Possible RAM or CPU issues may cause screen freezes.
Possible RAM or CPU issues may cause screen freezes.
arrived finally so I turned off DOCP and RAM to the 3600 MHz you mentioned. It’s already running the prime95 test for a few minutes after manually overclocking RAM. Should I concentrate on speed or also experiment with timings/voltage? Also, am I still concerned about performance loss when using four sticks—it’s likely there will be some impact. The gap between 6000 MHz CL32 and 3600 MHz timings seems substantial since they’re both set to auto.
Adjust the clock settings and voltage to match the XMP standard. If you lack a label with the numbers on the RAM, turn on DOCP and check the timings and voltage in the BIOS, noting them or capturing a photo. After that, turn off DOCP, configure the settings manually, and gradually increase the speed as instructed. In practical scenarios, these adjustments usually have minimal impact, making it mainly valuable for advanced overclockers aiming to break records or for enthusiasts who enjoy the process.
I followed your instructions, DOCP modifies these settings (DOCP III profile - 4800 MHz). The overclock tuner [Auto] now points to [DOCP Tweaked], memory frequency [Auto] is set to [DDR5-4800 MHz], Trcd [Auto] and Trp [Auto] both at [38], Tras [Auto] at [70], DRAM VDD Voltage [Auto] is [1.10000], DRAM VDDQ Voltage [Auto] matches [1.10000], and PMIC voltages are synced to [Sync All PMICs]. Memory VDD Voltage [Auto] is [1.10000] and Memory VDDQ Voltage [Auto] also [1.10000]. Based on the HW info, the small table in the attachment matches what you described, and using the "enthusiast" profile should work fine. It looks like the prime95 large test has already run for 30 minutes, so this profile should perform well—feel free to adjust frequencies and timings further if needed.
It’s mainly based on trial and error. The DOCP settings seem quite flexible. Also, your voltage is only 1.1 volts, which means you’d need to increase it to 1.35v as recommended by the profile for higher XMP clocks and timing. With four sticks, you won’t be able to reach the most precise timings at the required frequency, but you can still find acceptable settings. Consider boosting the frequency first to establish a baseline, then fine-tune the timing. Frequency plays a much bigger role in performance.
I'm currently operating at the speeds listed in the table. The previous run took about 4800 seconds, and this latest test lasted around 50 minutes. With more relaxed timing, you might reach up to 6000.
Your timing is 38-38-38-70-116 in that screenie. However, continue raising frequency until you reach the limit. Reduce the step size near instability, then gradually tighten the timing. Once you believe a stable configuration exists, let it run overnight.
I observed that the lowest settings always show the first number low. It seems a configuration like 6000 MHz 38-38-38-80 could work. Atm, 5600 running at 30-36-36-75 for 15 minutes, then let it run for at least an hour and try 6000.
Well, 5600 at 30-36-36-75 worked fine for 75 minutes. Then 6000 at 40-40-40-80 failed right away, so I think 5600 is the maximum limit and those settings are already the toughest. Likely 5600 with that pattern will let Prime95 run a full overnight test and hopefully it will succeed.
It seems you're wondering if it's safe to conclude stability after 8 hours. You might want to confirm with more details or check the specific guidelines for your situation.
It's happening again... forced UPDATE keeps occurring. I attempted to reduce RAM speed but now everything freezes on my screen—two RuneScape clients plus video in picture mode. Even after running Prime95, it works fine. I suspect a Windows update issue, though I'm still on the latest BIOS and chipset updates. My GPU is up-to-date too. Anyone have ideas? Also, I lowered RAM speed to default after some tries, but it didn't help.