Welcome Guest ( Log In | Register )

Outline · [ Standard ] · Linear+

 AMD Ryzen /Threadripper OverClocking V1.1, big or small, share your results here...

views
     
CAL V
post May 3 2020, 02:39 PM

Getting Started
**
Junior Member
277 posts

Joined: Dec 2011


QUOTE(owikh84 @ Apr 28 2020, 08:27 PM)
Good news for B-die lovers, ADATA 3600C14 @ 1.45v is coming to our local market soon.
*
Now available on Shopee for RM1,049 biggrin.gif
1 set left only, very limited indeed.

» Click to show Spoiler - click again to hide... «

CAL V
post Jul 9 2020, 11:01 AM

Getting Started
**
Junior Member
277 posts

Joined: Dec 2011


Newbie first attempt on ram overclocking here, so go easy on me pls sweat.gif

Here are my results after some tinkering last night and it's stable at 300% memtest just now.
MSI X570 Unify | G.Skill Trident Z Neo 3600C16 (B-die) 2x8GB | DDR4-3600 14-15-15-30-1T @ 1.41v | SOC voltage @ 1.1v (refence to owikh84 format)
user posted image

user posted image

I've used fast preset timings from DRAM Calculator by 1usmus v.1.7.3 with XMP imported from Thaiphoon.
user posted image

Initially I tried vDIMM @ 1.35v and vSOC @ auto with safe preset timings, no go failed to boot so I had to bump up vDIMM to 1.4v, vSOC @ 1.1v.
It did boot fine and I got a little ambitious, I tried 14-14-14-28-1T with those voltage, it will boot but memtest error at around 60%.

End up punching in the current settings, and set vDIMM @ 1.41v. Passed 300% memtest but I know things can be improved.

Everything on CPU is Auto, cooled by wraith stealth while I'm waiting for some custom wc parts to arrive. Temps are high at 83c and I'm reluctant to hammer it now.

A few things that raised some question, the vDIMM readout in HWinfo is quite a bit higher than what I set in BIOS eg: vDIMM @ 1.41v in BIOS but appears to reach 1.436v in readings. vSOC is still close enough.
Another weird thing is the memtest coverage, some threads are at 400% already but some just only reach 300%, most results that I referred to had about 10-20% difference only. sweat.gif

Any sifu feel free to comment on it and I'll see how I can improve. thumbup.gif
CAL V
post Jul 9 2020, 04:37 PM

Getting Started
**
Junior Member
277 posts

Joined: Dec 2011


QUOTE(nrw @ Jul 9 2020, 11:55 AM)
Have you tried if your 3600 is running on a 1900 FCLK?
I'd be choosing 3800C16 over 3600C14 any day.

Seeing you have the same board I do (which is brilliant for memory OC on higher frequencies btw) I can confirm that MSI loves to bump memory voltage. I'd be inclined to believe 1.436V or something very close to that is your actual voltage despite setting it to 1.41V in bios.
*
I just did some tests, it's not posting at 1900 FCLK, failed the silicon lottery I guess. It's an early 2020 batch, but still not good enough for that. doh.gif
1867 FCLK posted with no problem.

I see, I'll try higher frequency settings in the future. But for now, focusing on finding the right settings for normal usage.
CAL V
post Jul 9 2020, 06:12 PM

Getting Started
**
Junior Member
277 posts

Joined: Dec 2011


QUOTE(nrw @ Jul 9 2020, 04:54 PM)
You could play around with the VSoC, VDDG and VDDP to see if you can get 1900 FCLK running stable.

If that doesn't work out you can still  just use 1867 (3733) then.
Still mostly better even if you are going for C16.
*
I tried applying VSoC 1.15v, VDDG 1.05v, VDDP 0.925v. Still doesn't boot.

