Zen2 Voltages aren't exaggerated, Zen2 is a MESS

[Master Disaster]

1 minute ago, Darkjacky said:

My 3900x likes to request 1.5v stock bios settings. Temperatures spike to 95 celcius and it runs at 3.95-4.025GHz. Decided to take a look at voltages and max boost clocks and found that it can run stable at 4.3GHz all core at 1.275v LLC maxed. Temps never go above 90c now. Pretty sure I could lower the voltage slightly.

I will say that I did a poor job on applying my thermal paste but I cant exactly redo it right now as I don't have more. Once I get my thermals under control I don't think 4.4GHz is that far off.

I greatly undervolted the SOC too as it seemed to make a difference in the heat output and it runs stable at lower voltages. I tried all the way to 0.85v which ran stable. I settled at 0.95v just in case.

 

Honestly if the CPU runs at such high voltages out of the box you should probably not worry about it too much.

The important question, how the eff did you get hold of a 3900X?

[Darkjacky]

21 minutes ago, Master Disaster said:

The important question, how the eff did you get hold of a 3900X?

Ordered it the date of release. Well technically a day later since EU times . It arrived the 18th, 10 days after ordering. Lucky I guess.

You are right though I have been hearing a lot of stories about processor stock being awful.

[wolfsbane3083]

The only thing I've found for me that works is switching to the Windows High Performance Power plan and changing the mim/max states to 1%/100%. This lowers my idle and game temps/voltages quite significantly

[Master Disaster]

35 minutes ago, wolfsbane3083 said:

The only thing I've found for me that works is switching to the Windows High Performance Power plan and changing the mim/max states to 1%/100%. This lowers my idle and game temps/voltages quite significantly

According to AMD staff over on Reddit this is actually not accurate

Quote

What about Ryzen Balanced vs. Windows Balanced Plan?

By now, you may know that 3rd Gen Ryzen heralds the return of the Ryzen Balanced power plan (only for 3rd Gen CPUs; everyone else can use the regular ol' Windows plan). This plan specifically enables the 1ms clock selection we've been promoting as a result of CPPC2. This allows the CPU to respond more quickly to workloads, especially bursty workloads, which improves performance for you. In contrast, the default "Balanced" plan that comes with Windows is configured to a 15ms clock selection interval.

Some have noticed that switching to the Windows Balanced plan, instead of the Ryzen Balanced Plan, causes idle voltages to settle. This is because the default Balanced Plan, with 15ms intervals, comparatively instructs the processor to ignore 14 of 15 clock requests relative to the AMD plan.

So, if the monitoring tool is sitting there hammering the cores with boost requests, the default plan is just going to discard most of them. The core frequency and clock will settle to true idle values now and then. But if you run our performance-enhancing plan, the CPU is going to act on every single boost request interpreted from the monitoring tool. Voltages and clock, therefore, will go up. Observer effect in action!

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=218464f06b0f46ed8574e1698ebe3862&utm_source=embedly&utm_term=cbls9g

[wolfsbane3083]

17 minutes ago, Master Disaster said:

According to AMD staff over on Reddit this is actually not accurate

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=218464f06b0f46ed8574e1698ebe3862&utm_source=embedly&utm_term=cbls9g

That's the thing though, even the balanced power plan wouldn't settle my idle temps/voltages. I understand the difference between the plans but even chrome causes the voltages to peg themselves at 1.45v  (just using CPU-Z to see voltages). Even then, I have a full custom loop with 360/280mm rad and I'm hitting over 75c under load, these are definitely some toasty chips.

[nO_d3N1AL]

I made a post on this too. Idle voltages when in BIOS show 1.47v. But when I start maxing out the cores in a program this go down to around 1.3.

 

I agree that the whole observer effect thing is nonsense, since the BIOS, CPUinfo, HWMonitor and Ryzen Master show the same thing.

 

I've come to the conclusion that everything is fine. My temps are very stable even under load, and at idle around 45c but load is about 60c.

 

If you're worried about it I recommend turning off Core Performance Boost, which drops voltage to 1v and sometimes below.

[AvogadrosDog]

I made a post about this too.

