Question How high can I raise the Vcore on a Ryzen 7 3800x?

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Feb 24, 2020
14
0
10
I have the Vcore right now at 1.45V with the cpu clock ratio at 44.5 from 39. It's stable right now. It's not stable at 45. Should I keep raising the Vcore? Everything else is in Auto including Vcore SOC and Global C-Control State. Motheboard x570 aorus pro wifi. Thank you.
 
Feb 24, 2020
14
0
10
It's both current AND temperature. The way to reduce temperature at a given clock, though, is to lower voltage.

As far as voltage, the only thing I've ever seen from AMD about a safe voltage is to keep it in auto, let the processor determine what is safe for itself. It automatically reduces voltage, AND clocks, to a safe level based on temperature. If you've ever heavily stressed a Zen2 processor at a heavy load (like Prime95, small FFT, AVX) when in full auto, no PBO, you might find it lowering voltage down as far as 1.2 volts, with a clock speed right at or just above it's spec'd base frequency. That's about the best way to figure what's safest.

Ok, because we can't measure the current that goes through the processor and deal only with the voltage, which is basically the same thing given they are direct proportional, can I raise the voltage over 1.5v as long as I keep the temperature down under stress with an adequate cooling system, let's say below 75C?
 
Ok, because we can't measure the current that goes through the processor and deal only with the voltage, which is basically the same thing given they are direct proportional, can I raise the voltage over 1.5v as long as I keep the temperature down under stress with an adequate cooling system, let's say below 75C?
see what I wrote above....about LN2.. Yes you theoretically could, with adequate cooling AND a VRM capable of stable, even power delivery on your motherboard.

But I wouldn't go above 1.5V. That's because that's what AMD limits their specified voltage range for processor operation to. I also don't think it's possible to get the heat out fast enough...just not enough surface area on the AM4 heat spreader to move it. Not considering the temp differential you could get from a water cooled, or ambient air cooled, system. So it's all quite theoretical, unless you're talking sub-ambient or LN2 cooling.

And one other thing: it's patently incorrect to suggest voltage and current are directly proportional. The V-I curve of a semiconductor junction is really so UNproportional that engineers usually try very hard to stay below the 'knee' of the curve. That's the point where current starts going up way, way faster as voltage increases by sometimes very small amounts.

Overclockers generally don't care...we bump up the volts and buy expensive cooling solutions to overcome the heat the designers try to not generate at all by staying in the most efficient part of the curve. But even an overclocker needs to drop a few mV to keep heat in check sometimes, and the ensuing loss of stability means dropping clocks too...which, by the way, Ryzen's boost algorithm does automatically!
 
Last edited:
Feb 24, 2020
14
0
10
It's both current AND temperature. The way to reduce temperature at a given clock without changing cooling is to lower voltage since (voltage * current) = power / heat.

As far as voltage, the only thing I've ever seen from AMD about a safe voltage is to keep it in auto, let the processor determine what is safe for itself. Left to itself, it automatically reduces voltage, AND clocks, to safe levels based on temperature. If you've ever heavily stressed a Zen2 processor at a heavy load (like Prime95, small FFT, AVX) in full auto, no PBO, you might find it lowering voltage down as far as 1.2 volts, with a clock speed right at or just above it's spec'd base frequency. That's about the best way to figure what AMD considers safest. Some call it the FIT voltage and different for each processor, programmed at manufacture based on silicon performance in binning tests.

Go watch some overclocking on LN2 (there's quite a few techtube videos). They'll raise voltage as high as the VRM controller allows...even modifying it to allow voltages way above the safety limits. That can be 1.55 - 1.6 maybe more even. The processors can draw 200+ amps current, at 5+Ghz. How can that be safe? even for just few minutes to demonstrate operation and benchmarks? You can be sure they don't want to degrade their processors...many times they're 'gold samples' representing the best silicon they could find after many hours of testing and sorting. They need them operating as expected for their competitions.

The answer is, of course: they're maintained at a temperature of -100C or more on LN2. So electro-migration is being held to an absolute minimum at the extremely high current densities they're operating under.

