Did i win the silicon lottery with my Ryzen 1600 & ROG STRIX B350-F GAMING ?

[Equinox187]

As the title says did I win the silicon lottery with my Ryzen 1600 or is it normal or am I just not pushing it far enough as i have seen others with a similar OC but a with much higher core voltage.

The system:



CPU: Ryzen 1600 OCed
Motherboard: ROG STRIX B350-F GAMING
Ram: 16GB (2x8GB) DDR4 3000Mhz G-Skill Trident Z 15-16-16-35 - F4-3000C15D-16GTZR
Graphics Card: GTX 970 soon to be a GTX 1070
Power EVGA 650w GQ Gold
Cooling: Generic Chinese pump and res / Magicool G2 240mm Rad / EK Water Blocks EK-Supremacy MX - Acetal / Mayhems pastel blue coolant /Pump running at 50% and rad fans running at 50%.

So the reason I ask is i currently have the Ryzen 1600 over clocked to 3.8GHz and nothing special with that until it comes to the voltage, I have seen others with the same OC but they seem to be running closer to the 1.35v some even closer to the 1.4 but in my case with a +0.03750 offset on the voltage when running prime 95 and 100% usage across all cores it sits stable at 1.286v after 3 hours testing and never goes over 50c with all cores at 100% (see bellow image ), and under normal use it idles at 25-27c and goes to about 36c under gaming. So did i get lucky with the silicone lottery and should i push the chip further or are other people just seriously overvolting their chips?

[st0neh]

3.8 really isn't all that much regardless of voltage. It also depends on how you're running Prime95, some settings are a lot more intensive than others. And 3 hours really isn't much time.

[Equinox187]

st0neh wrote:
3.8 really isn't all that much regardless of voltage. It also depends on how you're running Prime95, some settings are a lot more intensive than others. And 3 hours really isn't much time.

As I already said the 3.8 was nothing special but also its hardly not all that much given that its 400Mhz over the boost clock and 100Mhz in most case away from the chips max or 200 if you are really lucky and the difference between 3.8 and 4.0 is negligible, but that's getting away from the point of my post that i have looked all over and many are having to use 1.37v of voltage or more at 3800 and since this post i pushed it to 3.9 with another small boost to offset and voltage maxed at 1.306v with all cores loaded under prime 95 for 6 hours.

As for prime 95 Under torture test I have tested using all of them SmallFFTs / In-place large FFTs and Blend, as for time if you understood the point of prime 95 you would know that 3 hours is more than enough when only testing core stability, yes if you are testing memory stability then you want to run for 12 or more hours but as far as core stability testing goes you really do not have to test for that long at all as testing cores is not the same as testing memory as any instability will usually present its self the second you load all the cores to 100% or soon after if there is any voltage drops during the 100% test and if you get past the first hour you can pretty much guarantee that the cores are stable as they would not have got that far if they where not and any issues after would be memory related.

[st0neh]

Equinox187 wrote:
any instability will usually present its self the second you load all the cores to 100% or soon after

That's really not the case though. I've had CPU overclocks run stable Prime95 and fail after 22 hours.

[JustinThyme]

Prime 95=CPU to heater conversion tool.

Run something real world in a loop like realbench to determine stability. Once it heats up to max temps and runs 15 mins after that if it didn't fail its stable.
Ive had machines run for months then after a windows update not run stable any longer, just back off the clock a little or increase the Vcore a little. Sometimes they just lock up after a few days or a few months or years. Just remember you are running a device past its manufactured design and stability past that point is not a guarantee, neither is longevity of the chip.

[st0neh]

Realbench is a horrible stability test. It's just nowhere near intensive enough.

[celab]

try using cpu-z internal cpu stress test and lynx ver.0.65. than go into a gane,. prey and bf4 are working good for me. they usually crash within minutes ^^