Ryzen 3900X overclocking

[zorro20010]

Hi, guys!
Strongly need your help.
Currently my CPU is overclocked up to 4700/4600 MHZ at 1,49V. When I try to reach 4800 even on CCD1, PC reboots after 10-30 minutes. I have tried 1.6 CPU voltage - doesn't help.
Do not worry about temperature, I have custom liquid cooling and temperature doesn't exceed 56 degrees when using 1.6V.
What should I do to reach 4800, to hit this magic frequency? Try 1.7V? Or may be to change some settings in BIOS?

 

[EarthDog]

What are you using to test the CPUso that, at a whopping 1.6V, only reaches 56C? That voltage is too high for daily use. I also believe when you actually stress test it, temperatures will not be under control at that voltage.

 

[zorro20010]

What are you using to test the CPUso that, at a whopping 1.6V, only reaches 56C? That voltage is too high for daily use. I also believe when you actually stress test it, temperatures will not be under control at that voltage.

I use XMRig and Prime95. Prime 95 heats CPU up to 65 degrees. But it doesn't answer the question how to hit 4800 MHZ level.

 

[EarthDog]

I've never heard of XMRig (after googling, that is mining software, not a CPU stress test). What stress test in Prime 95 do you run?

But it doesn't answer the question how to hit 4800 MHZ level.

A bit of patience would be great. I'm asking some other questions as it's a bit unbelievable the voltage you're running and the temperatures you're getting. The voltages you want to run are way too much. So, short answer, you don't have any voltage headroom as those volts are dangerous for 24/7 operation. AMD states 1.35V or so when manually setting things (well aware it goes over that with boost bt default). But 1.5V daily is a ton already, regardless if temps are in order.

What motherboard do you have? BIOS updated to the latest version?
What memory kit do you have, exactly? What speed is it at?
Please post some images of CPUz...the main screen, mainboard, memory and SPD, please. You can host here if you are in 'advanced' reply.

 

[Johan45]

The memory looks like 3600 CL14 at 1:1
As E_D says that voltage is too high regardless of temps. The last recommendation I have seen which was a while ago was 1.35V or 1.45V with a chance of some CPU degradation over time.

 

[freeagent]

Even 1.325v is probably too high under manual control for 24/7 operation under all loads. But if you don't care sure give it 1.7 and see if she do what you want..

 

[bassnut]

The 3900x i had would bench 4700 ish depending on the bench being run but for stability, for me that meant being zble to run X265 for extended runs 4hrs + 4.3ghz and that would run 62 to 65*C on a real big custom loop. Pushing those volts you will see degradation, I might have seen some from pushing my chip and my benching was minimal.

To answer your question 4.8 is probably not reasonably atainable. If you get it it will be some little tweaks that will help most and more volts ..... not recommended.

 

[zorro20010]

Guys, you are looking wrong way )) I am not asking about temperatures, they are fine. I have custom liquid cooling with '-30' temperature in the loop. So the question is how to reach 4800?
Motherboard: Gigabyte X570 aorus master
Memory wouldn't help you as it is stable on it's frequency and working without reboot on CPU speed 4700. So reason is not in memory.
I do not believe in fairy tales about degradation and do not care about it. The problem is that 4700 some magic frequency. Even 4750 goes to reboot.
I don think that if 4700 1,49V is stable 4750 1,6V magickly becomes unstable. May be I need to make some more tweaks in BIOS or rise some other voltages? May be P-state...I do not know what way to go...

- - - Auto-Merged Double Post - - -

You think that additional 25 MHZ requires more than 0,1V difference?

 

[EarthDog]

Degradation is not a fairy tale, lol, but if you don't care, we don't!

RE: memory, that may be true, but who would know without you providing the kit and current settings? We deal with complete n00bs and people who have a clue, but these questions are always asked (nice to know you're sub-ambient cooling...........good info out of the gate, that!). For giggles, have you tried your overclock with the memory at JEDEC instead of XMP? That likely won't change things, but worth a try to 100% isolate the issue.

You think that additional 25 MHZ requires more than 0,1V difference?

Yes. You're at the end of the line for those CPUs, really. To the point where you're at the voltage wall where you need significant chunks of voltage for little MHz gains.

 

[Johan45]

Yes that next 25 MHz probably does require a 0.1 V difference. The frequency to voltage curve with Ryzen is very steep at a certain point. Even at -30°C the temperature is high enough that the CPU voltage requirements are too high to remain stable. I also have a chilled loop, maybe my CPU wasn't as good as yours but here's where mine ran out of gas 4.5 GHz for hard benchmarks at 1.5V.
I have done a lot of benchmarking also with liquid nitrogen. Your CPU isn't likely to go much faster without more cold, it's just the way it is.

