Originally Posted by Unknownm
can POST 3600Mhz SAFE but fails prime95 4Hours in with complete restart (no bsod just restart)
If I apply higher voltage , prime95 fails faster
SOC (1.05v to 1.15v)
CLDO_VDDP & CLDO VDDP (950 / 950)
When I turn on the PC the BIOS hangs on DRAM Yellow light. I restart and everything POSTs as well
Edit: having cad bus @ 30-30-40-60 gave me 4h prime instead of 2h
SoC gets more unstable/diffrent the longer you have it under usage/load. I've noted that it usually requires more voltage at longer runs, if voltage is low on a cold boot it can work for some time but after it's warmed up it will start to require increased voltages to stabilize as it may suddenly start to throw errors/bluescreens after several hours. Let it cool down and if will settle down after a while but it is easier to just increase the voltage instead to push the edge away from occurring.
procODT and Cad Bus settings have a effect on how stable or fast it can occur. Higher values I've noted usually cause it faster unless they are just a bad/unstable combination. They all play together.
I can for sure say HEAT! is a big factor on this. Open your chassis put a fan on your RAM or backside of motherboard and you can probably run much longer.
Something is drifting with it's tolerances and causes issues at longer loads when it heats up. As a point you might even need different CAD BUS settings for a warm system compared to a cold one. (they might error on cold boot but they will work when the system gets hot enough)
Found most of this testing out 3733-->3800Mhz memory speeds. At the edge of insanity.
NEW: Biostar x480GT8 ~4.0Ghz CPU, 3200Mhz RAM.... not really better
OLD: Ryzen 7 1700 @ ~4.0Ghz, Gigabyte GA-AB350 Gaming 3, F25 BIOS Agesa 188.8.131.52
4x8Gb Kingston HX434C19FB2K2/16 3466C19 1.2V @ 3733Mhz 16.(17).184.108.40.206.tRFC 568 1T 1.490V (Micron E-die 16nm)
Corsair LPX 2x8Gb 2666C16R @ 3200Mhz 14.(17).220.127.116.11...(260) 1T1.470V (Micron B-die 25nm)
Last edited by Nighthog; 05-15-2019 at 08:54 AM.