 

The first time I ran my built PC (yesterday) I realised after about 2 hours that once loading HWinfo64 my voltages were pegged at 1.55v on stock. I ran cinebench to see if it would lower the voltage during load and it didn't, the temps rose to 90+ degrees Celsius and then the computer shut off (last recorded temperature was 104 degrees Celsius just before the machine cut - this sounds like the last episode of Chernobyl haha). I let it sit for about 10-15 minutes whilst I researched and then started it up. Voltages never hit that high again and I also downloaded the chipset drivers and switched to ryzen balanced power plan. Voltages now sit at 1.4v-1.48v with occasional falls to 0.9v and at some point the voltage evens falls to 1.1-1.3v. 

 

My only worry is temps. Last night I managed to start my pc and load cinebench with HWinfo64 on. Voltages were about 1.3-1.4v and temps sat mostly at 60-65 degrees Celsius with final temperature max at 73 degrees Celsius. Scored 4700 roughly in Cinebench R20. Glad it's better but if I leave the computer on for an hour or so the temps average starts to rise. Worringly so, I reach these temps on Cinebench - scared to even try Prime95 or AIDA64. I've never been an early adopter so maybe this is normal. I'm reluctant to use my PC and I'm worried I damaged my CPU when it hit high temps and cut off (is it something to be worried about or am I overthinking?). 

 

I am running the Wraith Prism - eventually opting for a Corsair 240-280mm AIO which should help. I don't think we should criticise people for being worried about something they've invested a lot of money into - it's fair to question if everything is okay and ask a wider and more experienced audience for some help or advice. That's what brought me to LTT forums and I've been happy ever since.

 

P.s. I'm running a 3700X on an MSI X570 MEG Ace

[xg32]

51 minutes ago, AvogadrosDog said:

.

The stock voltages on some, or maybe most x570 are rather aggressive, 1.5v oob is insane, as long as boost clocks don''t surpass 4.2 1.4v should be enough for all cpus regardless of silicon lottery, and should be 100% safe to run 24/7 as long as temps are under control.

 

I only went ham on op because he blamed the platform and then these forums for recommending zen 2, which is ridiculous.

[MrMoonboots]

I lambasted the forum for not being able to maintain consistency on their recommendations. I know it's not LTT forum's fault. Its just annoying.
Two weeks ago I posted here and could not get an answer to my question before everyone got done quizzing me about why I would ever want to buy a 2700X when the 3700X was right around the corner.

Now I'm back with the 3700x, with legitimate concerns, met by fanboys like xg32 angrily telling me I should've known better, it's "My fault the zen2 is sitting at 1.4 at its default stock manufacturer settings with no changes" and its "My fault for not just waiting" with no context to understand any of these issues could pop up prior to the build.

Tell me more about how I should have "Just researched every one of the hundreds of possible failures that could occur in every part, including the unreleased ones i was rec'd prior and if it happens to me I shouldve just expected it" it's incredibly productive in answering the question

[MrMoonboots]

I am awake now and going to try some recommendations I found on other forums and report back some results (In between and after work when I can)

[MrMoonboots]

4 hours ago, wolfsbane3083 said:

The only thing I've found for me that works is switching to the Windows High Performance Power plan and changing the mim/max states to 1%/100%. This lowers my idle and game temps/voltages quite significantly

I'm surprised to hear high performance had this effect, but then a lot of recommendations seem to have had different effects for different people. I'm going to make a point to play with these settings today

[AvogadrosDog]

2 hours ago, xg32 said:

The stock voltages on some, or maybe most x570 are rather aggressive, 1.5v oob is insane, as long as boost clocks don''t surpass 4.2 1.4v should be enough for all cpus regardless of silicon lottery, and should be 100% safe to run 24/7 as long as temps are under control.

 

I only went ham on op because he blamed the platform and then these forums for recommending zen 2, which is ridiculous.

I definitely agree with this, the X570 boards seem to be uber generous with voltages and I'm suprised that MSI, in my case, have only put up two bios updates for my board and most of their X570 boards - thought they would be in a hurry to "fix" these issues (make voltages kinder to the system and its temps) but again, they're probably at the mercy of AMD and their AGESA updates and also it's not just about voltages; current is also part of the equation.

