Occasional issues with stable performance on a functioning system when XMP is activated
Occasional issues with stable performance on a functioning system when XMP is activated
I've experimented with disabling XMP for some time now, but I've grown accustomed to the slightly quicker startup speeds. Even a small delay feels like a long wait. Now I'm looking into the cause of this problem. System details: Windows 10 Pro 64-bit, MSI Z390-A PRO ATX (BIOS version E7B98IMS.180), I7-9700k processor with OC enabled, 36 cores, 47 threads, voltage at 1.008 and 1.256, RAM 32GB (4x8MB), Corsair Vengeance (16GB, 3200MHz, 2133MHz actual), AORUS Elite RTX 3060 GPU, storage setup including ADATA XPG SX8200 Pro 1TB M.2, SK Hynix Gold S31 500GB SSD, and a Seagate Barracuda 2TB HDD. Power supply is an EVGA GQ 850W semi-modular unit with 80+ Gold certification. This entry comes from a previous post where I was just trying to identify the source of instability, so you might find it useful if it contains insights not covered here. To clarify the situation: when XMP is activated, the system runs smoothly for several minutes after booting. However, under moderate load—like loading a game—it may abruptly restart, sometimes without any clear error message in Event Viewer. Most times it ends with a blank screen followed by the MSI logo, occasionally stuttering briefly as if about to crash, then returning to the logo. Very rarely does it trigger a BSOD (DRIVER_IRQL_NOT_LESS_OR_EQUAL). I still have an old dump file named sl6irk.dmp attached, though I'm uncertain if it's directly connected or just a coincidence from that time. I'm hoping this issue can be resolved with software fixes; finding a solution would save me a lot of stress if I manage to recover a few corrupted files. Thanks!
1. Restore your core voltage back to the default setting. The 9700K will work just fine at 47 volts across all cores without changing the voltage.
2. Update your NVIDIA drivers and reinstall them. This is what James Tillenger mentioned. Also, configure your RAM speed in BIOS to 3200, as he explained—this means using the intended speed, not necessarily overclocking.
I faced several odd problems with my Asus Tuf B550-PLUS paired with a Ryzen 5900X. Initially, it wouldn’t wake from sleep, then it would power down abruptly during games, followed by an unexpected no-video start that turned out to be a POST failure—no diagnostic indicator. After roughly four months, I partially disassembled the system, rechecked connections and RAM/GPU placement multiple times, but the DRAM debug light stayed amber. My MSI B450M didn’t show any similar warnings. Eventually, I learned that the GPU, storage, and one DIMM were moved to an ASRock B450M-HDV with a 3600X chipset. This setup had its own set of problems, though it was still functional. Installing the second DIMM led to a blank screen and no boot; it would only play audio during games afterward. I found out that Qualified Vendor Lists helped, and using RAM from the ASRock board’s QVL resolved the issue. Interestingly, both configurations used identical timings and voltages (16-18-18-36 @ 1.35V). The Asus board replied “could not duplicate issue.” After reassembling in a different case with RAM from the QVL, everything worked smoothly. It seems skipping the recommended QVL steps can cause trouble, particularly with Ryzen processors. By the way, the only person I’ve talked to who knows about those DIMMs was someone who bought them for an older Coffee Lake (8th-gen Intel) build a while back. This suggests that certain RAM types might cause instability depending on the processor brand or board used. Updated: September 12, 2022 by An0maly_76
search ram ics in thaiphoon burner, mixing different ram ics may lead to instability since each ic is unique. You can confirm the ics by uploading a thaiphoon burner image, then power on xmp and adjust voltages gradually—try raising vccsa and vccio by +0.1v, then increase dram volt by +0.05 to +0.1v, and slightly lower uncore frequency (100-300mhz). Not all changes need to happen at once.