VCCIO/SA voltages for i9-10850K OC
VCCIO/SA voltages for i9-10850K OC
Checking the suitability of the voltages depends on your needs. At 1.26-1.278 V for VCCIO, you're aiming for a stable setup under heavy load with high-core count and a higher core voltage. The 1.336 Vcore and 2x8 GB RAM at 4000 MHz CL15-16-16-36-52-770-2T should work well if the BIOS settings are correctly adjusted. The SA voltage you're using (1.32) is close but may cause instability under full load, so consider bumping it slightly higher or verifying stability. The IO voltage seems a bit low for your configuration—raising it could improve performance and reliability.
I was attempting to reduce the Trfc voltage but even minor changes didn't hold steady at 1.5V. I also tried CL15-15-15, though it doesn’t work for my setup. What alternatives would you suggest for RAM settings? I think this is the optimal range for better performance since XMP isn’t reliable here, especially with 10th generation chips and this particular memory type. I’m not using advanced BIOS options like many timings—just sticking with the preset values.
First thing is tRAS, it should handle 28 without issues, and tRC can operate a bit lower than 52. Increasing VCCIO and VCCSA might bring the tRFC value below 770, which is quite high. Most kits I've worked with manage 300 at 3800MHz, especially since my overclocking experience leans toward AMD. It would be surprising if you can't reach something closer to 350-400 by adjusting those voltages.
I’m using an MSI memory try it preset in BIOS for my RAM since I don’t want to adjust all the timing settings myself—it’s complicated for me. I’ve also heard that Trfc and lower timings aren’t too risky and won’t significantly boost performance. What do you think? Should I try those? Could I increase voltages further? I notice my system is near stability limits, and Cinebench R23 shows a high temperature. Also, my current settings cap at a 4.6x CPU cache ratio, which might affect overclocking possibilities.
I noticed on my 7700k tweaking some secondary timings improved memory benchmarks by about 5-10%. Copying standard BIOS settings might help reach peak performance. If you're investing in B die memory like the Patriot Viper kits, tune them similarly—most sub-timings should stay low. In R23, using excessive voltage for cooling is inefficient; the stress test is manageable, but if temps exceed 80-85°C, cooling won’t support stable operation.
I understand I have strong CPU cooling thanks to the Fractal Celsius+ S36 Dynamic setup. It seems my chip struggles with high temperatures during overclocking, especially under stress tests like R23. You mentioned it’s not a typical hard test and that it generates a lot of heat, which matches what I observed—my processor stays around 80-85°C during those sessions, sometimes spiking to the 90s before throttling. In contrast, my room stays cool, and the cooler is well-mounted with good thermal paste. The sudden temperature jumps only appear in Cinebench, suggesting something specific in the stress conditions. Despite this, other tests show stable temps near 85-88°C for hours, which is normal for everyday use and gaming. My system has remained stable continuously for about two weeks now. Appreciate your prompt replies!
If you're facing the toughest stress tests, it might mean your hardware isn't handling them properly—possibly due to thermal throttling, incorrect AVX settings, or an AVX offset issue. Benchmarks like Linpack Xtreme and Prime95 should push more load than your chip is currently experiencing. Cinebench typically outperforms others by 200-300MHz, with world records reaching up to 7.04GHz for certain configurations. Monitor the actual core voltage via HWiNFO64, not just VID. It seems your system runs normally under normal conditions but spikes unexpectedly during testing, possibly because of cooling limitations. With strong cooling, a 5.3-5.4GHz Cinebench single pass should be achievable.
I was actually executing the most intense and demanding stress tests available, including those you referenced (linpack xtreme, prime95 small ffts) and more. My chip wasn’t throttling under any circumstances, I kept monitoring everything via hwinfo64. I didn’t have any AVX offsets or similar features; everything ran at full capacity. I checked the Vcore directly—I’m familiar with voltage checks—and it showed normal readings, no spikes or sudden changes. It might be a BIOS-related issue that isn’t captured in hwinfo, possibly affecting certain CPU settings. Running a single Cinebench R23 pass at 5.1 GHz across all cores was too power-hungry, nearly causing overheating. I suspect the CPU could safely run at 5.0 GHz under constant stress, which would be unusual for this model. It’s possible the silicon itself is subpar, especially since it was already rejected during manufacturing at 10,900K.
I’m going to challenge the idea that R23 is too light. Most users won’t run their PCs at even a fraction of its potential, so it’s mainly useful for quick checks. I also disagree with the claim that stability with certain specs means it’s not a real stress test. As with any product, limits come from how it’s meant to be used—not just theoretical limits. If every manufacturer and laptop brand treated it as stable under all conditions, they’d all face issues. They’re just not suited for everyone’s needs.
For instance, a regular car transmission isn’t built for long hauling heavy loads, even if it seems capable. It doesn’t match the demands of its intended role. If a truck can’t handle towing a trailer for six hours, that’s not a flaw—it’s just outside its normal scope. Similarly, rifle parts don’t need military-grade specs unless you’re in special service. For everyday shooters, it’s fine as long as it performs within expectations.
In tech terms, compact designs usually trade off performance for efficiency, often causing thermal throttling. On high-end CPUs, pushing them hard can hit a thermal ceiling, leading to reduced performance or shutdowns. This doesn’t mean the product is unstable—it just means it’s operating near its limits in a realistic scenario.
Most people aren’t using their systems at extreme levels, so they don’t push them beyond what R23 can handle. For gamers and multitaskers, stability isn’t always the top priority. If you’re comfortable with the expected behavior, it’s fine. But if you want to push it further, you’ll need to adjust your expectations.
My take is that most users are casual or light users. They should focus on what matters for their needs rather than chasing absolute stability. R23 works well for testing maximum CPU temperatures in typical use cases. It’s not perfect, but it’s reliable enough for the average person. If you’re unsure, stick with what fits your situation.