Errors in memory during memtest86 execution
Errors in memory during memtest86 execution
This weekend I set up a new system (PCPartPicker list). BIOS is now the latest version, along with all drivers and Windows 11. CPU and GPU stress tests passed (Cinebench 23 and Heaven). Passmark memtest86 reported 6 errors on test 8. A report from MemTest86-Report-20231203-181232.html showed no issues after reseating RAM, disabling XMP, and retesting. Running MemTest86 again with XMP enabled produced 4 errors on test 8: MemTest86-Report-20231204-063729.html. In my view, errors in MemTest86 shouldn't occur at all. Up to now I've only experienced Civ VI crashing twice, but I want that not happen during work. I'm unsure what to do. Since the tests pass without XMP, it seems the problem lies with XMP settings. Changing those might help or I could just replace the memory sticks and buy new ones. Edit: PCPartPicker Part List CPU: Intel Core i7-14700K 3.4 GHz 20-Core Processor ($401.60 @ B&H) CPU Cooler: Alphacool Eisbaer Pro Aurora 69.98 CFM Liquid CPU Cooler (Purchased For €185.00) Motherboard: MSI MEG Z690 UNIFY ATX LGA1700 Motherboard (Purchased For €185.00) Memory: G.Skill Trident Z5 RGB 32 GB (2 x 16 GB) DDR5-6400 CL32 Memory (Purchased For €130.00) Storage: Samsung 990 Pro 2 TB M.2-2280 PCIe 4.0 X4 NVME SSD (Purchased For €135.00) Video Card: Asus ProArt OC GeForce RTX 4070 12 GB Video Card ($679.00 @ B&H) Case: Lian Li LANCOOL II-X ATX Mid Tower Case (Purchased For €0.00) Power Supply: SeaSonic FOCUS GX-750 ATX 3.0 750 W 80+ Gold Certified Fully Modular ATX Power Supply ($144.99 @ Amazon) Operating System: Microsoft Windows 11 Home OEM - DVD 64-bit ($124.99 @ Amazon) Total: $1350.58 Prices include shipping, taxes, and discounts when available Generated by PCPartPicker 2023-12-04 03:46 EST-0500
The details are kept confidential. Without understanding your CPU, motherboard, or RAM configuration, it's hard to pinpoint the exact problem. It might be an overly fast XMP profile (like DDR5 8400) or a faulty memory module—testing one stick at a time could help. Providing this info could guide you toward a more effective memory stress test, since tools like MemTest86 aren't the most sensitive for detecting errors; you might achieve faster results, such as identifying issues within minutes instead of hours.
It's typical what you see. The setup isn't overly extreme, so it should function on most 14th generation processors. Possible causes include faulty memory card, defective motherboard, incorrect CPU placement, or outdated BIOS version. Most likely, the CPU mount is the issue since fixing that is simpler. First, test each stick separately, focusing on slots 2 and 4 for stability under XMP settings. If one stick keeps failing consistently, it points to a faulty stick requiring return merchandise authorization. If a specific slot repeatedly fails, it suggests a motherboard or CPU mount problem—consider reinserting the CPU. For both sticks and slots, a BIOS update might be necessary. With 14th gen chips, stress tests like Y-Cruncher VT3 and TestMem5 with the 1usmus_v3 profile are useful. VT3 evaluates memory controller performance first; if it detects problems within minutes, it can catch frequency-related glitches. If the problem seems tied to a particular stick, try running it at JEDEC speeds for a month or wait for MSI to release an improved BIOS. Note that TestMem5 1usmus_v3 usually identifies memory issues in 5-10 minutes and can spot high-frequency faults faster than standard diagnostics.
A contact frame might enhance the CPU connections. Adjusting the speed or timing could help. Edited December 4, 2023 by leclod
You're checking if the thermalright contact frame affects XMP settings. The same configuration should work, right? You tested both sticks in slots 2 and 4 with XMP enabled, passed memtest86, but got errors during reverse placement. The VDDDQ is 1.400V and VPP is 1.800V—would raising the voltage help? Also, could reverting to an older BIOS fix the issues?
I attempted to raise the DRAM voltage to 1.410V. It appears successful as both MemTest86 test 8 and TestMem5 ran without errors using the 1usmus_v3 preset. I plan additional testing later when time allows, though memory sticks likely aren't faulty.
It seems unusual. These components can lead to memory problems in certain setups and specific boards. I’m using one on my Unify-X without any issues, so it’s unlikely to be the cause. For troubleshooting, I might revert it back to stock and see how it performs. The voltages I was talking about refer more to IMC levels rather than DRAM values. You can raise those (1.55V is acceptable for long-term use if SK Hynix is correct), but anything above 1.43V should include active cooling for G.Skill kits because their heat spreaders are not very effective. With IMC voltages, there’s no concern, and these levels actually contribute to stability in DDR5 configurations. The top three are CPU VDDQ, CPU VDD2, and System Agent (probably called SA voltage in BIOS—check the manual). For VDDQ, aim to keep below 1.45V, VDD2 around or below 1.65V, and System Agent at or below 1.45V. These targets are generally optimal, so reaching them shouldn’t be too high. VDDQ is typically set near 1.35V, though newer BIOS versions suggest lowering it can help (1.25–1.3V worked well on my system). VDD2 usually sits between 1.4 and 1.5V, but at 6400 MHz you probably don’t need to push it to 1.5V. System Agent is heavily dependent on the CPU; I wouldn’t doubt a 1.2V setting works at this speed. Raising the memory voltage might assist if timing is the issue, though that’s unlikely. If you try it, feel free to share your results. In theory yes, but older BIOS versions don’t support the 14700K, so you’re limited to the latest updates.
I wasn't sure at first, but if each stick works alone yet causes an issue when combined, it’s worth checking another setup. Cooling in the lancool II isn’t ideal, so moving things to a different case might help. I should have understood better then. You seem to have found the voltages they were talking about.