Question PC experiencing failure during stress tests with XMP despite having good stock performance
Question PC experiencing failure during stress tests with XMP despite having good stock performance
Hello everyone, I recently assembled my first PC after eight years and have been attempting to enable XMP functionality, but it consistently fails during stress tests. With the default configuration (no XMP), the system has successfully completed an OCCT test, Memtest86, and Heaven benchmark, and appears to operate normally under these conditions. However, when I activate XMP or manually adjust the timings, errors appear in both OCCT and Memtest86. The RAM is already listed on the motherboard’s QVL, and I’ve reinstalled the BIOS, updated the CMOS, refreshed drivers, and reinstalled Windows 11—yet it didn’t resolve the issue.
Error messages indicate CPU issues detected on a specific core (see attached images - https://imgur.com/a/hFMLwJ8) and viewing details here: https://imgur.com/a/hFMLwJ8. Running without XMP or memory overclock settings isn’t an option, as I’m determined to get it working with the price I paid. I have no idea what additional tests I should perform since I don’t have another RAM stick, CPU, or motherboard, and because the stock build passed all tests, the retailer seems reluctant to accept a single part return and will likely ask for a full system diagnostic. It would also be unhelpful if I return a part and the same problem recurs or if the new unit fails again.
Are there any other recommendations you could try before spending money and forcing the store to run a comprehensive check? Or is there a component you think might be the culprit?
Additionally, I’ve observed some concerning issues: during boot, the RAM’s RGB lights flicker (almost like a short circuit) but then return to normal. It’s unclear if this is typical for my RAM and just part of the initialization process. Occasionally, the PC freezes into a black screen upon restarting, requiring me to manually press the restart button.
Memory Uncore Frequency changes from 5000Mhz to 800Mhz after a failed test. The screen sometimes glitches during OCCT with XMP enabled. I’m not sure if this is related to the GPU or ultimately points to the RAM.
Parts list:
- CPU: Intel Core i7 14700KF
- Motherboard: Gigabyte Z790 AORUS ELITE AX LGA 1700 ATX v1.1
- Memory: Thermaltake TOUGHRAM XG RGB D5 32GB (2x16GB) DDR5 7200MT/s C36
- PSU: Corsair RMx SHIFT Series RM850x
- GPU: 1660 Super (anticipating it’s outdated, awaiting the 4070 Super Ti)
- SSD: Kingston FURY Renegade 2TB M.2 NVMe PCIe Gen4
- Case: Fractal Design Meshify 2 Compact RGB
- Cooler: DeepCool LT720 360mm
- OS: Microsoft Windows 11 Home 64-Bit
Your insights are greatly appreciated.
Did you attempt the newest BIOS version FHe?
Verify the voltage should be around 1.45V when XMP is active.
Update: only 16GB of RAM appears in the list from the motherboard, not 32GB.
Also, it's a SK Hynix M-Die and not an A chipset.
Replace the RAM.
Are these installed in the A2 and B2 slots?
Inspect the CPU socket pins on the motherboard for any bending or discoloration.
Is the CPU operating on standard clocks?
For reference, run userbenchmark.com with XMP and share the link, for example: https://www.userbenchmark.com/UserRun/28977730
Is the ram a single matched 32gb kit bought in a single package?
Or, is it two identical 16gb part numbers?
If the latter, there is no guarantee that it will operate at advertised speeds.
If the ram fails the xmp memtest then I would assume that the ram is defective and you should contact the ram seller for a RMA.
If you want to fix this yourself, enter the xmp specs yourself in the bios and gradually add voltage past the xmp default.
Although rare, the CPU might have a weak memory controller. This situation falls within the "silicon lottery" when purchasing CPUs. Are the RAM sticks passing tests at any rate faster than the JDEC standard? Consider enabling XMP and then reducing the speed to 6000 mt/s for testing.
I just tried using XMP defaults at 6800 Mhz and it worked stably in OCCT for at least five minutes. At 7000+ MHz it caused errors within a few minutes. I also encountered 'breakpoint' errors when attempting to power off the PC after testing, which suggests a possible RAM issue or memory failure.
https://imgur.com/a/aDiEsu4
View: https://imgur.com/a/aDiEsu4
I read about IMC but doubted it was the case. The 14700KF at 7200Mhz seems modest compared to better boards that reach 8000Mhz or higher. So, I’m considering the motherboard or RAM first, but I can’t be sure until I test with different components.
Yeap am on the latest FhE bios.
Have reseated the RAM a few times to make sure and they're defintley in A2, B2.
I did check the CPU once and didn't see any bent pins on the mobo or chip. Didn't check for discoloration but i'll check again as a last resort. Voltage shows it's actually pushing to 1.455 V as well (jumping to 1.470V MAX).
The memory module P/N is
RG33D516GX2-7200C36B
which on the QVL shows it's actually Hynix A... Which makes it worse that it's not pushing 7200mhz then right? Not sure what you meant by the 16gb as that should be per stick. This is the exact product page =
Product Page
CPU is stock as well unless the MOBO software has some sort of weird auto clocking feature i'm unaware of.
It's a 32gb kit, and the product page is available. Should it function properly? I ran some tests and found issues. At 7000 (compared to the advertised 7200 max) operating at 1.45V, it's causing errors. It looks like raising the voltage would be necessary to maintain stability at this speed.
I believe the RAM package isn't performing well right now. If you find a comparable speed set still causes issues, it's likely the motherboard or the CPU memory controller for your particular CPU is underperforming.