 

[Nebulous]

Dam that's alot of voltage I'm at 4.9Ghz with 1.4v. If your chip doesn't pass 4.7Ghz with all that voltage, then you have to come to the realization that's your chip's limit. You know how the silicon lottery works, right?

 

[zorro20010]

memory, that may be true, but who would know without you providing the kit and current settings?

My memory is overclocked, of course ) Standard clock is 3200 CL16. I have overclocked up to 3600 CL14. 15 cycles of TM5 usmus config - no errors. I have tried 3800 but failed(((

- - - Auto-Merged Double Post - - -

I have done a lot of benchmarking also with liquid nitrogen. Your CPU isn't likely to go much faster without more cold, it's just the way it is.

Well I hoped that voltage can solve the issue, but I have faced very similar issues: running CPU at 4700, if the liquid temperature is 0 degrees or higher, computer reboots even if I put 1,55V. And it doesn't matter that CPU temperature is about 65 degrees. It seems that I need more cold...but does the higher voltage compensate higher temperatures? For example CPU temperature is 40 degrees and it is running 4700 under 1,5V. If the temperature would be 60 degrees, would higher voltage help me to save CPU frequency?

- - - Auto-Merged Double Post - - -

Unfortunately ryzen master doesn't allow to set voltage higher, than 1.55V. So every time I need to setup it through BIOS. I think I will try 1.65V tomorrow and will see.
About degradation: even if it is not a fairy tale, CPU are constructed to work 100 years and more, even if I face degradation process and the CPU life would reduce twice, I am not going to use CPU for 50 years )))

- - - Auto-Merged Double Post - - -

I'm at 4.9Ghz with 1.4v.

all 12 cores? 4900 at 1,4V? Really? wow..you have diamond CPU )))

- - - Auto-Merged Double Post - - -

For giggles, have you tried your overclock with the memory at JEDEC instead of XMP? That likely won't change things, but worth a try to 100% isolate the issue.

Yep, I have tried to reach the same speed at 1200 MHZ memory frequency - the same result.

 

[Johan45]

Well I hoped that voltage can solve the issue, but I have faced very similar issues: running CPU at 4700, if the liquid temperature is 0 degrees or higher, computer reboots even if I put 1,55V. And it doesn't matter that CPU temperature is about 65 degrees. It seems that I need more cold...but does the higher voltage compensate higher temperatures? For example CPU temperature is 40 degrees and it is running 4700 under 1,5V. If the temperature would be 60 degrees, would higher voltage help me to save CPU frequency?

That's the point I'm trying to make here. Voltage and temperature are all related. The lower the temperature, the lower the resistance in the CPU which allows operation of same frequency at lower voltage levels. They're all intertwined.
Yes more voltage may help but at some point, it won't until the CPU is colder. You're at that point IMO.

Unfortunately ryzen master doesn't allow to set voltage higher, than 1.55V. So every time I need to setup it through BIOS. I think I will try 1.65V tomorrow and will see.
About degradation: even if it is not a fairy tale, CPU are constructed to work 100 years and more, even if I face degradation process and the CPU life would reduce twice, I am not going to use CPU for 50 years )))

It is real and too much voltage to certain parts of a CPU can instantly kill it. I have ran 2.0 V into a Ryzen 2600 but I was also at -170C at 0C it would probably kill it

all 12 cores? 4900 at 1,4V? Really? wow..you have diamond CPU )))

Neb is running a 5800X which is next generation Ryzen

 

[zorro20010]

It is real and too much voltage to certain parts of a CPU can instantly kill it. I have ran 2.0 V into a Ryzen 2600 but I was also at -170C at 0C it would probably kill it

But what is killing CPU? Voltage, temperature or current?
If 2V at -170 doesn't kill CPU, it seems to me that temperature is killing CPU, not voltage or current. If it would be voltage (current), than it doesn't matter what temperature you have. Just imagine, what would happen if you plug any TV or other device in 380V instead of 220.

 

[EarthDog]

It can all kill it. Just because it's good for a couple of ln2 sessions doesn't mean degradation isn't real. I've had Chips degrade as have many here at this overclocking website.

Do as you wish with it... but you're at the end of line with that CPau it seems.

 

[zorro20010]

but you're at the end of line with that CPau it seems.

I also think so (( it seems, that I need to stop here and wait for zen 3 refresh or upgrade my chiller )))

 

[zorro20010]

Johan45 what temperature on CPU did you have using liquid nitrogen?

 

[Johan45]

Johan45 what temperature on CPU did you have using liquid nitrogen?

To be honest, I don't monitor the actual CPU temperature when using LN2. There's a temp probe in the bottom of the pot which is very near the CPU IHS this is connected to an external handheld fluke type meter. When running harder benchmarks the pot temps can drop significantly but still remain around -160C and close to -190C on idle.