 

I've never seen the board hit higher than 1.5v after the incident - I think a one-off fault may have occurred somewhere. The VID is also mega generous (anywhere from 1.43 to 1.45v, with periodic voltages sitting at 1.3-1.4v). My solution at the moment is either to run an offset of -0.05v maybe even -0.1v with PBO on and/or just wait for AMD to release their AGESA update after the last one turned out buggy.

 

Side note: do you think that after my PC hit high temps and shut off that my CPU may have either got damaged or accelerated degradation? I like to think the PC shutting off as a sign that at least the safety measures are in place and working and saved my CPU?

 

 

[MrMoonboots]

41 minutes ago, AvogadrosDog said:

Side note: do you think that after my PC hit high temps and shut off that my CPU may have either got damaged or accelerated degradation? I like to think the PC shutting off as a sign that at least the safety measures are in place and working and saved my CPU?

I'm concerned about this with my random shutdown yesterday, too but I'm hoping its just a BIOS update issue as Disaster stated, I'm certain my temps didn't merit a shutdown but if it keeps happening I may run another stress test and/or reapply my thermal paste if necessary. I keep hearing these CPUs are pretty tough and can handle a lot and I'm really hoping thats the case.

[xg32]

1 hour ago, AvogadrosDog said:

.

 

It already happened, i wouldn't worry about it now that you have the voltages under control, if it's stable on everything atm then likely it didn't degrade at all.

 

If -.1v does work then it's a very efficient setting that i'd settle on and run 24/7.

[Master Disaster]

3 hours ago, MrMoonboots said:

I'm surprised to hear high performance had this effect, but then a lot of recommendations seem to have had different effects for different people. I'm going to make a point to play with these settings today

Keep us updated with your results please, I'm interested in hearing them.

[Mister Woof]

/watches this thread while wearing the Intel dunce cap, smugly smiling.

[Master Disaster]

9 hours ago, wolfsbane3083 said:

The only thing I've found for me that works is switching to the Windows High Performance Power plan and changing the mim/max states to 1%/100%. This lowers my idle and game temps/voltages quite significantly

Holy crap, this actually works. Thanks dude, I've been messing with power profiles for hours and have never seen it drop below 1.46v until I tried this. Now its dropping down to 0.9v, I've seen 1.1v and 1.4v is only when its working.

[AvogadrosDog]

So, quick update:

 

I've done the following tests on Cinebench R20 as it's quick and gives a rough idea of thermals and voltages under load (I know Prime95 or AIDA64 are the go-to's for OCers and for verifying systems stability).

 

Tests:

a) Windows High Performance power plan with minimum processor state at 1% (PBO and all voltages set on AUTO on bios): voltages go to 1.35v-1.4v with temps at 70 degrees Celsius max. Clocks sit at 2.2 GHz at idle with voltages of 0.9-1.1v. Score: 4600

 

b) Ryzen Balanced power plan with minimum processor state set to 1%, everything else left same as prior test: voltages fall to 1.2-1.3v under load with max temps of 71 degree Celsius. Clocks idle at 1.4-1.48v with occasional spikes at 1.5v. Score: 4700

 

c) Ryzen Balanced power plan set to normal with everything else left on bios the same as the prior tests: exact same results as Test b). Max temps 71 degree Celsius. Score 4700.

 

d) Ryzen Balanced power plan at normal, with offset of -0.10v but rest of bios CPU settings were left the same as prior tests: voltages pegged at 1.3-1.4v under load with temps around 74 degree Celsius max. Score: 4200

 

Not sure if this helps or not anyone but these are my findings. I'm beginning to get used to my Ryzen processor and honestly seeing the load voltage drop to 1.187v when running Cinebench at complete stock settings with max temps of 71 degrees Celsius on stock cooler made me feel better. If it wasn't for the prior issue with voltage pegged at 1.550v and the computer overheating and shutting down I'd be okay right now. On the other hand, seeing my CPU get through all these tests and comply with everything I've thrown at it since (played around with setting manual fixed voltages earlier today and then played with offsets and that) and the CPU reacting and POST-ing fine with all these configurations make me feel like everything is fine with the CPU and the high voltages at idle and low voltages under load is just how Zen 2 behaves.

