction methods. In practice, real-world benefits from overclocking are limited, possibly only a slight edge. If you possess the skills and resources to replace faulty parts, it’s worth considering—especially with a solid cooling solution. For most setups, sticking to default speeds is safer.
I adjusted the same setup on my TR. Set all core frequencies, fixed voltage for consistent performance and it runs smoothly with minimal power use. The main challenge was testing various voltages across different conditions and workloads. For everyday workstations, I monitor CPU activity and maintain optimal frequency and voltage settings to achieve peak efficiency. This approach delivers strong performance while maintaining stability—tests cover multiple scenarios, including AVX support and extended stress tests. It also improves thermal management, reduces cooling load, and lessens component strain overall. Yes, rather than defaulting to a high turbo setting that hits 280W TDP, I opt for locking all cores at 3.8GHz, accepting around 215W consumption for just a few extra watts when idle. For overclocking, I’ll push the 13900KS above 6GHz once the build is complete.
What? I disagree on many points except clock stretching, a term I don’t recognize well... Using this precise way of talking, the ratio of 13900K to 13900KS and even high-end AMD users should be completely out of the question in the OC/tuning community! There are plenty of gray areas in the tech world regarding unlocked processors. I’d categorize them into three groups: starting below the base clock, mid-range between base and turbo speeds, and the third for over-turbo settings. In the first two groups, I’m confident those issues aren’t real because the CPU stays within its limits but is locked in. For the last group... it’s a bit ambiguous. If the tuning is gentle, I think the locked speed and voltage won’t cause much wear over about five years, or so it usually lasts before a part needs replacing. The wear will likely increase with more extreme tuning and less efficient cooling. For instance, my 3960x runs at 3.6Ghz under the base speed of 3.8Ghz but stays locked on all cores at the lowest voltage for extra stability (my setup is 1.0v). In idle it uses only a few extra watts (around 80), compared to if it ran at 1.8Ghz (around 70). That difference is almost insignificant in terms of heat, lifespan, or performance stability.