PC stops working entirely when running Prime95's memory test.
PC stops working entirely when running Prime95's memory test.
I recently upgraded my CPU from an R3 3200G to an R5 5600 on an A320M motherboard. I have two 8GB DDR4 3200MHz modules (Kingston Fury).
With the default BIOS (3200), the PC boots in dual channel but no input is detected.
I can boot into Windows and perform RAM tests; it worked for three minutes before stopping, though I’m unsure if it could last longer. It doesn’t shut down under normal conditions.
After applying a manual underclock, the system runs smoothly.
However, during a Prime95 "memory controller & ram" stress test, things changed: setting dual channel to 2800 caused shutdown after 20 seconds, while 2666 lasted 30 seconds, 2400 lasted a minute, and 2133 ran for three minutes.
Here are the full specifications:
Motherboard: Asrock A320M-HDV R4.0
CPU: Ryzen 5 5600
RAM: 2x 8GB Kingston HyperX Fury 3200MHz
GPU: RX 560 XT / RX 470D (used only 75W)
Power Supply: MSI A650BN
Storage: 1 SATA SSD (128GB) and 1 NVMe SSD (500GB)
One hard drive
Additional tests confirmed the CPU and GPU don’t trigger shutdowns under stress.
Hey there,
Honestly, I wouldn't be using Prime or any similar test with an A320 mobo. Their VRM's are weak and not built for that kind of stress. Why are you going for it? There are game crashes happening—something isn't stable. Just remove the undervolt and any other overclocking settings, and try with stock specs. As recommended, make sure you have a fresh BIOS, because older ones won't work with your CPU. Also, clear the CMOS after updating the BIOS.
Welcome to the forums, newcomer!
Your GPU is an Rx 560xt / rx 470d (it only uses 75w). You should elaborate on this.
mobo : Asrock A320m-hdv r4.0
BIOS version for your motherboard?
I have two 8gb DDR4 3200mhz memory units (Kingston Fury).
Did you buy them together or spaced out over a few months?
PSU : MSI A650BN
How long has the PSU been in your build?
It is suggested to upgrade to the newest software release for better memory compatibility. This could address the ongoing instability problem.
The GPU is an Aisurix RX 560XT, with limited details but essentially a BIOS updated RX470D.
The BIOS version was 7.20 with AGESA 1.2.0.7.
I acquired the RAM about 1½ years prior to getting the CPU and GPU. The entire system was a 3200G prebuilt unit.
At that time, using an old inexpensive Korean PSU I could run the PC at around 2999MHz. It never shut down during gaming; no stress tests were performed back then.
Recently I purchased a new MSI PSU because I encountered a problem about a week ago after upgrading the CPU. The issue occurs when trying to play CS2—after 30 minutes it shuts down mid-game.
I attempted to underclock the RAM and disable the CPU boost clock, but the PC still shut down during gaming and remained stable afterward.
Later I tried compiling shaders on a Unreal Engine project, which also caused shutdowns. To finish the compilation, I reduced my RAM again, then performed a RAM test. The system still shut down.
Hey there,
Honestly, I wouldn't be using Prime or any similar test with an A320 mobo. Their VRM's are weak and not built for that kind of stress. Why are you doing it? The crashes are happening because something isn't stable. Just remove the undervolt and any other overclocking settings, and try with stock specs. As before, make sure you have a fresh BIOS—older ones won't work with your CPU. Also, clear the CMOS after updating the BIOS.
The information provided discusses various troubleshooting steps and considerations for system stability. It mentions updating BIOS versions, adjusting RAM speeds, and addressing motherboard VRM limitations. The user also notes challenges with resetting BIOS settings and booting issues after changes.
From a technical standpoint, you're still pushing your RAM beyond its rated limits at 2666MT/s (1333MHz). If you manage to adjust it to 2133MT/s, it's likely the standard (non-overclocked) JEDEC specification. Exceeding this would be considered overclocking in my view.
CPU-World indicates that the 5600 can operate at DDR4-3200 rates.
https://www.cpu-world.com/CPUs/Zen/AMD-Ryzen 5 5600.html
Are you certain the voltage is set to 1.2V at 3200MT/s? I assume any XMP-based overclock would raise the voltage for DDR4 RAM to a standard 1.35V, which is typical for speeds above the stock 2133/2400 DDR4 JEDEC rates.
I’m aware you use Kingston RAM, but I found this information on the Corsair website, which should apply to your setup.
https://help.corsair.com/hc/en-us/articl...ing-memory
By default, DDR4 operates at 1.2V, while many memory kits support around 1.35V with XMP. Gradually increase the voltage until stability is achieved; it’s advisable not to exceed 1.4V for safety.
If your RAM remains at 1.2V at 3200MT/s, it’s likely the system won’t boot. The BIOS should automatically adjust to a standard 1.35V once you enable XMP. You’ll need additional voltage for higher frequencies.
When needed, manually set the RAM voltage to 1.35V at DDR4-3200 before applying BIOS changes.
I struggled to get an old Ryzen 2600X to boot with DDR-3000 RAM. The 2600X only supports up to DDR4-2933, but I resolved the issue by adjusting the CL(CAS) value for both DIMMs from 18 to 20 at 3000MT/s (actual clock speed was 1500MHz).
In your situation, if 1.35V doesn’t resolve the problem, consider trying CL=23 or CL=24 at DDR4-3200. You’ll need to consult the manual to modify CL after selecting DDR4-3200 in the main menu.
When I reset the CMOS, it sets the system to #1JEDEC 1.2v 3200mhz CL22, so I need to remove one RAM stick and boot from a single stick.
I'm not sure if that was the right approach...
My motherboard's VRM is really weak—it's an A320M (the microATX model). I feel okay with 2666.
Thank you for the image. It’s really helpful and differs from my previous A320M motherboards since your BIOS offers two configurations for 3200MT/s: one with relaxed timings of 22-22-22-52 at 1.20V, and another with an XMP setting of 18-18-18-36 at 1.35V. Observe the significantly quicker 18-clock cycle settings at 1.35V compared to the slower 22-clock options for 1.20V. Less clock cycles mean faster memory performance. You might notice the PC runs a bit smoother with the 18-18-18-52 timing than with the 22-22-22-36 setting. If you have a copy of AIDA64, check the Memory Benchmark tests (Read, Write, Copy, Latency) at 2133MT/s and whatever overclock you reach. The difference isn’t huge, but it can improve usability for older systems.
If you prefer not to use the XMP 1.35V 3200MT/s option or the non-XMP 1.20V 3200MT/s setting that’s unstable, increasing the tCL(CAS) from 22 to 23 (or even 24) won’t affect power usage. Boosting the tCL clock cycles slightly slows the RAM but won’t harm performance and might resolve the 3200MT/s stability problem. The same applies at 2933MT/s.
I own several older A320M mATX boards that work well with XMP at 1.35V.
The additional 0.15V may slightly strain the RAM and CPU’s memory controller channels, though I doubt you’ll be concerned about system failure in 10 years (with XMP enabled) rather than 11 years (without). Other changes could occur in the meantime.
That’s great. It’s unnecessary to stress over minor speed gains from overclocking RAM beyond its maximum.