I did try something I found online to enable LN2 mode under AMD overclocking, and then go back to set 1900 FCLK in bios, everything else auto including ram. It did boot into Windows after some struggle, but in Windows, the fabric clock was fluctuating at around 1200 FCLK only. Unfortunately the guy only mentioned good luck stabilising FCLK from there and nothing else doh.gif

Maybe I'll just go back and play with 1867 (3733), it was spitting error at 3733 16-16-16-32 @ 1.42v, maybe I'll try bumping up the voltage little by little. Dram calculator suggested up to 1.48v sweat.gif
CAL V
post Jul 15 2020, 09:09 PM

Getting Started
**
Junior Member
277 posts

Joined: Dec 2011


QUOTE(nrw @ Jul 9 2020, 06:28 PM)
sometimes for voltages and stability more is less wink.gif

Try 1.05-1.1V for SOC.
Try VDDG within ranges of  0.95-1.05V (synced meaning iod & ccd)
Try CLDO VDDP in ranges from 0.855 - 1.05V

Once it boots test stability by running aida64 stress cache and streaming a high resolution video at the same time.

If your usb devices/mouse is jerky that will mostly be down to VDDP/VDDG, if your sound is intermittent or fully crashes that'll be either the two above or vSOC.

Good luck. Pretty often if you are willing to spend some time there, you can find a combination of these 3 voltages that will allow your IF to run stable at a higher frequency that auto settings will allow you to.
*
I tried, still 1900 FCLK is impossible on this R5 3600, guess it just doesn’t like it at all, almost instantaneously goes into 07 post code.

I’m currently playing around with 1867 FCLK and 3733Mhz on ram. Don’t have much time to tighten the timings yet.

QUOTE(k!nex @ Jul 14 2020, 11:32 PM)
Dont put vSoC at 1.15V. You put 1.1V also good enough already. You might need to up your VDDP to 1V . Dont use that LN2 mode.
In your BIOS, there is 1 option call  UCLK DIV1 MODE, you set UCLK==MEMCLK .
FCLK put 1900Mhz. RAM set to 3800Mhz , but you need to loosen your timings or up your VDIMM. From my experience , if you wanna do CL14 at 3800Mhz with a Samsung B-die, you need 1.5VDIMM and enable GearDownMode. And TRCD I cannot make it fully stable at 14 neither, I can do 15 only for 100% stable.
I think you are overly conservative with your memory voltage.
*
I tried your method, still couldn’t boot with 1900 FCLK, this cpu just can’t do it I guess. Will play around at 3733 and hopefully bring it to be stable at C14.
CAL V
post Jul 16 2020, 02:06 AM

Getting Started
**
Junior Member
277 posts

Joined: Dec 2011


After some tweaking and here's where I am now. HCI memtest 200% stable (late already, I'm going to bed sweat.gif )

MSI X570 Unify | G.Skill Trident Z Neo 3600C16 (B-die) 2x8GB | DDR4-3733 14-15-14-28-1T @ 1.48v | VSOC @ 1.1v | VDDG @ 1.0v | VDDP @ 0.925v
user posted image

Changed to use HCI memtest instead of runmemtest as the version I used just felt a little odd somehow. Per nrw's suggestion before, I could have ran into insufficient memory and caused the memtest coverage to differ significantly. My available system memory in task manager was 14.2GB, running 1200MB*12 memtest definitely blew that.


QUOTE(nrw @ Jul 15 2020, 11:18 PM)
have a shot CAL V
What I got here is somewhat similar to what you have suggested, but I was already running memtest just now. I'll adjust those timings again. 66.3ns is an improvement over 3600C14 last time but still far from good, some got theirs to 61~62ns. Maybe at 1867 fclk instead of 1900 hindered my chance of going that low on latency?

Does clock speed affect the time taken for memtest? Average clock speed was down to 3.8Ghz and temperature got up to 84c. I also tried your method of running aida64 stress cache and playing 4k video on YT simultaneously. The temps spiked to 92c after 2 minutes and kept rising even with 3 SGT1850 blowing full speed at it sweat.gif. Really need to get my custom wc running asap.

This post has been edited by CAL V: Jul 16 2020, 08:37 AM

 

Change to:
| Lo-Fi Version
0.0348sec    0.55    7 queries    GZIP Disabled
Time is now: 28th November 2025 - 07:23 AM