Help with getting B-die stable on an X470 board

Sumanji · 25 Sep 2020 at 21:44

Hi All,

Having some trouble with getting my RAM stable at its rated timings - would love your guys help please!

I'm using this kit of G.Skill Samsung B-die PC3200 on an Asus Crosshair VII (X470) with a 2700X (full spec in sig)

Was previously using The Stilt's 3200MHz safe profile that is built in to the BIOS, but had to turn off 1T command to get it stable (was Memtest error free for 12+ hrs when I first built this rig).

Recently I have been getting black screen / CTD in Destiny 2 and Borderlands 3, so wanted to double check and make sure it wasn't a RAM problem. I just ran Prime95 (large FFT's) now and it threw up an error after ~9 hours.

Maybe the RAM is slowly pooing the bed, but it's also possible that it was never fully stable to begin with... Is there anything specific I need to do when setting it up? E.g. other voltages or timings, chipset, infinity fabric. Could a weak memory controller on my 2700X also be an issue?

Thanks!

Su

P.S. I tried to post with the XMP (DOCP) profile and I think it failed and reset back to super slack timings of 16-22-22-53-75-2T @ DDR3200. I am running Prime95 again at these settings to rule out straight faulty RAM.

Joxeon · 25 Sep 2020 at 21:51

First I'd try a tad more DRAM voltage maybe up to 1.4 and see if that fixes it but like you say it could also be the memory controller so reducing ram speed by a notch may also fix it.

Sumanji · 25 Sep 2020 at 22:04

Thanks Joxeon - I was running at 1.4V DDR voltage before (when I was using The Stilt's settings), so I don't think it's that.

Which voltage would I play with to stabilize the memory controller (too many options on Ryzen!). Is SOC voltage the right one?

Cheers,

Su

EsaT · 25 Sep 2020 at 22:19

Sumanji said:
Thanks Joxeon - I was running at 1.4V DDR voltage before (when I was using The Stilt's settings), so I don't think it's that.

Samsung B-die is known from being very voltage tolerant because of older bigger manufacturing node.
But what it doesn't like at all is heat with about 50C being limit before stability goes over the edge of canyon.

So is case cooling still working well?
No dusty fans/filters?
No new more power hungry/heat spewing components?

Sumanji · 25 Sep 2020 at 22:45

Thanks for the reply EsaT! Hmm, it has a 140mm fan blowing straight at it and I keep my case pretty well maintained (pics here).

Right now I'm running Prime 95 Large FFT and the DDR max temp reading from HWMonitor is 44C (at 1.35V VDDR).

It's possible the temps were running higher when I was at 1.4V VDDR, but the fact that the system is failing to post at XMP settings makes me think the issue is more likely to do with voltage/timings?

Cheers,

Su

Sumanji · 27 Sep 2020 at 20:32

Update:

Been running the RAM at 2933MHz (14-14-14-34-2T) the last few days and it *seems* fine. (Prime95 Large FFT over night no errors in ~12hrs).

So I am more sure that it's an issue with CPU not being able to keep up with the RAM. The Asus default VSOC seems to be 1.1V+, which afaik is already pretty high. Maybe my CPU is just a bit of a dud on the IMC side?

Cheers,

Su

Sumanji · 3 Nov 2020 at 14:47

Another update, for those following this saga

Since 2933MHz was solid, I wanted to see if I could push it a bit higher...

Seems to be ok at 3066MHz using The Stilt's safe profile (14-14-14-28-50-1T). It's been running Prime95 overnight without errors (so far), but will give it a proper workout with gaming over the weekend, which weirdly seems to be more sensitive to RAM stability!

I do still think the root cause of not being able to run at 3200MHz is a weak IMC though - any ideas on how I would go about testing this theory? So many voltages and different settings...

Cheers,

Su