[Master Disaster]

One minor issue I've noticed is the time from pressing power button to anything showing on screen is silly long, sometimes it can be 15 seconds before the Asus splash screen appears.

 

Like I say, it's a minor issue but definitely didn't happen with my 2700X in the exact same board.

[bignaz]

X470 has nothing to do with his problems. 

 

Everyone knows all bios's for all chipsets have issues. Everyone knows software voltage readings are innaccurate.  

 

I know for a fact sensors on my board show .06v higher then what the cpu is actually getting.  Meaning my 1.5v so in reality it's 1.44v. hell some programs won't even show the voltage right even undervolted it still shows normal. 

 

 

Here is an example.

 

 

12 core 24 thread run. Cpuz report 1.404v actual mesured voltage is 1.344v coretemp reports 1.404v as well.

 

Hwmonitor shows the fallowing.

 

1.294

1.288

1.294

1.313

1.281

1.288

1.313

1.294

1.288

1.313

1.288

1.288

 

Safe voltages for an all core load. 

 

Idle 

 

1.473

1.481

1.469

1.488

1.469

1.488

1.469

1.481

1.469

1.463

 

 

Average is around 1.38 for everything. 

 

Voltage reporting is all over the place and programs are way off. I tested on a $20,000+ tektronix scope that's in calibration at my work.

 

My findings are the software is reporting the highest voltage. That voltage was +/- 1% of what I know my mobo's sensors report high by .06v. 

 

 

Now idle. 

 

Cpuz reports. .876v. coretemp .9875v and hwmon reports .988v. 

 

Hwmon and coretemp are not reading my offset. Cpuz reads it but reports the highest voltage of the 12. 

 

 

Your volts will go low during all thread use. I have the absolute best wc loop money can buy without going below ambient  and I'll tell you that an all core 1.44v run brings that loops to it's knees.  On auto my all core voltages are around 1.3. single core in the 1.4 range and that's ok. Light use the chip can handle that

 

 

 

My advice till the bio's get fixed and programs read more accurately are as fallowed.

 

Set soc voltage to 1.0-1.1 max. High soc can and will kill Ryzens 

 

Sec vcore on. -.100v offset. This will hurt performance but safe guard if your motherboard is trying to pump a ton of voltage to the cpu.  Turn that 1.5 to a 1.4 that's safe. 

 

 

Hell I got a -.100v offset but hwmon is showing a high of 1.5v. trust me. Your 3900x gets 1.5v your going to know by the the thermal shutdown.  Again not showing the offset. 

 

Just relax. Keep in mind your temps are going to be high. This is 2 x 7nm die with very focused heat off-center. Most heatsinks are designed for heat in the center not off-center. 

 

 

And just wait for the bios  to roll out. 

[MrMoonboots]

12 hours ago, wolfsbane3083 said:

The only thing I've found for me that works is switching to the Windows High Performance Power plan and changing the mim/max states to 1%/100%. This lowers my idle and game temps/voltages quite significantly

I tried this too, and funny enough it works great! My voltages vary enough between 0.9 and 1.4 I'm certain if I checked hwinfo64 the average might be around 1v or so.
With PBO off my CPU isnt boosting like crazy anymore at the slightest nudge either. Thanks so much wolfsbane! This is what Ill be using for now while I wait for more information.

I was recommended to look into turning on "Cool N Quiet" in my BIOS settings by a few people but for the life of me I can't find it anywhere. Is it not an ASUS option? Is there an equivalent I might look for?

@Master Disaster I will definitely try a few more things here soon, two hours and I'll be off work and I'll try some things and post results.

 

[paulmohr]

cool and quiet should be on by defualt. If you didn't go in and turn it off its probably on already.

[MrMoonboots]

So I'll be honest, I didn't do too much testing after the settings wolfsbane suggested helped so much. I've actually been enjoying my PC, leaving it on, it hasnt had those random shutdowns I experienced. (Settings for me being PBO off, Windows High Performance with min/max states between 1%-%100

What I'm seeing now in CPU-Z makes a lot more "Sense" to me, I suppose.
Instead of my idle voltages sitting at 1.4 and dropping to 0.9 occasionally, it has completely flipped.

