cancel
Showing results for 
Search instead for 
Did you mean: 

OCing Ryzen 1800X

dusank
Level 7
Hello,

can I please get some guidance for OCing my CPU. I've been following this guide:

I have my memory set to DOCP Standard (3200MHz).

I've tried OCing with the following settings (everything else is on Default/Auto):

CPU Core Ratio: 40
CPU Core Voltage: Offset 0,025 - 0,0625
Core Performance Boost: Disabled
CPU LLC: Level 1
Global C-State Control: Enabled
Fast Boot
CSM: Disabled

My PC boots up fine at vcore offset 0,025 but with AIDA64 System Stability Test running the PC freezes after cca. 1,5 minutes.
I've tried increasing the offset as recommended in the guide but the only thing that changes is that the PC freezes after a longer time (18 minutes @ offset 0,0625).

I've tried all the values between 0,025 and 0,0625.

Is it safe to go higher? What is the safe maximum for vcore offset?

Are there any other settings that might help stabilize my CPU at Core Ratio 40?
250 Views
38 REPLIES 38

Sigtran wrote:
Hm, having some difficulties replying to this thread...

Anyhow, here is how much I managed to sqeeze from Cinebench, that nicely respond to memory speed and timings (an CB bias option in bios)

Unfortunately, I cannot make my cpu a 4GHz fully 100% stable, so for every day keep 3.9GHz at 1.39V.

70817

WinRAR is also a good benchmark that VERY HEAVILY respond to memory timings, much more then CPU speed 🙂 For example 3.8GHz CPU with 3466CL14 memory with optimized timings is way faster then 4GHz CPU 3200CL14 default memory subtimings. Try it...



You have pretty tight timings there. I've tried to run with tRAS = 22 or 28 and tRFC = 256 or 260 but unfortunately it was not stable for 20minutes under AIDA64 SST.

With memory set at 3466MHz my motherboard refuses to boot (black screen, had to reset CMOS).

At 3333MHz it runs stably under AIDA64 SST for 20 minutes with GearDownMode enabled.

70841

70832

70833

Sigtran wrote:

WinRAR is also a good benchmark that VERY HEAVILY respond to memory timings, much more then CPU speed 🙂 For example 3.8GHz CPU with 3466CL14 memory with optimized timings is way faster then 4GHz CPU 3200CL14 default memory subtimings. Try it...
70819


WinRAR Benchmark:

70838

70839

I've tried running AIDA64 SST with only system memory stress test enabled. It runs for two hours already and the system appears to be rock solid.

Up until now I have been running CPU, FPU, Cache and Memory stress tests all at once with AIDA64 SST and with many settings I've tried it crashed/froze before the 20mintute tresshold.

How do I interpret these results? If OCed CPU run stably for hours and same CPU OC with tighter memory timings and higher memory frequency tends to be hardly stable for 20 minutes and stress testing only memory with same timings and frequency is absolutely stable does that mean that my ram OC destabilizes my CPU OC?

Should I try decreasing the CPU multiplier until I achieve at least a coulpe of hours AIDA64 SST stability with my new memory timings/frequency and then possibly try even tighter memory timings?

unknownmiscreant wrote:
If you try tune your ram a bit more you'll likely be able to get up to 100 more cinebench points... I run 3466 CL14 b-die with tight subtimings at 1.42v (with active cooling) on a 1700X at 4GHz. The ram we have looks to be the same stuff, just with different heat spreaders and rgb cansr xD. So if you can up the voltage a little to get it stable it'll be worth it in my book.


I had to up the ram voltage to 1,4V already and I dont have any additional cooler on my ram sticks. I really doubt that 1,42V would make my ram stable at 3333MHz ...

What stability criterion are you using? My PC was running at 3333MHz just AIDA64 SST did not run stably for 20mins.

unknownmiscrean
Level 8
Strange, I could get 3600 CL14 booting at 1.47v although cinebench crashed my machine. I could probably get it stable if I upped the voltage. I did menage to get 1 (rather iffy) run in and there was no noticeable performance difference between 3466 and 3600.

Thats pretty good scores you're getting. Good work.

One thing to note is I've found 1701 seems to be slightly kinder to tight sub timings. 1701 boots with my RAM oc, yet 3501 bootloops on q-code 05.

unknownmiscreant wrote:
Strange, I could get 3600 CL14 booting at 1.47v although cinebench crashed my machine. I could probably get it stable if I upped the voltage. I did menage to get 1 (rather iffy) run in and there was no noticeable performance difference between 3466 and 3600.

Thats pretty good scores you're getting. Good work.


Thanks I am happy with it, just a little bit concerned about stability as it cant take much more than 20mins of AIDA64 SST and sometimes I think even reaching the 20mins might be problematic.

What stability criterion are you using? (AIDA64 SST, RealBench Stress Test, ... how many minutes?)

dusank wrote:
Thanks I am happy with it, just a little bit concerned about stability as it cant take much more than 20mins of AIDA64 SST and sometimes I think even reaching the 20mins might be problematic.

What stability criterion are you using? (AIDA64 SST, RealBench Stress Test, ... how many minutes?)


I ran 5 and a half hours of memtest 86 to verify the memory. In terms of CPU I can run around 20-30mins of prime95 (idk if its avx) although I think that was when I was running 4.05GHz. I have however run around 10mins of solid cinebench, and some daily usage stuff like video editing. Since I never manage to fully load the CPU having it not 100% stable at full load for extended periods in not an issue for me.