Question Voltage on Ryzen 7 3700x

[AdrianVaughan]

Hi, all.

I just bought a new 3700x and put it in my b350 Tomahawk board. The only thing I am worried about is the voltage that is going on.

Is this safe being round a CPU core ? Because on my 1st gen, it was never this high.

Thanks

 

[CountMike]

Way too high, shouldn't go over 1.375 under full load. You may want to limit it to that much or less in BIOS. Also use HW Info https://www.hwinfo.com/download/ for more accurate readings.

 

[CosmicDance]

This is a long post so just the following first paragraph sums it up if you do't want boring with the finer details!

These voltages are perfectly safe for your CPU but I understand why you are concerned as they are higher than they need to be for normal smooth running.
This is due to both the Ryzen 2000 and 3000 series of CPU s having their voltages set notoriously high by the BIOS.

There is more to the CPU's voltage but to put it more plainly....
The CPU's Precision Boost Overdrive (PBO) system which raises the speed during load balances out voltage, demand for CPU power, heat and thermal throttling amongst other factors so that it never exceeds its thermal safety temperature but always delivers the best performance.
This performance curve is where overclockers manually intervene in order to achieve slightly better results.

These CPU s can run very happily at around 1.37 volts even in peak demand like benchmarking or heavy gaming.
The fact that they run at around 1.48 or more is down to the design and hence what you see on your screen via the hardware monitoring software.
The program HWinfo 64 gives a much more accurate and detailed display showing voltage on all cores, CPU multipliers which lower when the temperature threshold is reached so I do recommend this one if you want a deeper understand of what your hardware is doing in realtime.

So this is why people who overclock to a great depth will tweak every possible factor and often under volt their CPU manually in order to control temperature & voltage more efficiently and consequently squeeze a little more performance out of their chips.
They often reduce the CPU's core voltage by 0.100v which means it will be running at around 1.37 volts or even less but test it so that it's stable.

This is under load though and in Windows just browsing the internet etc the voltage is controlled and lowered automatically.
Usually anyway....
More so when idling but there have been complications with some PC set ups so using Balanced Power Plan is generally the best all round for idling to heavy gaming/intensive applications.
You can also switch to Windows Power Saving plan which means the CPU will be heavily throttled even under load and consequently idles at a very low speed with lower voltage.

AMD have published updates recently stating that some hardware monitoring software can probe the CPU too aggressively and stop ot idling properly which reflects higher voltages.

Under load the CPU is throttled when it gets to 85c on your hardware monitoring software which is actually 95c as their is a 10c difference between displayed and actual temperatures.
Now this is obviously extremely high and approaching the CPU's thermal cut off where it literally shuts down in milliseconds to avoid damage.

So the hardware is built to both protect and optimize itself along with the software to balance it out with peak performance.

Also as time goes on some motherboard manufacturers will release more optimised BIOS updates to balance out voltages and performance etc.

The newer motherboards like the X570 range are more efficient with with updates that maximise the CPU's faster transfer speeds for RAM and the PCI bus.
They won't necessarily lower the CPU voltage though as this is set by motherboard manufacturers in accordance with their collaboration and working with AMD to maximise performance.
It is AMD who optimise their hardware and circulate guidelines for the masses of possible PC configurations.


Andy

 

[MeanMachine41]

Hi Adrian

You will need to stress test to determine best voltage settings under load, according to any OC you decide and your cooling system. Keeping voltage as low as possible whilst maintaining stability is best for longevity and best kept below 1.4V

VID (Voltage Identification Digital) voltage is hard coded into the CPU to essentially provide a stable voltage to maintain stability according to the task or demand.

1.475V on the core is a little high for basic use (Your pic shows the system at idle) however not dangerous depending on your temperatures under load.

I have found CPU_CORE voltage is always a little high when set to AUTO. You can adjust to a fixed CPU Core voltage in Bios and fine tune with offsets to reduce heat and maintain stability when under load.

 

[AdrianVaughan]

I don't over clock, this is default, temp never go above 50oC when in use, I have found out that icue keeps it from going idle and it drops the voltage to 0.9 when that program is closed so should I set it in bois to 1.35v? Or maybe even 1.3?

 

[MeanMachine41]

Yes, Test it at both and see if it's still stable. You have nothing to worry about.

 

[AdrianVaughan]

ok i set the cpu vcore to 1.325

and this is what i have now...is this ok ? the Core Vid is still high tho @ 1.444? or is that normal ?

