• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Ryzen 7 3700X - 4,2GHz 1,25v

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Naco Omnis

Registered
Joined
Jan 8, 2020
Dear overclockers,
I just wanted to ensure, is it possible I manually overclocked Ryzen 3700X with only VCore 1,25v and SoC 1,05v? Everywhere I check the voltages are higher. After few tests the CPU is stable.
Can someone confirm I didn't do anything wrong?
 
That would depend on how long and what programs you used for stability testing. A few tests does not really tell us anything. Run something like Prime95 for 30 Min and if that passes you have at least a basic stability at least 2 hours using Prime or a similar product to really test your stability is needed.
 
Hi, thanks for your replay,
I will do some longer tests on weekend to be sure it works well. Thank you for suggestion.
I just realized that somehow in BIOS Voltages (VCore and SoC) mode has changed from Override Mode to AMD Overclocking (AMD has offset option). Which is better?
I have an additional Offset option but I don't prefer to use it.
I have one more question bothering me. My Voltages are constant all the time, even with no single program running (not mentioning HWiNFO64). Should I use some offset in that AMD overclocking mode? I've read 2 different opinions. Some people say it makes worse performance and some say it is good cuz of less voltage and lower temps...Or maybe Cool'andQueit option should be turn on? Or should I use CPU LoadLine Calibration Control?
Do you have an opinion about that?
 
Override is simpler to work with and you can implement it in conjunction with LLC to keep voltages steady or to supply a little voltage boost under heavy load.

I would suggest downloading and installing HWInfo64. Have that app open as you are doing a significant stress test. It will give you a lot of detail about voltages and temp for CPU and motherboard power components. Not sure what you have been monitoring your voltages with but your voltages may be higher than you think when under load.

AMD says 1.35 24/7 is safe voltage that should not cause undue silicon degradation.
 
There are some problems with stability on Override as I saw. Strange... Same Voltages as on AMD Overclocking.
I use HWInfo64 to check everything and what's concern me also is the fact, when my PC is idling or not on heavy load the Cores and CPU's voltages are the same (~1,25V):
Idle Voltage.JPG
and when I use for example Cinebench, they change like this:
heavby voltage.JPG
Isn't this strange?
 
It's normal for voltage to drop under heavy load. That's what LLC is for. You adjust the LLC to compensate for vdroop.

Offset may be giving you instability because of the swings in voltage as it's ramping up and down. In my experience, offset is a pain in the neck to calibrate. But your max voltage is quite low. Significantly lower than AMD's state threshold for accelerated silicon degradation. I wouldn't worry about that. But first, you have to make sure it's stable at those voltages. I would suggest a couple hours of OCCT or Prime95 or Realbench.
 
So you suggest a higher voltage with LLC turned on? I though lower voltages are what we are looking for.
There is something that bothers me, from time to time, more often when I start some programs/games, my PC freezes for 1 sec and makes a buzzing noise. I am afraid I might have did sth during my "overclocking".
Here's a sample:
https://files.fm/u/taggdsq3
 
Strange, I run OCCT for 3h and everything was ok (max temp 68) and next day I run Prime95 and i needed to turn it off when i saw 94 temperature after 20 min;/ Is it really so much more stress-testing than the others benchmark/tests?
 
What cooling are you using? 94c is getting warm but I'm not surprised running Prime95
 
Just common fortis 3 but still, I am surprised about such temperatures difference between prime95 and other tests;/
 
I think the reason P95 is driving up temps so high is that OCCT and many other stress testers do not contain AVX3 testing code. The newest Prime95 releases utilize AVX3 intructions sets. There is AVX, AVX2 and now AVX3 and they are progressively more demanding on the CPU. For this reason, many people have quit using P95 as a stress app or install older, pre AVX3 releases of it. AVX can also be switched off in the P95 ini file. The newer P95 versions drive temps up ridiculously and dangerously high; temps you would never see in real life use. Asus recommends not using it for stress testing when overclocking.
 
I ve just find out I had OC Genie 4 turned on. This allowed me to have such low voltage with 4,2. I don't know what it does because all options are the same as before. Strangely without that option I can't set voltage as low as before. But the strange bussing noise has gone. The temperature are a little bit higher then before (higher voltage, now I have 1,35V with LLC mode 4).
Is 45-50 degrees with only windows running aren't too high? Like I said before I have only Fortis 3 but with pretty good thermal paste.
So, anyone know what this Genie 4 does exactly?
And I have CPU LLC and CPU NB LLC in BIOS. Anyone knows what this second thing does?
 
