Is there a single-core benchmark or stress test that involves more than one core?
Is there a single-core benchmark or stress test that involves more than one core?
In prime 95 I'm attempting to replicate my overclock by increasing vcore and MHz through stress on one core. Rather than having one core at 100% and others varying, my cores stay around 50-70%. The same issue appeared when using the single core benchmark in cinebench. What adjustments should I make?
In prime 95 I'm attempting to replicate my overclock that increases vcore and MHz by putting stress on a single core. Rather than having one core at 100% and the others at random, my cores are around 50-70%, which isn't close to full speed. I also tried this approach with the single core benchmark in cinebench, but the issue persists. How can I resolve this???
In prime 95 you are attempting to simulate an overclock by increasing vcore and MHz through stress on a single core. Instead of having one core at 100% and others at varying levels, your cores are hovering between 50-70%, not reaching maximum capacity. The same issue appeared when using the single core benchmark in cinebench. How can this be resolved???
What you are observing is the operating system distributing the single thread workload across all available threads. This pattern can confuse users into believing their CPU is handling multiple cores when it is actually operating with a single core at suboptimal performance.
You might consider locking a core to the task or, in Windows, adjusting the advanced options in msconfig to specify one core. Rebooting may be necessary to apply these changes. For more stable testing, consider using OCCT instead of prime95 or IBT.
Also, I'm experiencing issues with turbo boost. I understand my first core can handle 48 GHz at 1.35 v (i5 4690k), but when all cores are active I reach 4.5 GHz. I adjusted the BIOS settings accordingly. The limits I set are: 1-core max 48, 2-core max 46, 3-core max 45, 4-core max 45 (voltage adjusted). I know all cores can reach 45 even at 1.32 v, but I encounter BSODs when the system boots with low voltage. It seems the design uses a multiplier for the weaker cores (3 and 4), which might cause the crash. Will following your previous advice help, or is this a natural characteristic of turbo boost?
Initially, your safety limit depends largely on the quality of your chip.
I believe your chip performance is at least average.
I recommend resetting all configurations back to their default state, including RAM settings.
Then slowly raise the multipliers on concert and observe your results.
Keep an eye on your vcore; values above 1.30v are not ideal, as higher voltage tends to increase temperatures.
Perform a stress test using OCCT—it employs more typical instructions compared to prime95 or IBT.
It will terminate the test once temperatures reach 85°C.
Watch your vcore closely.
If you hit your maximum, consider reducing the voltage slightly.
How high should you aim to go?
I keep RAM at its standard voltage. Higher speeds require more voltage, which can slightly affect OC performance.
Discrete graphics cards often work well without such adjustments.
See this article on Haswell RAM scaling: http://www.anandtech.com/show/7364/memor...on-haswell
After completing these steps, apply speedstep and adaptive voltage techniques.
This will allow your multiplier and vcore to decrease when the CPU is under light load.
I possess the previous release of prime 95 (excluding AVX instructions), my RAM is standard at 1866 MHz, and the temperature stays below 70.
I can handle 4.5 ghz at 1.29v, so I think that's a solid choice.