Vcore dropping formula
Vcore dropping formula
I am evaluating a 9900k on a z390 aorus pro with a current 4.9 overclock and -2 avx offset. The required vcore for running prime95 avx workload for an hour is 1.32. There seems to be no standard formula provided, but I can help you adjust the parameters if needed.
photonboy :
Additionally, even though Prime95 might not be realistic, I find it a solid sign that problems should be manageable if it succeeds. It could be an overestimate in terms of demand, but passing it would still be preferable, especially if the temperature exceeds what I'd consider ideal for sustained use. Just to make things clear, later releases like 27.7 through 29.4 support AVX code, which is clearly unrealistic according to my perspective, and represents a severe performance hit compared to other tools. In contrast, Prime95 version 26.6 handles a genuine 100% workload without any issues.
Never exceed such high temperatures outside controlled tests—stick with what you have. For instance, use Handbrake or similar tools where you can verify full activity in Task Manager for at least five minutes before checking the temperature. Also, consider lowering the frequency to 4.8GHz for better stability. You wouldn’t ever feel a 2% difference in everyday situations.
Photonboy suggests avoiding extreme temperatures outside controlled tests. He recommends using software like Handbrake and checking Task Manager for at least five minutes to see real usage. He also advises lowering the processor to 4.8GHz for better stability, noting that a small 2% difference in real life won't be noticeable. He mentions he might need to restart everything.
I suggest running at a significantly reduced frequency such as 4.6GHz for a week to confirm everything functions properly there, then proceed if you haven’t reached that stage yet.
Additionally, perform MEMTEST86 on the DDR4 memory. Execute a FULL PASS—create the stick, boot via BIOS, and keep default settings unchanged. Allow it to complete one pass, which could take about an hour for 16GB, or longer if errors occur.
Testing the CPU is unnecessary if the memory has already been verified.
*Also, a stable overclock is unpredictable because each CPU behaves slightly differently. Some may reach the same speed at a lower voltage, resulting in reduced heat and temperature.
I never push my systems near their limits. This way it’s hard to notice any drop. For another reason, your CPU can degrade over time, so a current borderline setup might fail after a year.
When you’re close to the limit, errors may appear without you realizing them.
My guideline is to aim 200MHz below the unstable threshold, using an appropriate voltage. It’s thrilling to experiment, but eventually most people prefer stability and focus on enjoying their setup.*
Additionally, although Prime95 might not reflect real-world conditions, I find it a reliable sign that problems are unlikely, provided it succeeds. It could be considered too demanding as a test, but I still value passing it even if the temperature exceeds my ideal range for sustained use.
Additionally, although Prime95 might seem too simplistic, I find it a reliable sign that problems won’t arise, assuming it succeeds. It could be considered too light for heavy demands, but I still value a passing result even if the temperature exceeds my ideal range for sustained use.
To better understand, later iterations like 27.7 to 29.4 support AVX instructions, which is clearly not feasible as highlighted by photonboy. This represents an extreme workload—nearly 130% capacity—that far surpasses any other tool.
In contrast, Prime95 version 26.6 operates without AVX and executes a complete 100% workload:
• Prime95 v26.6 – https://www.technic3d.com/download/overc...-v26.6.htm
According to Intel’s datasheets, TDP and thermal specs are confirmed without AVX support.
Refer to page 87, Section 5.1.1, first paragraph, second sentence:
8th and 9th Generation Intel® Core™ Processor Families Datasheet, Volume 1
https://www.intel.com/content/www/u...co...vol-1.html
To confirm, Prime95 version 26.6 is perfectly suitable for thermal evaluation.
"Stress" assessments differ significantly and fall into two types:
stability tests with variable loads, and
thermal tests with constant demands. Intel evaluates processors under a steady 100% TDP load to verify thermal data.
Prime95 version 26.6 Small FFTs is optimal for CPU thermal analysis because it delivers a consistent 100% workload with stable core temperatures. No other application matches Intel’s testing conditions closely.
Tools that avoid overloading or underloading your processor will provide a reliable thermal reference. Here’s a comparison of utilities categorized by thermal and stability tests, based on TDP percentages, averaged across six processor generations at standard settings, rounded to the nearest 5%:
All measurements indicate 100% CPU utilization in Windows Task Manager, reflecting processor usage rather than % TDP load. Core temperatures react immediately to power consumption (Watts), which is influenced by workload. Prime95 v26.6 Small FFTs offers a genuine and consistent 100% workload, meaning if core temps stay below 85°C, your processor should handle the most challenging real-world tasks without overheating.