CPU NB LLC targets the CPU IMC (Integrated Memory Controller) and cache I think.

Don't use the OC Genie 4. It's a poor substitute for manual overclocking. It also may hide manual overclocking options when operative.

45-50c sounds to warm for idling in Windows. High 20's to mid 30's are typically what we see in Windows while idling.

The Fortis 3 cooler is a big chunk of metal and would seem capable of giving better temps than what you are reporting.

The goal is to attain a stable overclock with the least amount of voltage possible to maintain that stability. LLC helps with that by supplementing the voltage under load. If the voltage under load drops to0 low you will get instability. LLC kicks in while under load, "Load Line Calibration" is what it stands for.

If If were you, I'd start over. Disable OC Genie 4 and put the system back to default, stock settings and check temps, both idle and load. Get a baseline before overclocking.
 
Last edited:
Ow, the stock settings was awful. It was set on auto VCore with 1,45 voltage most of the time, cool and quit amd mode on(i AM not fan but maybe should I enable it?), 60degreees on idling. Disaster...
This is why I started "overclocking"/changing things. To adjust it a little bit. I d never start OC with my poor knowledge and on this medium range setup. Now I am starting being addicted to change stuff:/
Well, I need to change thermal paste again then. It was a little bit dense imo so it might didn't spread enough and it is too thick layer of it now:/
 
Most of us don't try to spread thermal paste. We deposit a blob about the size of a pea in the center of the CPU face and then just clamp the cooler on. The clamping pressure of the cooler will spread the paste between the two flat surfaces without introducing skin oil from fingers. When bolting down the cooler, make sure you turn the mounting screws a little at a time (like one turn) in an "X" pattern to get even pressure. Of course, if this the clip on kind that doesn't apply.

What you want to do if your are content to run at stock frequencies is to manually set the CPU LLC to the next to the highest level. So if 5 is t he highest supplement, set it to 4 as an example. That usually will give a voltage supplement under load that will keep the CPU vcore about the same under load as what you set it to manually.

Turn off the genie.

Manually set a core voltage of say 1.25, save the setting and see if it will boot into Windows. Stress test with OCCT or Realbench for 15 minutes. If it's stable, go back into bios and lower the voltage core by .01 and retest for stability. When you lower the core voltage to the point where you can't pass a 15 minute stress test or it won't complete a boot, then raise it back up. Then do longer stress tests, at least one hour and better, two hours to confirm stability. Increase core voltage by .01 volts until it stable for a long test of 1-2 hours.
 
I did step by step what you suggested but stupidly I used "AMD Overclocking" feature instead "Override". I assume the only difference is the additional "offset" option (it combines Override+Offset I think or correct me if I am wrong). Sadly I can only set "Auto" or -25, -50, and so on... There is no "dsabled" option. Should I do it again with "Override" mode? In fact, I see while idling a little bit lower voltages then I set so this "offset" sadly works :/
IMG_20200110_090001.jpg

Can I leave 1.05 VSoC? Or is it too high also?
 
Last edited:
I hate my lack of knowledge and not presision explanation of features. Now I changed in BIOS Precision Boost Overdrive from disabled to enabled and my cinebench20 score rised 400 points. What this PBO exactly does? I know it allows to run beyond limits but how it works when I OC it manualy and turn it on in BIOS.
 
Yes, use Override.

I would leave the VSoC alone. If you only have .025 voltage increments to work with then use that increment. There doesn't seem to be the fine increments of control present with AM4 boards as there is with Intel chipsets, especially on the lower end AM4 boards. But it is what it is.
 
What about LLC? On 1,2 mode it increases voltages under load. On 3 mode it is stable (via the HWInfo), 4,5,6,7,8 modes are decreasing voltages under load. Since I set VCore on 1,2875v with 4,25GHz (low IMO) I think decreasing is not a good option so I left it on MODE 3 to avoid Vdrop. Am I correct in my way of thinking or something is missing me?
 
Okay, then. Put LLC to 3 mode. Try it. Experiement. That is what this is all about. I can't tell you exactly what will work on your system since every combination of components is different.

I sense you are up tight about all this. And that is normal for beginners. Maybe it will help if I point out that the components you are working with are not as fragile as you might imagine.
 
Back