 

[CosmicDance]

You are fine with those voltages and temperatures.
Run either a benchmark like Cinebench/UserBenchMark or an intensive game to check it's ok and stable without crashes.

Have HWinfo running prior to it and it will record your maximum values like temperature etc.
You can also observe these in realtime during benchmarking.
Cinebench can push CPU s to their thermal limit so you might get a very high temperature even up to 85c during multi core testing.
Thermal throttling will kick in on your CPU to keep this rising any higher.

 

[AdrianVaughan]

will do thanks very much

 

[TechyInAZ]

Ok I see the problem here,

OP you are using the VID to measure vcore. That is NOT the right sensor! You need to be paying attention to the SVI2 Core Voltage Sensor!!!
SVI2 is the actual voltage that the CPU is getting, VID is what's being requested, not actually being used.

1.306v on SVI2 is completely safe, I'd recommend going back to automatic voltage for stock operation as SVI2 is way lower than VID.

 

[AdrianVaughan]

ahhh ok i will do that

 

[AdrianVaughan]

so this is with it set to auto now , still ok ?

 

[rigg42]

Don't set a manual vcore if you are leaving the clock multiplier on Auto. You will kill your performance. Your single core performance especially. Run a Cinebench test before and after setting voltage manually and you will see what I'm talking about. The Auto voltage is fine and perfectly safe. Your CPU will only operate at that high voltage when lightly loaded and hitting max turbo like it is in your screen shot. Since there isn't a lot of current running through the CPU it will be perfectly fine. If you load all cores heavily the voltage will go down into the low 1.3 range. Your monitoring software is hitting the core with enough load to cause it to go into full turbo and the voltage it needs to hit that. When the CPU is truly idle it will hover at around .9 volt. Fully loaded it will be around 1.3. When lightly loaded and turboing up such as your screenshot it will hit 1.45-1.5. The CPU is doing exactly what it is supposed to be doing and this voltage behavior is perfectly normal. People sounding alarm bells about voltages and telling you to undervolt are misinformed and don't understand how these CPU's behave. They are literally telling you to make your CPU slower than a 1st gen Ryzen. Underolting the new 3000 cpu's doesn't achieve the same thing as it did on 2000 CPU's. People who haven't played around with these CPU's need to stop spreading bad information.

Read this:


View: https://www.reddit.com/r/Amd/comments/cbls9g/the_final_word_on_idle_voltages_for_3rd_gen_ryzen/?ref=share&ref_source=embed&utm_content=title&utm_medium=post_embed&utm_name=ecba227119304d4e9b83fca7b97f403b&utm_source=embedly&utm_term=cbls9g

 

[TechyInAZ]

rigg42 nailed it. That is exactly what you should be doing.

1.45v might seem high, but because the CPU only runs at those voltages on very low loads and only during low core count workloads, the amount of amperage going thru the chip is too low to cause degradation.

 

[J_I_M_M_Y]

Sorry to bump this thread but I've been having the same 'issue' with my 3600X & B450M Mortar. I've done plenty of reading on it, including about how certain monitoring software can put the CPU under load and give false idle readings. There do seem to be some mixed messages around this, though.

On one hand, AMD is saying that some programs aren't giving an accurate picture and we should use CPU-Z as that is more likely to show true idle voltage/temps. I've done that, but at stock settings I'm getting virtually the same core voltage (just under 1.45V) and temp (fluctuations between 40-52 deg C) readings at idle as I was with HWMonitor, Open Hardware Monitor and Ryzen Master. On the other hand, @TechyInAZ and @rigg42 are saying that this voltage is normal at idle, which to my amateur ears doesn't sound consistent with AMD's line about false idle readings.

(EDIT: I should add that I am seeing the voltage drop to the mid 1.3s under load.)

Also, if these voltages are normal at idle, should I be concerned about the temperatures? Surely jumping above 50 deg C at idle is on the high side? The Wraith Spire isn't the world's best cooler but it's not awful either.

I swapped from AMD Balanced power plan to Windows Balanced, which lowered the idle voltages and temps I'm seeing by about 0.1-0.15V and 5-7 deg C respectively. My quick benches on CPU-Z seem to indicate a modest performance hit as a result, though.

Any clarity would be greatly appreciated!

 

[rigg42]

Sorry to bump this thread but I've been having the same 'issue' with my 3600X & B450M Mortar. I've done plenty of reading on it, including about how certain monitoring software can put the CPU under load and give false idle readings. There do seem to be some mixed messages around this, though.

