Prime95 no AVX
Prime95 no AVX
I already understand my AIO's limits at 4.6ghz 1.26v 8700k maximus x formula. Now I just need guidance on configuring prime95 for daily stability checks. It works well in OCCT with CPU, cache, and memory tests—no problems. Checking FPU raises temperatures, and I think prime is even more problematic. I recall hearing about an older version that handles this but can't locate it.
The smallest FFTs will match the smallest FFTs. We prefer Small FFTs for ease and uniformity across different versions to avoid further confusion.
Don't lose hope or rush to judgments.
The surge in your Core temperatures was solely caused by AVX. As previously mentioned, executing Prime95 with Small FFTs without AVX will yield results that match Intel's specifications. Anything beyond that is an overload, and anything below is insufficient.
“Stress” tests differ greatly and can be split into two types: stability tests with varying workloads, and thermal tests...
Digicats, The problem with Prime95 you "heard about" mainly involves processors that support the AVX Instruction Set, like your 8700K. When any iteration of Prime95 using AVX code is executed on AVX-enabled chips, it can demand an unrealistic 130% workload. This may harm stability and put your processor under severe strain. That’s why the BIOS includes an "AVX Offset" (downclocking) feature to counteract the high core temperatures generated by these intensive AVX tasks. Real-world applications that rely on AVX code—such as rendering, transcoding, or some modern games—do not create such extreme demands. Therefore, as outlined in Intel’s datasheets, TDP and thermal specs are confirmed “without AVX.” Refer to page 87, Section 5.1.1, Thermal Considerations: 8th and 9th Generation Intel® Core™ Processor Families Datasheet, Volume 1 - https://www.intel.com/content/www/u...co...vol-1.html Prime95 Small FFT without AVX is best for evaluating thermal performance, as it matches Intel’s guidelines as a consistent 100% workload with stable core temps. No other tool can replicate Intel’s thermal testing requirements exactly. There’s a dedicated resource on the CPU forum titled "Intel Temperature Guide" https://forums. It might be worth exploring. Check Section 10 - Thermal Test Tools Refer to Section 11 - Thermal Test Basics (Pre-AVX Version) Prime95 v26.6 - https://www.technic3d.com/download/overc...-v26.6.htm Limit yourself to Small FFTs. For versions from 27.7 through 29.4, add "CpuSupportsAVX=0" to the local.txt file found in Prime95’s directory after the initial run. (Latest Version) Prime95 v29.8 - https://www.mersenne.org/download/ Click to disable AVX test options in this sequence; first AVX2, then AVX. Only Small FFTs should be performed. Core temperatures during Small FFTs remain consistent between the pre-AVX version (26.6) and all subsequent versions with AVX turned off. (CT)
P95 v26.6 lacks avx support
https://www.technic3d.com/download/overc...-v26.6.htm
Thank you for your responses. I aim to avoid generating excessive heat. My CPU never reaches those high temperatures during regular use. While gaming, I experienced around 76-78°C before lowering the clock speed to 4.6GHz. I want to be cautious and prevent damaging this PC with unnecessary voltages and heat. Would the lowest FFT setting be the best option? Previously, running small FFTs caused my CPU to reach 90°C in just half a second at 4.4GHz and 1.25V.
The smallest FFT implementations will match the characteristics of smaller FFT versions. We prefer smaller FFT tools for ease and uniformity across different releases, avoiding further ambiguity.
Don't lose hope or rush to assumptions.
The surge in your Core temperatures was solely caused by AVX. As previously mentioned, executing Prime95 Small FFTs without AVX will yield results that match Intel’s specifications. Anything beyond that is an excess, and anything below is insufficient.
“Stress” evaluations differ significantly and can be split into two types: stability tests with changing workloads, and thermal tests with constant loads. Prime95 version 29.8 Small FFTs (AVX disabled) works best for evaluating thermal efficiency, as it matches Intel’s guidelines for a consistent 100% load with stable core temperatures. No other tool can replicate Intel’s thermal workload as closely.
Software that doesn’t overload or underutilize your processor will provide an accurate thermal baseline. Below is a comparison of tools categorized by thermal and stability tests, based on TDP percentages, averaged over six processor generations at default settings, rounded to the nearest five percent:
Although these assessments cover workloads from 70% to 130% of TDP, Windows Task Manager treats each test as 100% CPU usage, which reflects processing activity rather than actual demand.
Core temperatures are influenced directly by power draw (Watts), which is determined by the workload. Prime95 v29.8 Small FFTs (AVX disabled) delivers a consistent 100% load, even when TDP exceeds normal operation through overclocking. If your Core temperatures stay under 85°C, your processor should handle the most challenging real-world tasks without overheating.
Give it a shot with AVX2 and AVX disabled. Your Core temperatures shouldn’t surpass those measured by RealBench.
If you wish to explore an 80% workload, run CPU-Z > Bench > Stress CPU.
• CPU-Z - http://www.cpuid.com/softwares/cpu-z.html
This is the typical operating range for Core temperature:
Temperatures above 85°C are discouraged.
Temperatures below 80°C are optimal.
Core temperatures fluctuate with room temperature.
TT floe riing 240mm is a garbage cooler for OC. I thought about Prime but didn't run it. Both options show high heat, which I can't handle so I can't fully test for stable OC. I'm just going to reset it to 4.4, cry, and then play.
Thanks again.