Question Is this sweet spot for Ryzen 5700X

[Mir96TA]

I was trying to play with Ryzen 5700X PBO setting
I would like to know, Am I at sweet spot ?
PPT=130 TDC=82 EDC=138
Curve Optimizer is Negative 30



I am using CINEBench R23



Thanks in advance

 

[Shmee]

I can't say for sure, but that is a nice memory OC you seem to have going there. What memory are you using and what VDIMM are you applying?

As for the PBO tuning with the curve optimizer, a -30 offset seems very optimistic. I believe I am running -9 all core on my 5950X, though I believe my scalar is 2X. What are you using to test for stability? (Hint: you can use the p95 with corecycler, to test each core boosting for stability)

 

[Mir96TA]

I didn't realize I was really O.C. the memory.
I thought I was running at stock.
This Ryzen can go upto 3600 Memory Frequency but memory I have it is good for 3200 only
Dram Voltage (VDIMM) is coming from XMP profile, I think Asus call them DOCP!

 

[Mir96TA]

I have not used to validate the stability.
I often use enterprise Machine, for example Windows's Server, Firewall management software. That cause heavy memory usage 90 to 95 % of the memory (26 GB to 30 GB).
I can not Offord instability.
My workload is my validation

 

[Mir96TA]

Here is my fine-tuned one.
PPT=140W TDC=90A EDC=150A
All Core have - 30
Scaler is 10 X
200 Mhz
All Core are as a cure optimizer Negative 30
Here is a Cine



I didn't O.C. I need 100 % stable and able to runs weeks non-Stop.
I am not sure, if I can further optimize ?
Any input is greatly appericated

 

[Shmee]

You realize that tweaking with pbo2 can cause instability? It is basically a form of OC, especially with the +200 MHz. That is why you need to stress test

 

[Mir96TA]

You realize that tweaking with pbo2 can cause instability? It is basically a form of OC, especially with the +200 MHz. That is why you need to stress test

No I didn't know.
I thought me running memory nearly to full gives enough sress
I guess I can P90 ?

 

[Shmee]

Yeah, you can run P95, or try core cycler, that can test individual cores and tell you which core fails, if any.

 

[Mir96TA]

I ran it, when I came back System reboot it, with UEFI screen CPU error, Temperature.
haaa aaaa
I guess I need better cooling

 

[Shmee]

More likely your tweaks aren't stable. The CPU should throttle as it gets too hot, but you are basically undervolting to the max, so I am not surprised it crashed. I suppose if your temps are high though, that doesn't help either. What cooler do you have? And when checking the event viewer, are there any WHEA errors?

 

[Mir96TA]

Well I have set Limit on UEFI 90C which is also TjMAX for my CPU.
So MB rebooted and stop at UEFI screen witch CPU Temp error.
Cooler I have is Corsair H55 Quiet
I am going to check WHEA errors.

 

[Mir96TA]

Only error I saw was Kernal-Power 41

 

[Shmee]

Only error I saw was Kernal-Power 41

You could try setting a temperature limit a bit lower, like 85C or 82C or something. This should help keep the core temps from getting out of control during stress testing, gaming and any rendering work.