Hey guys, I've spent the past week, day and night, trying to overclock 2 sets of Patriot 4400 CL19 B-Die without much luck and would appreciate some advice. The IF appears stable at 3800/1900 with auto/loosened timings (500%+ HCI Memtest, 8+ hours of P95 Large FFTs + GPU stress) but not having much luck tightening things from there. Tried just about every guide and dug up every thread I can find but as far as I can tell, even very safe/conservative timings (e.g. as described by @Taraquin
in this guide: A guide to ram overclocking on Zen 3
) cause memtest to throw errors.
One thing I've noticed is that I'm unable to get voltages above 1.45V without causing the system to become unstable. From what I've seen, there are many examples of people running on >1.5V without an issue on these kits in a 4x8 configuration (e.g. @Bloax
in this thread: https://www.overclock.net/threads/a...0-besides-xmp-profiles.1798106/post-28977880;
or buildzoid in his video), but the moment I go over 1.45V, things get shaky on otherwise stable settings. Can anyone illuminate for me why this is? My understanding is that B-Die is supposed to scale with voltage and easily hit 1.5.1-6V with active cooling.
At first, I thought it was a heat issue since the 4 sticks are quite tight underneath an NH-D15 cooler, so I attached a spare 120mm Noctua fan as a top intake fan directly above the RAM, but it made no difference. The DIMM voltage-related errors usually happen sooner rather than later (<100% coverage on memtest) so I figure that's not the reason. I tested various RTT and CAD BUS settings based on what I've read on the forums, but again no luck with any of these at 1.50-1.55V (using settings that are otherwise stable at 1.45V):
The best I've managed to do is 3800/1900 at CL15 with loose secondaries and GDM disabled at 1.45V. If I try CL14, it will have issues around 1.45-1.48V (BSODs or significant instability), whereas higher voltages seemingly run fine but throw errors in Memtest after a while. I would be fine with CL15, but it seems unstable even with secondary timings that should not be a problem for B-Die based on what I've read:
(The above failed Memtest at 160% coverage with primaries of 16-17-17-17-33)
When I first started out following the basic steps from this guide (MemTestHelper/DDR4 OC Guide.md at oc-guide · integralfx/MemTestHelper
), even trying to get tRRDS/tRRDL/tFAW to 4/6/16 or tWR/tRTP below 20/10 with everything else on default/loose timings was giving me problems.
I'm currently trying to do this thoroughly and tuning each timing individually starting with primaries but it takes a while because I tend to get memtest errors around 100-400% depending on the setting (TM5 with 1usmus v3 config usually passes 3 cycles when DIMM voltage is set to 1.45V). It's quite disheartening to spend so much time for terrible results so if anyone has some suggestions on what I could try, that would be great.
Current settings in ZenTimings:
Among other things, I have also tried:
-Many combinations of VSOC/CLDO VDDP/VDDG
-Different fan curves, blasting every fan at 100%, etc.
-Tuning secondaries/tertiaries first while leaving primaries loose