Does a stable OC offer significantly reduced vcore compared to the Silicon Lottery's estimate?
Does a stable OC offer significantly reduced vcore compared to the Silicon Lottery's estimate?
I am a new OC with an 8700k processor that made it through Silicon Lottery. They rated it stable at 5.1Ghz with 1.412 vcore and -2avx offset. That definitely got very hot. I’ve since adjusted to a 5Ghz setup with 1.35vcore and -1avx offset at medium settings. So far, the system has passed an hour of realbench stress testing, a one-hour Prime95 small FFT without AVX enabled, and another hour of Prime95 blend tests. It appears stable so far, though I haven’t run it for multiple hours yet. The temperatures stay in the low to mid 80s to high 80s.
My main concern is why my vcore is significantly lower than what Silicon Lottery suggested, even though my clock speed is only 100mhz slower. Does that extra 100mhz really need a full 0.6v for stability?
I feel like I might be missing something here because these experts seem to expect a much higher vcore. My settings are working fine, but it’s puzzling.
Below are pictures of my current configuration. Any advice or suggestions would be appreciated?
If you're working with offsets, it could alter the situation, especially if you're testing without them but intend to use them later. This would likely prevent AVX from reaching the same heat levels as when it's active. For lasting performance, I'd never push a daily driver beyond 85°C. The intel temperature guide notes that thermal issues and VT shifts can cause this problem.
The challenge with overclocking lies in balancing your goals for the hardware and the demands you place on it continuously. You need to assess whether the benefits justify the risks. As you've experienced, pushing the processor to higher speeds like 100MHz demands significantly more power than lower speeds would require, which brings us to the concept of diminishing returns specific to that chip. In fact, it's generally advised to keep voltages below 1.3v if you plan to use the chip for an extended period.
Heat also negatively impacts performance, and while a proper air or water cooling system can help, many overlook how high voltages affect the internal components of the processor and board. This damage becomes irreversible once reached.
Would you consider lowering the voltage slightly while reducing memory frequency to 3200MHz? If you're aiming for record speeds, that's acceptable. However, if your project is intended for everyday use, it's better to prioritize stability over maximum performance.
-I'm confused about why lowering the RAM speed would help. Does it actually reduce stress on the CPU memory controller? I recently purchased this RAM and sold my G-Skill 32GB 3200MHz kit to make room for the D15 in my system, so I could perform overclocking. What advantage does slower RAM bring here? -I don't want to risk damaging my system or causing major performance drops. It's my main computer but I still want some improvement. Over the past three years, I've been using these chips without much issue. I think I'll probably upgrade again next year after Alder Lake is released. -How sure are you about that 1.3 volts? I saw this guide online: https://forums. it shows a graph indicating that 1.4V is the point where degradation starts for 14nm chips, while 22nm chips might risk it at 1.3V according to that source. -My main worry is data loss or corruption. I need to be absolutely sure my system stays stable and avoids any problems. My daily tasks include music production, light photo editing, and gaming.
Yes, reducing the memory speed does lessen the strain on the IMC, which is located within the CPU. This generally helps minimize some causes of CPU overheating or instability, though I never suggest doing this unless absolutely required. It's better to keep your memory at its designed frequency, timings, and voltage, and only slightly lower your CPU overclock if necessary.
For the information from your PM, the only advice I could give is whether you carried out the extensive testing procedures they suggested in my overclocking guide. If you did, maintaining a voltage below the recommended maximum for Intel temperatures, staying within the thermal limits, and avoiding errors during testing would mean you're safe to proceed—provided you don’t see any other issues. Every CPU behaves differently, and changing settings can affect performance, but I still recommend following the ITG guidelines.
I’m not very confident about stability with these settings, but it’s doable. You should test it yourself. The results will show what works best.
There are certain configurations that make you question stability, mainly the clock and voltage settings. I’m less sure about others, some I’ve turned off in previous tutorials.
If you're working with offsets, it could alter the situation somewhat. However, if you're conducting tests without offsets but intend to apply them after testing—since using AVX in games or apps would likely prevent reaching the same thermal limits—then it might not be necessary.
For lasting performance, I would never push a daily driver beyond a temperature of around 85°C. According to the Intel temperature guide, you'll discover that due to thermal effects and voltage adjustments, the recommended threshold is lower:
This should only be set during testing. Avoid pushing your CPU to reach 95°C during tests because you're uncertain it will ever hit that level in real use. In fact, many have observed this issue. If you were simply experimenting without concern for longevity, then go ahead—but don't apply these settings to any important or long-term system. My advice is based on experience and the Intel temperature guide, which I consider a reliable source for processor data. Of course, your CPU is ultimately in your hands.
We provide these suggestions based on observed results and extensive testing, especially from Computronix. This is just my perspective, supported by various resources. Ultimately, it's your decision whether to prioritize safety or performance.
Keep in mind that unless you're constantly monitoring CPU temperatures, you can't be certain you'll encounter situations where the system exceeds normal loads. That's why we suggest maintaining a moderate buffer.
You should check the advanced settings on the tweaker panel—there should be an option for frequency per core. It's also wise to turn off Intel Speedshift, enable Intel Speedstep, and either leave C-states enabled or disable them. This is largely a personal choice, but it often results in lower peak temperatures under similar conditions. CPUs can jump from 800MHz to over 5GHz quickly, so noticeable changes are possible. In most cases, cores will reduce their speed briefly when idle, which usually happens in microseconds rather than causing consistent lag.
These are my recommendations. Your preferences may differ. I have some concerns about stability with these adjustments, but I encourage you to verify them yourself. Performance gains can be significant, especially under continuous load, and modern CPUs handle such changes smoothly—often without noticeable delay.
Thanks for all the assistance and details everyone provided. It was a couple of weeks filled with effort to try to figure things out. I consistently received "stable results" at lower vcore, yet I was taken aback by how minimal the performance improvements were in actual scenarios. After reading posts from various OC forums where people mentioned encountering failures during the 47th hour of prime95 runs or similar issues, I realized that this barely noticeable gain doesn’t justify the risk to my system’s stability. Since this is my main tool for daily tasks, I won’t invest much time in verifying its reliability. I might try it again once I transfer all important work to a new system and focus on playing games on this one.