Ok, so basically doesn't really matter how much one raises the voltage as long as the temperatures are under control degradation does not happens.
 

zx128k

Reputable
It was posted by zx128k. It's a valid one I guess. Basically zx128 says that electromigration happens when the amperage increases due to high voltage applied to the cpu and doesn't really matter how cool one keeps the processor the degradation still happens because of the high current induced which causes to displace ions and knock down atoms. You say that as long as one keeps the dissipated heat caused by the high current within acceptable limits degradation doesn't take place. Of course dissipated heat is caused by high amperage which is direct proportional with the voltage. If you raise one the other one automatically raises. I was under the same impression that as long as the processor is properly cooled doesn't matter the voltage. I'm in doubt right now. Which one is right?

Be careful there temperature does affect things. We have not fully determined what is safe. In ohm's law voltage and current are like so,

ElfcSznvuS50x-itIUmxZU4liyw2IODsK45GEE8gp_dv3dTgJXa6pO3vGFKyp615i8b1L3T8k6Dy6CF0UTf6ZIJzf84wVMPQ2yLe2PlxUuJ8N9Tno52d_PzzHEpRKxvG4l2Pd8xczUmColDawNowaO5j7YX9I45WvUD1drM6h6J6QT9DFWmDV-XA5kfe4lM


Not all devices follow Ohm's law and statements on the internet about cpu's are often full of non sense. You have to see the information yourself and make that decision yourself what you want to follow. You can't determine what temperature it is safe to apply what vcore voltage. We don't have that information. I don't know what way the cpu boosts. All I have is the information available online like everyone else. Some of it is going to be bullcrap. Just remember we have just speculation here. Nothing more or less.

Even the statement from AMD is given in its full so you can make up your own mind.

Going colder helps the CPU in a number of ways which allows higher vcore and higher core frequencies.

Speculation is not the same as facts or the need to think for yourself.
 
Ok, so basically doesn't really matter how much one raises the voltage as long as the temperatures are under control degradation does not happens.
ummm... that's not really right because degradation ALWAYS HAPPENS as it's an inevitable effect of current flow (moving electrons that knock loose ions). Current is flowing all the time, while the processor is in operation. It just occurs at a faster rate when it's hotter.

What I'm really saying is, in a fixed voltage overclock with TEMP's fully controlled, no matter how low you make the voltage at heavily loaded condition it's probably not safe. Certainly not 'safe' as what AMD makes it when their algorithm is occasionally hitting voltages down to 1.2V under heavy, hot loads and lowers voltage to 200mV in frequent deep sleep states (around 90% of the time in my case) during normal workloads, such as general computing tasks. Even gaming my processor spends over 50% of it's time in deep sleep. At one time it was a rule of thumb that 1.325V was considered 'always safe'. But now a lot of people who fixed an overclock at that voltage and went about heavy tasks without a care are already finding their processors degrading; not staying stable in a stress test they used to pass. It turns out that 1.325V 'safe voltage' was a misinterpretation of a post an overclocker made early on in the release of Ryzen 3000.

Also, since temperature does follow voltage, and temperature has to be controlled in extreme loads, not just at idle, it pretty much makes a practical limit to how high you can leave voltage to remain stable in loaded conditions anyway.

And lastly, you have to remember there is a voltage stress limit: the stress limit is the point where (simply) voltage chips away at dielectric insulators in the CPU which also lowers the voltage needed to take out the next chip. Every time it hits the (ever decreasing) stress voltage, it chips a little more. Eventually it chips enough to short out junctions and stop working. In contrast (and also simply), degradation due to electron migration makes conductor resistance increase inside the processor, requiring ever increasing voltage (with increased heat production) to remain stable at a given clock speed.

AMD's spec'd voltage operating range is 200mV - 1.5V which doubtless provides some margin to the stress limit. So AMD's saying the maximum operational voltage is 1.5V, which their algorithm lets the processor reach when boosting in light/bursty loads.
 
Last edited:

zx128k

Reputable
There are several factors contributing to the CPU power consumption; they include dynamic power consumption, short-circuit power consumption, and power loss due to transistor leakage currents:

eb3d0b2f8836bd8cdf7c14d14a5b8d0126514e84


The dynamic power consumption originates from the activity of logic gates inside a CPU. When the logic gates toggle, energy is flowing as the capacitors inside them are charged and discharged. The dynamic power consumed by a CPU is approximately proportional to the CPU frequency, and to the square of the CPU voltage:

79198534a73f4231242a253ca69beef559d608ad


where C is capacitance, f is frequency, and V is voltage.

