overclocking help please!!!

topprospect

Honorable
Aug 10, 2013
291
0
10,790
so i have been having this issue with overclocking my amd fx 8350 from 4ghz to 4.4-4.5 ish. ive followed every guide. from overclock.net to tomshardware you name it, i dont freeze at all follow what they say exactly they also have some motherboard etc. but when i always go to prime 95, one core ALWAYS fails. ALWAYS. doesnt matter what vcore i have. from 1.3 to 1.5. either one or cores fails. before the test i always have the vcore i want then when i run it, it goes from 1.4 right down to 1.287 every time and stays there know reason why... HELP PLEASE I NEED A GUIDE.
 


full specs please!
 
Overclocking creates heat which raises resistance in the circuits which creates more heat which raises resistance which... until you toast the processor. Dropping the voltage is the processor/BIOS's way of saving itself.

You have reached the limits of the overclocking ability of your current processor/mobo/cooler combination.

Prime95 just generates prime numbers. When used for stress testing it tells you the limits of your hardware which you have reached.

If your system runs OK withh all your other tasks, you're good but be aware you are on the fine line of failure. The longer you run it hot the more you shorten the lifespan and the more likely you are to get data errors and data corruption.

Or back it off one notch voltage and speed wise for safety.

http://www.tomshardware.com/forum/266064-29-absolutely-pass-prime95
 

topprospect

Honorable
Aug 10, 2013
291
0
10,790


sorry.

MB: asus m5a97 r2.0
GPU: r9 280x
cpu: amd fx 8350
cpu cooler: CM evo 912
case: CM HAF 912
 

topprospect

Honorable
Aug 10, 2013
291
0
10,790


the temps are fine bassicly any load besides 100% for long periods of times. will i still get optimum performance even if 1 core fails at around 10 min full load?
 

Benevolence

Honorable
Dec 2, 2013
378
1
10,860
To add for the sake of discussion: CPU's often have small manufacturing faults, and the quality of a CPU varies dramatically by batch. A classmate of mine worked testing mobileCPU's for intel. I gathered that the difference in quality on all cpu's is really all about luck from a consumer standpoint. There's enough redundancy built into the CPU for it to operate normally; however when you push it to the limits, these microscopic defects start to affect performance. It's no guarantee that your CPU will match someone else's OC stability, as there are variations even within the same batches and platters. Depending on your cooling solution, you may be able to get better stability with a better CPU cooler, but I wouldn't count on that alone giving you the ability to clock higher than where you are at now.
 

topprospect

Honorable
Aug 10, 2013
291
0
10,790


well thats good info, but what if a core fails, do i need to add more voltage or keep it the same, because voltage for my cpu is nothing doesnt matter what i put, like i could put like 2v and 1 core would still fail.
 


71c isnt acceptable for fx cpu. clearly that voltage is higher what your cpu cooler can handle. make sure you have sufficient airflow and better lower the voltage.
 

Benevolence

Honorable
Dec 2, 2013
378
1
10,860
Basically you're going to have to reel it in until it becomes stable again. As the semiconductors get hotter and have more current flowing through them, the resistance increases. Sometimes you have a segment in the die which is 100% OK under the tested conditions, but will fail once the designed perimeters are crossed. I'm studying photo lithography, and you'd be quite surprised how quickly quantum-scale defects add up in a microprocessor at <35nm. Focus first on lowering voltages to regain stability, then work on getting your cpu cooling solution up to par with the performance you are looking for.