CH7 + 2700X Overclocking

[Mordorr]

Then, please accept my apologies...

I know im new on this forum, so maybe i don`t know very well how things work, but i look always for common sense...

Please, if you want, give some more directions how to get some more juice from my rig..

Thank you!

[gupsterg]

AIDA64 is terrible to use on Ryzen gen 2 IMO. It creates excessive temps, yours are way higher than what I've ever seen.

SOC in one of your screenie is [Auto] and hitting slightly over1.1V. Some Ryzen 2xxx CPUs will start negatively scaling for RAM MHz attained when effective SOC is greater than 1.05V. I would say it is always best to set SOC manually than use [Auto].

When SOC is left [Auto] and RAM MHz is past AMD officially supported for CPU, the board set SOC based on a fixed "Auto rule" not based on say profiling silicon, if you go manual then you can profile it yourself. All 3x 2700X I have used on C6H/C7H will do 3466MHz with ~0.95V or there abouts.

[Mordorr]

I went there and change SOC to 1.00000....is it better

My actually configuration:






I didnt put here memory options.....

[gupsterg]

Yes SOC@1V IMO is good to use IMO, not too much or too little.

As you have F4-3200C14D-16GFX I would think 3466MHz should be doable, 3533MHz if you have good CPU sample, any higher is golden IMO. The motherboard and RAM kit is not the weak link in you attaining greater RAM MHz, it is CPU and or settings you use.

I would also bring down VBOOT (setting at end of External Digi+ Power Control menu) and VDIMM, always match these settings IMO, when you change either. 3200MHz should be possible with 1.35V. I have so far not had a single CPU out of 5x Ryzen gen 1, 3x Ryzen gen 2 and 1x Threadripper that could not to 3200MHz with 2x8GB Samsung B die at 1.35V. Think of excessive voltage as creating noise on signalling, even if RAM IC can take it. IMO aim to tune SOC & VDIMM to HW in use/OC profile being targeted.

Test using P95 v28.10b1 (link) and or v29.4b8 (link).

First test setup use default RAM MHz (2133MHz) and [Auto] timings, but have Precision Boost Overdrive enabled. Use custom setup of 16 threads, 128K, 128K, tick in place FFT. Run this for ~30-60min, this should tell you if CPU is handling MHz. I'd also run say ~30-60min RealBench Stress mode, Y-Cruncher is another stress test I use, this can have high peak temps but not as crazy as AIDA64 on gen 2 CPU.

Second test setup do not have Precision Boost Overdrive enabled, aim to use The Stilt Safe 3200MHz. Use custom setup of 16 threads 8K 4096K 12750MB in each P95. Then round of testing with RealBench, Y-Cruncher.

[gorpo]

Second test setup do not have Precision Boost Overdrive enabled

Curious, what is the reason for testing with and without PBO?

[gupsterg]

The test method is to determine if each OC type (CPU/RAM) is ok on it's own. Then theoretically when combined together should work as is or need very minor tweaks.

[gorpo]

The test method is to determine if each OC type (CPU/RAM) is ok on it's own. Then theoretically when combined together should work as is or need very minor tweaks.

Ah, I see. Thanks.

[gupsterg]

@gorpo

NP .

For this thread is some of my past test data and rerun today.

The Stilt has an app which can show you what VCORE the worst/best core uses. This app was shared around when gen 2/C7H launched, it will not run if clock date is few months after that time, so that's why in shared screenies time is not realistic for the batch of CPU being tested.

CPO Test data for 2700X 1805 SUS, 1825 SUS and 1835 PGS, link.

I regard 1835 PGS as best CPU sample as it has a) tightest worse/best core VCORE differential b) IMC allows 3600MHz.

Next is album link, this shows 1835 PGS doing PE: Default PBO: Enabled with 3600MHz using The Stilt 3466MHz timings preset. CPU was purchased near end of Dec 18, early Jan is 1st screenie in album showing 32K% PASS in RAM Test, then 8hrs P95 v29.4b8 (8K 4096K 12GB), 2hrs RB, finally ~12hrs P95 v28.10b1 (8K 4096K 12GB). Today I revisited the profile, I tested at least 6 back to back RAM Test runs. The first being a full post and 5 other tests are warm posts. Hopefully rerun P95, etc soon, settings txt in ZIP below.

1103_WC_PBOE_3600S_setting.zip

[Mordorr]

I confess i have dificulty in keep with all information about this board/memorys/CPU...

On old days, i put FSB 1:1, control voltage on CPU/Ram and start climbing until instability, then tune for best stability...

With this board, im completely lost (my humble fault).

Why can`t i reach maximum CPU mhz, undependable of Memory speed, and then start play with ram settings, or even the opposite?
I know this are stupid questions, but at this time is exactly what i fell about this combo Board/CPU/Memorys...

I would be happy with all cores (if this is even possible) at +- 4300Mhz and memory's at 3466?

Ok.

Gonna read everything already posted previously and try make a little more tuning?

Thank you in advance!

[Mordorr]

ok.
I did make "past" from your 3600 settings, load 3466 profile memory, save and then did make tests...




Seems some core did fail or stop working....i hope did choose well Prime options....
I forget that on UFI i did "relax" some values on memory voltage....but i can later post here what exactly.

Question:
How to export bios option to a pen, as notepad text?

Thank you in advance!

Edit:
Can`t believe i lost print screen from tests.....arghhh!

Repeat:




Seems 10 minutes after i start i have failures....can`t tell if it is memorys or cores......
More juice?