On one hand, AMD is saying that some programs aren't giving an accurate picture and we should use CPU-Z as that is more likely to show true idle voltages/temps. I've done that, but I'm getting virtually the same core voltage (just under 1.45V) and temp (fluctuations between 40-52 deg C) readings at idle as I was with HWMonitor, Open Hardware Monitor and Ryzen Master. On the other hand, @TechyInAZ and @rigg42 are saying that these voltages are normal at idle, which to my amateur ears doesn't sound consistent with AMD's line about false idle readings.

Also, if these voltages are normal at idle, should I be concerned about the temperatures? Surely jumping above 50 deg C at idle is on the high side? The Wraith Spire isn't the world's best cooler but it's not awful either.

I swapped from AMD Balanced power plan to Windows Balanced, which lowered the idle voltages and temps I'm seeing by about 0.1-0.15V and 5-7 deg C respectively. My quick benches on CPU-Z seem to indicate a modest performance hit as a result, though.

Any clarity would be greatly appreciated!

1.4+ volts is not normal on idle. If you are seeing that while sitting at the desktop it's probably because a lite load either from the monitoring software or windows background tasks etc. are loading the core and its boosting full tilt. So you are in fact not actually idle. This is the behavior I have observed on 5 different motherboards and 5 different 3000 cpu's. Seeing boost voltages that high when lightly loaded is normal and safe and is usually only a reaction to monitoring software when just sitting there watching the software with nothing else running. AMD's position on this is consistent with my own findings.

Assuming you have followed AMD's directions and are on win 10 1903, Chipset driver 1.07.07, have vcore set to normal instead of auto (if there is an option for it), have no other software open, and are still not seeing less than 1v in CPUZ then you have a separate issue other than the observer effect outlined above. I'm starting to think this is only going on with certain bios versions on certain motherboards. I've seen people posting about high temps at "idle" a lot with MSI. I don't have one on hand to test unfortunately and it may be why I have not seen this issue across multiple other motherboards. This appears to possibly be a bug where the CPU is actually just sitting at 1.4v regardless of load and not shifting dynamically like it is supposed to. If you read the AMD post carefully they seem to be aware of this and are probably working on a fix. A sub par cooling solution would be severely taxed in this scenario.

To anyone who is actually seeing this behavior after taking the suggested steps I suggest you fill out the form from AMD linked in the reddit post. It would be helpful to everyone if we can get this issue resolved so everyone can fully enjoy their new CPU's. In the mean time if you are comfortable dialing in a manual all core overclock at 1.35 or so volts it should get you most of your performance back with safe temps until a proper fix is available.

 

[J_I_M_M_Y]

1.4+ volts is not normal on idle. If you are seeing that while sitting at the desktop it's probably because a lite load either from the monitoring software or windows background tasks etc. are loading the core and its boosting full tilt. So you are in fact not actually idle. This is the behavior I have observed on 5 different motherboards and 5 different 3000 cpu's. Seeing boost voltages that high when lightly loaded is normal and safe and is usually only a reaction to monitoring software when just sitting there watching certain monitor software with nothing else running. AMD's position on this is consistent with my own findings.

Assuming you have followed AMD's directions and are on win 10 1903, Chipset driver 1.07.07, have vcore set to normal instead of auto (if there is an option for it), have no other software open, and are still not seeing less than 1v in CPUZ then you have a separate issue other than the observer effect outlined above. I'm starting to think this is only going on with certain bios versions on certain motherboards. I've seen people posting about high temps at "idle" a lot with MSI. I don't have one on hand to test unfortunately and it may be why I have not seen this issue across multiple other motherboards. This appears to possibly be a bug where the CPU is actually just sitting at 1.4v regardless of load and not shifting dynamically like it is supposed to. If you read the AMD post carefully they seem to be aware of this and are probably working on a fix. A sub par cooling solution would be severely taxed in this scenario.

To anyone who is actually seeing this behavior after taking the suggested steps I suggest you fill out the form from AMD linked in the reddit post. It would be helpful to everyone if we can get this issue resolved so everyone can fully enjoy their new CPU's. In the mean time if you are comfortable dialing in a manual all core overclock at 1.35 or so volts it should get you most of your performance back with safe temps until a proper fix is available.

Thanks for that. Yeah, I'd followed all the steps in the AMD reddit thread except setting the vcore to 'normal' - the only options I have are auto, override, offset and 'AMD overclocking'. I'd also submitted the form. The only software i had running (other than Windows Defender) was CPU-Z and voltage wasn't dropping while using AMD Balanced.

