[SOLVED] Ryzen 5 5600x too high voltage

[punisher_09]

Hello ! I have a new Ryzen 5 5600x and I made a few tests using Cinebench and OCCT. The max voltage during test with Cinebench was 1.337V and the temperature was 61.3C, and during test with OCCT max voltage was 1.375V and temperature was 65.5C.
I have PBO enabled in BIOS and I am using a Silentium PC Fortis 3 cpu cooler.
Also, sometimes, the voltage peaks at 1.400V, probably when only 1 single core is in use.
In Cinebench R20.060 I got 4318 pts (multicore).
I read on different forums that the max voltage for this Ryzen 5600X is 1.375V, while on other forums, I read that the max voltage is 1.500V. Can you please clarify this ?

 

[drea.drechsler]

... Can you please clarify this ?

When Ryzen CPU's boost to max clock the boost algorithm will raise core voltage as high as 1.5V to keep stable but only with light bursty loads. When the processing load lasts longer, uses more threads and/or involve more intense instructions (like AVX) the algorithm will pull back on voltage, and maximum boost clocks, to keep the processor safe.

That's "as high as" but not necessarily 1.5V. If it does it's safe because the CPU's not really very hot yet. It's core current and temperature that degrades CPU's, not voltage, but voltage is the only knob we have to turn that can reduce both core current and temp. That's exactly what the boost algorithm does (in addition to reducing clock) when lowering it to 1.3375V during Cinebench.

Can't say why someone recommends that "1.375V". IMO, it's way too high for a fixed overclock and I'd never run a fixed voltage for PBO since it's highly desireable to let the algorithm lower it to cool off the cores whenever it can. Also, even though there is intermittent OS activity when idle the CPU spends the vast amount of time with various cores in a C6, deep sleep state, with voltage less than 1.00V. So locking it at any voltage is more harmful than letting the algorithm do what it was designed to do.

 

[DimkaTsv]

Also, sometimes, the voltage peaks at 1.400V, probably when only 1 single core is in use.
In Cinebench R20.060 I got 4318 pts (multicore).
I read on different forums that the max voltage for this Ryzen 5600X is 1.375V, while on other forums, I read that the max voltage is 1.500V. Can you please clarify this ?

Stock voltage settings of Ryzen are pretty high for ensuring full stability in PBO, as it is automated curve. It is pretty easily fixable with PBO Curve Optimizer, but depth of undervolting you must know yourself
More than that, Ryzen 5XXX series undervolting also allowes PBO to increase frequencies more than stock (but by default no more than 4650 mHz), you can also override that in PBO settings too, and get not only lower voltages consumption and, most likely, heating, but get even bit more performance... Also, you can look up into Core VID X in then CPU section of HWinfo, and not CPU Core Voltage

Like my current PBO UV CO is (all negative) 30-30-27-20-20-30 (Core performance order 5-1-2-4-3-6) and boost overdrive is +50, my average voltage on CPU Core voltage is 1.1 - 1.17 (and CPU Core VID X usually not higher than 1.13) , with peaks up to 1.35 for VERY short time (very rarely and most likely VDROOP stuff). I was more focused on undervolting my CPU though than performance gain, even though i got my performance gain too (CPU-Z is 650.6/5135.3 ... well not quite fair result, as it is done on fresh and cool system, and CB20 is 4598). Why boost overdrive +50, and not +200? ... I lose stability too easily get more voltages and don't want to look for C1234 CO values again. . .

Only problem you may get is stabilty after UV, so highly recommend you to test it with prime95 on 4 threads! On 4 and 2, not on 6 or 12... Also if 2 or more threads fails at once can be with a minute interval between them) it not specifically means that these threads voltage is too low, but it can mean that previous thread got too low voltage and lost stability