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.
the only thing that comes to mind is achieving more stable results since cores don’t fluctuate too much in speed and you can reach higher clock speeds when using all cores. It seems unlikely you’d hit 4.7-4.8ghz with all cores active, though I haven’t tested it closely. The main drawbacks would be idle performance and temperatures, which probably won’t make a big difference. I was thinking about balancing power usage and performance—around 5 to 10 watts versus 10 to 20 with all cores on. Under heavy load, you’d see higher frequencies or lower power consumption depending on the configuration. Compared to PBO setups, this could be a real advantage. The main downsides are idle draw and thermal readings, but those effects are likely minor. What I was about to mention: when things get busy, you’ll notice more stable frequencies or better efficiency, especially with power-saving features still active. On the other hand, 30 to 40°C under load isn’t huge, though it’s worth noting that single-core speed is still limited. The main caveat is single-core performance—it won’t give a big boost. Overall, the trade-offs seem manageable if you’re willing to experiment and tune things properly.
Try OC on a real chip with full Northbridge voltage control. Those Core 2 units still exist but are standard. A few older series suffered and deteriorated over time. I’ve broken a couple of them along with Athlons back then. Running at 1.4V is minimal; in the 90s we used 3.3V stock, which was tough to manage—only FSB, voltage, and sometimes unlocking the multiplier. Damaging them was straightforward. Today’s chips are more forgiving, but pushing them hot continuously won’t help much. I managed a K6-2 once and reached 433.25 MHz, still functional for XP and late ’90s games. The Q6700 stable at 3.6 on water worked well for a while, though I had to lower clocks about two years ago; it stopped working a couple of months later. It ran its whole life, even handling streaming tasks like YouTube and Twitch. You might have better results elsewhere. By the way, the Cedar Mill chips weren’t available, even at low prices—only Prescott models were accessible, which didn’t handle higher clocks well.
I believe we're on the same page but I'm not an expert, so I thought it would help to get opinions from others. The issue is that my friend misunderstood the concept of locked core voltage—it doesn't indicate high power consumption during idle like it does under load. Voltage stays fixed, but actual load determines how much current flows (IOUT) in amps. Most people don’t fully grasp this. I’ve learned this recently after years working with BIOS and unlocked CPUs. Honestly, you might be right. Recently I built a custom water-cooled PC for a student, and it was surprising how tough it was to keep the 13900K running cool in a 2 x 360 loop. As someone who uses AMD hardware, I was particularly taken aback—my system would throttle constantly during stress tests even though it was in stock. On the other hand, as an enthusiast, I had the opportunity to get a professionally built 13900KS with a sealed directdie and supercool technology. I plan to test it on a Mora 420 to see if it can maintain performance at 6GHz+ on P cores and 5GHz on E cores (though I’m hoping for better results).
Pre 775 and am2 are now quite scarce here in India, so I won’t be able to snag any of those vintage chips. Still, 20 years back wouldn’t feel the same as it does now—CPUs have changed a lot. Not old enough to remember physical tweaking or to know the right tools for modern boards. I don’t have the expertise to run a basic voltmod on today’s processors, which is why those older chips seemed tougher than they actually were. I’ve tested 1.8V through some Wolfdales before, ran a 5GHz BIOS, and got around 70°C in BIOS—even at 1.85V with an E5800, it stayed stable at about 85°C idle. Only the I3 540 failed so far, but honestly, a chip worth $2 and a mid-sample only capable of 4.5GHz at 1.5V is pretty cool. I’ve also booted 5GHz on an X5660 without any issues. So far, the biggest loss has been the I3 540. Still, it’s surprising that something this cheap could last so long with such low voltages. I’m in a rough middle ground—no access to new hardware, but I’m not letting old standards scare me off. For now, I’m sticking to 1.35V VTT max, up to 1.6V for 32/45nm, and 1.7V for 65nm. The results above suggest maybe 1.65V VTT is too much unless the core isn’t running at 3GHz. I’ll keep testing desync uncore memclk on Nehalem after midterms. It’s wild what I’m seeing. In contrast, Sandy Bridge seems like a total mess compared to Nehalem. So far, only the I3 540 has failed, and even that might be due to fragility. I’ve seen Ryzen 3100 and 3300X handling 4.5GHz at 95°C without issues, so times have definitely shifted. I’m not worried about safety anymore—just go with what works.
Yup, 1.41 got it stable, my sample was not that great, needed more voltage to be be stable. Tried for more but couldn't get it to post. It used to be a file server and last was used as a steam cache / game server. After the CPU went up in smoke (SMDs shorted out on the CPU), I've replaced it with a conservative Q9550 with a mild 3.166 GHz OC on it, just because it would be a waste of a perfectly good motherboard otherwise. My best results were on a P4 2.4 that went to almost 3.0 (2.9 something with the multiplier) stable with the Thermaltake Aquarius external watercooler. Still have the unit in storage but the pump was leaking and it has absolutely not head pressure whatsoever compared to a DDC or D5. Modern CPUs are not as much fun anymore to OC, too much safety like modern cars, maybe I should get my hands on some LNG.... that would be fun. Anyway good luck to all your OC adventures out there.
That's the point—chips will wear out over time, and it depends on how cautious or experienced you are. Generally, it's safer to undervolt now rather than overclock, whether for GPUs or other components. ^This is more about software behavior than hardware itself; it makes sense!
The smds often seem to cut themselves off unexpectedly. It probably isn't related to the OC; my e5645 died and it happened while idle on a table. Once I turned it back on, it heated up extremely quickly—almost instantaneously, burning hot in just one second. A normal working 6-core Westmere would take three to five seconds. The temperature rise is wild, and it’s clear the OC isn’t handling high loads well. Even with lower voltages, newer chips aren’t as robust as older ones, which I noticed when I tried running 1.6/1.7V without issues. I can’t push beyond about 1.5V without noticeably losing performance. Heat management is a big issue too—cooling would be necessary for any real use. Setting up an OC is complicated and not worth it unless you’re really committed. The 2900 screen seems to need more than just a basic setup; it’s a significant upgrade. It’s affordable because you only need a simple bare PCB with the right chip, making it cheaper than buying regular high-end rigs while offering better speed. There’s also a real challenge in fine-tuning the timing and frequencies, especially at higher speeds, but performance at those extremes can be tricky.
Companies are overdoing it with stock levels and conservative voltage settings to push sales of lower-quality components. They’re being very lenient with voltage, but older hardware still holds up well even at 1.7V. I’ll need more testing to see if any issues arise, especially since there’s a cheap W3680 available or I can just keep testing Nealem because RAM speeds are fine without a reliable ASUS board that slows it down.
This method also offers performance through undervoltage control. My setup increases the maximum frequency across multiple cores (a claim often made online) and operates more efficiently, consuming slightly less power.