I'll use HWInfo tonight instead and see if that shows anything different. Failing that, I might go for a manual overclock and see how that performs (although this will negate any power saving options and keep the clock speed up, yes?). I'm not opposed to just leaving the power setting on Windows Balanced to keep the voltage under control, though. The performance hit isn't massive and I've just come from an overclocked 3570k, so I'm not going backwards!

 

[rigg42]

Thanks for that. Yeah, I'd followed all the steps in the AMD reddit thread except setting the vcore to 'normal' - the only options I have are auto, override, offset and 'AMD overclocking'. I'd also submitted the form. The only software i had running (other than Windows Defender) was CPU-Z and voltage wasn't dropping while using AMD Balanced.

I'll use HWInfo tonight instead and see if that shows anything different. Failing that, I might go for a manual overclock and see how that performs (although this will negate any power saving options and keep the clock speed up, yes?). I'm not opposed to just leaving the power setting on Windows Balanced to keep the voltage under control, though. The performance hit isn't massive and I've just come from an overclocked 3570k, so I'm not going backwards!

HWInfo does cause the cores to boost on the Ryzen balanced plan. That is still what I'd use for monitoring an overclock though. Ryzen is pretty easy to OC. I'd start at 1.35 vcore with medium LLC at 4.1 ghz and work your way up in 25-50mhz intervals using Intel burn test as a quick proof of concept. You can usually get pretty aggressive on LLC just make sure a full load isn't boosting voltage above what you set in bios on SV12 TFN. Once you hit the wall you can fine tune and run some long term stability tests. I like an hour of P95 26.6 and 8 hours of realbench with half of the RAM. I try to keep sub 80 c in these type of tests. Good luck.

 

[drea.drechsler]

.....
Is this safe being round cpu core ? cos on my 1st gen it was never this high ...

Thanks

Just do yourself a favor and go read this:

View: https://www.reddit.com/r/Amd/comments/cbls9g/the_final_word_on_idle_voltages_for_3rd_gen_ryzen/


TL/DR: many monitoring utilities can make the CPU erroneously report high voltages. Be careful what you use, and also be careful how you go about your 'overclocking' adventures.

In fact, I don't think it's gonna be very fruitful to get into it too deep at all until BIOS's are a bit more stable. If it ever proves fruitful... it's far more likely to just suck way more power and generate a lot more heat all for a tiny bit of performance gain.

 

[rigg42]

Just do yourself a favor and go read this:


TL/DR: many monitoring utilities can make the CPU erroneously report high voltages. Be careful what you use, and also be careful how you go about your 'overclocking' adventures.

In fact, I don't think it's gonna be very fruitful to get into it too deep at all until BIOS's are a bit more stable. If it ever proves fruitful... it's far more likely to just suck way more power and generate a lot more heat all for a tiny bit of performance gain.

Yeah I already posted this earlier in the thread. I think he is having a different issue related MSI's Bios where the voltage just stays at 1.4 all of the time. I can't confirm without an MSI board to test but I keep seeing this issue pop up on MSI boards around here.

I'm specifically looking for reports where the voltage is stuck at a particular value, or a small range of values, around 1.4V--no matter how long you sit there and watch it.

This is a different issue than the observer effect.

Considering his voltage seems to be stuck at 1.4v regardless of the load the power usage argument doesn't really hold water. Even with a working bios unless going completely overboard on voltage the heat and power are negligible anyway based on my testing. Maybe 10 watts more at idle and typically uses about the same power and temps at load at 1.35 vcore. It's not a laptop. Who cares about 10 watts at idle. The one disadvantage to manual OC in the real world seems to be a slight decrease in single core depending on the CPU and how well it OC's.

 

[J_I_M_M_Y]

HWInfo does cause the cores to boost on the Ryzen balanced plan. That is still what I'd use for monitoring an overclock though. Ryzen is pretty easy to OC. I'd start at 1.35 vcore with medium LLC at 4.1 ghz and work your way up in 25-50mhz intervals using Intel burn test as a quick proof of concept. You can usually get pretty aggressive on LLC just make sure a full load isn't boosting voltage above what you set in bios on SV12 TFN. Once you hit the wall you can fine tune and run some long term stability tests. I like an hour of P95 26.6 and 8 hours of realbench with half of the RAM. I try to keep sub 80 c in these type of tests. Good luck.

Thanks, really appreciate the input.

