first time with i5-3570k asking about overclocking
first time with i5-3570k asking about overclocking
I was running stress tests on my OC of the i5-3570k, increasing the frequency from 3.4GHz to 4.2GHz, which should be fine based on what I've observed and seen in Aida64. However, during the test, I noticed unusual readings on the graphs. The second image shows the readout still active after the test, with values significantly off. It's worth paying attention to both the CPU clock minimum and the graph fluctuations. What could be causing this?
As my first OC, I hadn't even begun experimenting much yet. I just selected an OC presave from the BIOS for 4.2GHz and tried to boot it to check if it worked.
CPU - i5-3570k
Mobo - Asrock z77 extreme 4
PSU - 750w osc
Ram - 2 x 4gb g.skillz @ 1600MHz (speed set to 1600MHz)
Edit 1: I briefly restarted Aida64 to check the CPU voltage under load using cpu-z, which showed fluctuations between 1.29v and 1.32v.
The 1600 current has reached its limit as there hasn't been any activity since the test was mentioned (this is functioning properly). You should adjust the multiplier to 44 and set the voltage to 1.24. After launching OCCT and RealTempGT, expect a consistent 4.4 overclock. If instability occurs, increase the voltage by 0.02 when stable, then reduce it slightly if you achieve a steady result while maintaining good temperatures. I usually raise the voltage by 0.02 to ensure stability. Your current voltage is quite high for 4.2.
It's not a big deal. I notice unusual spikes in the graph when using OCCT. Your voltage seems to fluctuate, indicating the "Auto" setting for CPU voltage or Vcore might be incorrect. Try manually setting it at 1.20 and check if stability is maintained at 4.2GHz across all four cores (42x multiplier, 100 BCLK). If not, gradually increase the Vcore by .05v increments until stable (around 1.205v, 1.210v, etc.). If stable, adjust slightly lower values with .001 changes to minimize voltage for stability. The motherboard's built-in Vcore controller usually provides more power than needed. Also, remember to perform a long stress test—hours rather than just 30 minutes—to properly assess stability. I've experienced crashes after nearly two hours of extended testing.
All this depends on the CPU cooler you're using, which you didn't mention.
The 1600 current has reached its limit as there hasn't been any activity since the test was mentioned (this is functioning properly). You should adjust the multiplier to 44 and set the voltage to 1.24. After launching OCCT and RealTempGT, expect a consistent 4.4 overclock. If instability occurs, increase the voltage by 0.02 when stable, then reduce it slightly if you achieve a steady result while maintaining good temperatures. I usually raise the voltage by 0.02 to ensure stability. Your current voltage is quite high for 4.2.
The 1600 current has reached its limit because there hasn't been much activity since the test. (this is functioning properly)
Adjust the multiplier to 44 and set the voltage to 1.24. Run OCCT and RealTempGT; they should stabilize at around 4.4. If instability occurs, increase the voltage by 0.02 when stable, then reduce it slightly if you achieve a steady overclock while maintaining good temperatures. I usually raise the voltage by 0.02 to ensure stability. The current voltage is too high for 4.2.
I plan to do this today. I allocated about 8 hours to test a stable CPU and possibly a GPU overclock. Thanks for the guidance; I'll review it multiple times as needed.
10tacle :
That's nothing to worry about. I get weird graph spikes with OCCT. However, your voltage is jumping around so that means you must have "Auto" selected as the CPU voltage or Vcore setting. What you should try is set that at 1.20 manually and see if it will maintain stability at 4.2GHz on all four cores (42x multiplier, 100 BCLK). If it does not, crank up the Vcore by .05v at a time and keep testing until stability is reached (1.205v, 1.210v, etc.). If it does, go backwards in the same manner.
You can then fine tune the voltage going in between +/- the .05v with .001 changes when you do get stability (1.214, 1.213, etc.) to get the least voltage necessary for stability. The motherboard's onboard Vcore controller always pumps more juice than is necessary. Also FYI you really need to run a stress test for hours, not 30 minutes, to truly get an idea of stability. I've run into crashes after nearly 2 hours of a stress test.
All of this, of course, is dependent on what kind of CPU cooler you have, which you did not mention.
Thanks for the info. I'm going to be fine tuning all today to try to get a stable OC manually and then I'll be stress testing for 8 hours over night. Also my cooler is a evo 212 so I cant push this thing to the max because I don't got water cooling for the temp.
We appreciate the details. I plan to fine-tune everything today to achieve a stable overclock manually, followed by an 8-hour stress test overnight. My cooler is an Evo 212, which limits me from pushing the system too hard since I don’t have water cooling for temperatures. With a 212, you should be able to reach around 4.3GHz-4.4GHz while staying under 80°C during load. Are you using HW Monitor to monitor temperatures and voltages? Concerning the CPU voltage (Vcore), do you set it automatically or are you manually adjusting it like in the example chart? Usually, the motherboard’s BIOS applies more voltage than needed for a specific overclock (the auto setting). Overclocking really involves a lot of trial and error, but I haven’t experienced any corruption before. Speaking of, if you update any software—like the latest Windows update or install new programs—make sure to revert everything to the original speed. Hopefully your BIOS supports custom boot settings, allowing you to keep one configuration at stock speed and another at your overclock level, avoiding the need to manually adjust each time.