ryzen 7 1700 safe overclocking?

[fra]

hi guys...i was wondering if there are any agreed upon "safe" values i could input that would pretty much guarantee a stable OC?
lets say for example : "set voltage to "X" and set speed to "y" and you should be fine" (and im not looking for some crazy oc numbers her.nothing wild like 4,0/4.1...a nice 3.6/3.7 would be more than enough)

my build:
-ryzen 7 1700
-msi b450 tomahawk
-ddr4 pc3000 (2x8gb)
-corsair vs series 550watt
-gtx1060
(got stock fans on all components and im running one fan in the front of the case and one in the back)

 

[ShrimpBrime]

Well all the processors will request a different votlage at X overclock.

I'd start at 3.6ghz at 1.3-1.35v and stability test it for a bit. If it passes at 1.3v, lower the v-core slightly and test again until unstable. Then bump it back up a little to ensure stability.

But really nobody here really knows what your processor can do, you'll have to do all the testing to find out.

 

[fra]

Well all the processors will request a different votlage at X overclock.

I'd start at 3.6ghz at 1.3-1.35v and stability test it for a bit. If it passes at 1.3v, lower the v-core slightly and test again until unstable. Then bump it back up a little to ensure stability.

But really nobody here really knows what your processor can do, you'll have to do all the testing to find out.

what increment would you suggest?say i do 3.6/1.3

what would the next bump be?

 

[EarthDog]

You can try starting out by reading this guide on the front page.
https://www.overclockers.com/amd-ryzen-overclocking-guide/

I haven't really played with these enough otherwise.

Cookie cutter settings may nt work is the problem. Too little voltage and it isn't table, too much and it is too much heat, etc.

 

[fra]

ok guys so i set the cpu to 3.6 and voltage to 1.3....the two system fans running at full speed (1000rpm each) and the cpu fan goes to 100% after reaching 85C(185F)

i ran stress test in prime 95 for about 20 minutes....for the first 20 minutes it was rising very slowly up to around 79C(174F)c...and then it spiked very quickly to 95C+(205F)

any thoughts?

 

[EarthDog]

Thoughts are, according to that guide....... temps are too high.

 

[fra]

i mean,thoughts on what to do

 

[EarthDog]

The name of game here fra, with any overclocking, is to achieve a clock speed with reasonable voltage and appropriate temperatures. The guide suggests how that is done for your CPU.

In a nutshell, if temps are too hot, you need to do one of three things...

1. Lower the voltage (while making sure the system is stable)
2. Lower the overclock (and subsequently the needed voltage)
3. Get a better cooler for the CPU.

What heatsink are you using for the CPU?

 

[trents]

At that CPU core voltage you need better cooling, at least for the CPU and maybe faster fans for the case to improve air flow.

Have you tried lowering the CPU core voltage more? You may not need 1.3 volts to be stable at 3.6. My wife is running a Ryzen 1600 and it takes 1.28 vcore to be stable at 3.8 ghz. I'm guessing you need to lower the vcore.

Also, I would steer you away from using Prime95 as the stress tester. It may be unnecessarily harsh for modern CPUs. Try the Realbench stress test. If you can pass it for two hours you are likely to be stable for most real world use scenarios. It will not drive temps up so high as Prime95.

If your cooling would permit it, you would be safe for 24/7 use up to at least a 1.35 core voltage so you can see how your cooling is limiting your overclock at the present.

I would also experiment with your LLC setting to see if "voltage droop" under load is causing you to have to use excessively high core voltage to compensate, which would drive up temps.

Make changes to CPU core voltage in .01 increments.

Stay where you are right now with core voltage and run the Realbench stress test for 15 minutes using 8 gb RAM. If you pass, increase your CPU frequency multiplier by 1x and rerun the stress test for 15 minutes. If you pass that, raise the core frequency multiplier by another 1x. Repeat that until you fail the stress test and then back dow 1x. Then run the stress test for two hours. Make downward adjustments as necessary to core frequency multiplier or upward adjustments to to the core voltage in order to pass the two hour stress test if you fail to complete it. The Realbench stress test has a built in CPU temp monitor so that is helpful. Keep temps no higher than 90c.