When logic gates toggle, some transistors inside may change states. As this takes a finite amount of time, it may happen that for a very brief amount of time some transistors are conducting simultaneously. A direct path between the source and ground then results in some short-circuit power loss. The magnitude of this power is dependent on the logic gate, and is rather complex to model on a macro level.

Power consumption due to leakage power emanates at a micro-level in transistors. Small amounts of currents are always flowing between the differently doped parts of the transistor. The magnitude of these currents depend on the state of the transistor, its dimensions, physical properties and sometimes temperature. The total amount of leakage currents tends to inflate for increasing temperature and decreasing transistor sizes.

Increased heat in a wire generally means higher resistance. Also increased heat in a semiconductor could lead to higher conductance which can cause transistors to not work properly. Increased temperatures inflate the leakage currents in transistors.

Increased leakage is a common failure mode for a cpu. It results from non-catastrophic over-stress of a semiconductor device, when the junction or the gate oxide suffers permanent damage not sufficient to cause a catastrophic failure.

So if you lower temps, you can increase voltage while keeping the same peak core temperature.

There is always a level of voltage were damage will happen even on LN2. Voltage is like a dam. Pressure increases until a hole forms, then the current is the water flowing out the hole.

With a high enough voltage, you can overcome the resistance of air and form an arc across a gap. Overcome the insulation between the plates in a capacitor. Break the insulation between traces in a cpu. Cause the breakdown of transistors.

Autoco plete can go to he'll.

 
Last edited:
I recommend everyone read this article:
...

Yes, a very good read.

The best take-away I got from when I first read it is this quote:

"...In a way this means there is no such thing as a “safe” voltage. Simply running a chip damages it, even at stock, regardless of voltage. So what we mean when we say “safe voltage” is effectively something like;

'The voltage at which it’s expected that the chip will not be damaged or destroyed so fast that we regret overclocking it'... "
 
Last edited:

zx128k

Reputable
Yes, a very good read.

The best take-away I got from when I first read it is this quote:

"...In a way this means there is no such thing as a “safe” voltage. Simply running a chip damages it, even at stock, regardless of voltage. So what we mean when we say “safe voltage” is effectively something like;

'The voltage at which it’s expected that the chip will not be damaged or destroyed so fast that we regret overclocking it'... "

Indeed and AMD states stock voltages are the safest. The quote, "'The voltage at which it’s expected that the chip will not be damaged or destroyed so fast that we regret overclocking it'... " end quote, seems to be the question. Also there are issues with some of the reports of degradation. With the 3600x he only did 3 hours or prime 95. So that begs the question if the cpu was stable in the first place.
 

schmuckley

Distinguished
BANNED
Jul 18, 2011
809
5
19,165
I have the Vcore right now at 1.45V with the cpu clock ratio at 44.5 from 39. It's stable right now. It's not stable at 45. Should I keep raising the Vcore? Everything else is in Auto including Vcore SOC and Global C-Control State. Motheboard x570 aorus pro wifi. Thank you.

No. Back it off and lower your OC.
Did I seriously just suggest someone back off an OC?
I did, and that's what's needed now.

Unless you're doing something extreme, keep it below 1.45v and see what clocks you get stable.

People are having conniptions and saying 1.37 is too much. I say no, but that's just an opinion.

Trust me, I can show you how to give a processor 1.9v.
Don't do that unless prepared to.
 
Last edited:
Feb 16, 2020
45
11
45
. I just bought a 360 AIO from Enermax. Only 5c reduce compared to a Synthe Ninja 5. Honestly, I was expecting way better performance from a 360 water cooling. Not there.

Please check out gamers nexus recent posts about the efficacy of that aio. They are gumming up something fierce and enermax has come under a lot of fire this past week for it.
 

zx128k

Reputable
The overclocking of the cores is only a small amount of performance. With an overclock to 4.4GHz all cores most of the performance is from setting to IF to 1900 and ram to 3800 cl14 with tightened timings. You take a risk overclocking the cores for a few hundred points in time sky cpu. The ram overclock/IF overclock can increase performance from 10k up to 11.5k time spy cpu.
 
Please check out gamers nexus recent posts about the efficacy of that aio. They are gumming up something fierce and enermax has come under a lot of fire this past week for it.
Oh yeah... i've read/watched both articles from GN about the Enermax coolers. They really are good performing AIO's but support some kind of chemistry or biology that gums up the micro-fins in the water blocks. It's really quite sad as the biocides you could put in the liquid mix to prevent this sort of thing isn't really a big trade secret or anything.