When my computer sits idle now with these settings it does so at 0.9, with occasional peaks of 1.4V. (They happen about every 3-4 seconds. I'm not sure if that's too often or not, but its better than it was)

The "Ocean tide" effect of my fans has slowed, I hear them spin up on their own far less.
Also, my temperatures at idle have settled to around 35-45 and fluctuate FAR less

So maybe the Zen2 is made to be able to handle 1.4V all the time for years on end and is totally cool with freaking out and boosting its clock speed to a full 4.2GHz anytime I open an empty text file.. Somehow a part of me still doubts that. I feel like if this were the case AMD would have already been saying so very loudly by this point, (AMD Rob said occasional peaks to 1.4 were normal, he did not specify sustained and in fact was hoping to get information from people experiencing this)

But if I can let my PC relax at idle like it is right now, potentially extending its lifespan, and being able to take full advantage of the 3700X when I want to, why the hell wouldn't I do that instead?

I feel like/hope that whatever recommendations and fixes AMD comes out with will probably end up with people seeing similar results. Because this seems a lot more sane.

The one thing I'm curious about is just a question I have from a lack of experience. I know the max safe voltage on these under load is around 1.45, sometimes even idling I see my voltage peak to 1.515 at highest. It's occasional, but I do see it go over 1.5 Is  that something to worry about?

[Master Disaster]

13 minutes ago, MrMoonboots said:

So I'll be honest, I didn't do too much testing after the settings wolfsbane suggested helped so much. I've actually been enjoying my PC, leaving it on, it hasnt had those random shutdowns I experienced. (Settings for me being PBO off, Windows High Performance with min/max states between 1%-%100

What I'm seeing now in CPU-Z makes a lot more "Sense" to me, I suppose.
Instead of my idle voltages sitting at 1.4 and dropping to 0.9 occasionally, it has completely flipped.

When my computer sits idle now with these settings it does so at 0.9, with some occasional peaks of 1.4V.
The "Ocean tide" effect of my fans has slowed, I hear them spin up on their own far less.
Also, my temperatures at idle have settled to around 35-45 and fluctuate FAR less

So maybe the Zen2 is made to be able to handle 1.4V all the time for years on end and is totally cool with freaking out and boosting its clock speed to a full 4.2GHz anytime I open an empty text file.. Somehow a part of me still doubts that. I feel like if this were the case AMD would have already been saying so very loudly by this point, (AMD Rob said occasional peaks to 1.4 were normal, he did not specify sustained and in fact was hoping to get information from people experiencing this)

But if I can let my PC relax at idle like it is right now, potentially extending its lifespan, and being able to take full advantage of the 3700X when I want to, why the hell wouldn't I do that instead?

I feel like/hope that whatever recommendations and fixes AMD comes out with will probably end up with people seeing similar results. Because this seems a lot more sane.

Let's just hope AMD don't release a new driver update and undo all our work in getting the CPUs running how we want them.

[Dubba]

I just built myself a system pretty much like yours, OP, and it's the first AMD build I've ever done.

3700X on a Asus x470-F with Noctua DH-15 and 2x16GB of 3000MHz CL15 RAM at 3200MHz, and this thing is running HOT. BIOS reported voltages howering around 1.485 and temps of almost 60°, in BIOS, but it was a really hot weather, 32°C inside.. After installation I proceeded to play Doom 2 hours straight and had no problems and got boosts to 4.3GHz even.

 

Anyways, I get your fustration. There's so much happening behind all the settings that doesn't seem to make any sense. I didn't see any effective idling before I updated to 1093 windows and installed chipset drivers. After reading this thread I want to try lowering voltage a bit to see if it helps at anything but I wouldn't worry about the high voltage numbers we're seeing (even though I'm crinching a bit coming from a 1.05 Xeon build). I'm just afraid that we have to wait a bit more to get properly working BIOS versions and I don't want to tinker with Ryzen Master right now because it seems like you can try to apply settings this motherboard doesn't like at all. (FCLK 1700MHz not possiblu whaduheck)

 

We can blame AMD and all the motherboard manufacturers for not learning from the past and providing us with the beta program once again