I have prime 95 installed already so I'll look at manually setting/testing voltage and clock speed, although with Windows Balanced I'm getting about 4.1GHz under load with vcore at mid 1.3s anyway, while also having the benefit of low core speeds at idle. It will be interesting to see how my benches change (if at all) going from that to manual settings.

One of the difficulties with the MSI AM4 beta bios (not sure if its the same with other brands) is that there are so many CPU features set at 'auto' at stock - the various boost modes, for example. Presumably I should be setting these to disabled where I can if I want to manually set vcore and clock speed. My previous overclocking experience was with a very different Asrock z77 bios that didn't have so many automatic features.

And re your response to drea.drechsler - yes, CPU-Z with no other software running shows core voltage at 1.448V with very little fluctuation at all, when I'm using AMD Balanced with BIOS settings virtually at stock. Hopefully the next bios update from MSI (or chipset driver update from AMD) fixes the issue so I can just set and forget.

 

[rigg42]

I have prime 95 installed already so I'll look at manually setting/testing voltage and clock speed, although with Windows Balanced I'm getting about 4.1GHz under load with vcore at mid 1.3s anyway, while also having the benefit of low core speeds at idle. It will be interesting to see how my benches change (if at all) going from that to manual settings.

One of the difficulties with the MSI AM4 beta bios (not sure if its the same with other brands) is that there are so many CPU features set at 'auto' at stock - the various boost modes, for example. Presumably I should be setting these to disabled where I can if I want to manually set vcore and clock speed.

It likely won't matter if manually setting multiplier and voltage. There is no need to disable a bunch of stuff.

 

[djxpress]

I have this issue as well, following

 

[jdq1412]

People saying that its fine for the cpu to be running at 1.4v+ is fine at low usage since it doesnt have much current running through it, dont know what they talking about. My 3800x sits at 4.3ghz 1.4v+ at idle all the time and temps above 60c. Im using a 240mm aio and my cpu hits over 80c while installing winrar. Amd says its fine for the cpu to run at 1.4v+ and high temps, because they wont admit that theres a problem with their chips. They lied about the boost clocks just to look good to the audience. They released a bios update addressing the boost speeds which are now achievable on a single core for like 0.5sec with optimal cooling. My 7700k ran at 5.0ghz at 1.35v and never went above 75c under full load.

 

[rigg42]

People saying that its fine for the cpu to be running at 1.4v+ is fine at low usage since it doesnt have much current running through it, dont know what they talking about. My 3800x sits at 4.3ghz 1.4v+ at idle all the time and temps above 60c. Im using a 240mm aio and my cpu hits over 80c while installing winrar. Amd says its fine for the cpu to run at 1.4v+ and high temps, because they wont admit that theres a problem with their chips. They lied about the boost clocks just to look good to the audience. They released a bios update addressing the boost speeds which are now achievable on a single core for like 0.5sec with optimal cooling. My 7700k ran at 5.0ghz at 1.35v and never went above 75c under full load.

Actually you don't know what you are talking about. Go read up on Ohm's law and electromigration and then get back to us. As long as the CPU isn't pushing a high current load it will pump the voltage and clocks and use up temperature headroom. Voltage is meaningless unless you also talk about current draw. This is how the 3000 CPU's are supposed to work. It isn't an Intel CPU and doesn't behave the same.

Unless your CPU is running above 1.325v or so on a high current all core load (like a stress test) It's doing what it's supposed to do. This is a possibility and I have read about people having issues with the motherboard constantly running at 1.4v regardless of load. This is clearly a motherboard/bios issue. So unless you have some useful data to provide including your hardware specs, bios settings, and screenshots from HWI64 at various loads, then maybe you should pipe down.

The boost issue was "legit" and they fixed it so get over it. Intel also markets it's max boost as single core. They also pretty much lie about the TDP of their CPUs. If you want your Intel CPU to meet TDP it's going to be at base clock under full load. You are comparing thermals to a CPU with half the cores and threads. Pop a 240 AIO on a 9900k at 1.35v and 5ghz and see how that works out.

All of that being said your temps do seem a bit high. You are not actually observing idle behavior. That much is clear. You need to be at the desktop with everything in the background killed and nothing but CPUz or Ryzen Master running if you want to observe actual idle behavior.

Assuming you aren't just creating observer effect with your monitoring software, don't have the wrong chipset driver installed, have the proper power plan selected, and have a properly working bios, I'd bet your cooler isn't working properly. Or it isn't installed and/or pasted properly.