Ensuring consistent performance during voltage adjustments
Ensuring consistent performance during voltage adjustments
Baseline set to stock with a default of 4.5 for ring. Uncore and ring speed limits matched if available. Memory baseline uses stock XMP or a stable known OC. 71ns is considered slow but acceptable for your memory performance. Unknown die type; BIOS typically provides details. After downloading aida64, open tools → Cache and memory benchmark → start test. Avoid background processes as they may affect accuracy. You can also check memory latency separately via the left menu "Benchmark". TestMem5 v0.12 (best configurations).zip is included. Also, move the "Extreme1" profile to TM5.exe for a more precise evaluation.
take a look at the burner screenshots for both sticks to figure out the issue. Overclocking blind seems risky, especially since others have already done it and you can reference their setups. It’s not great quality, but it’s better than waiting three weeks to test which sticks work best in a trio. Some sticks clash badly, so patience is key. The goal is stability, not just speed.
There are some well-tested Micron 16Gb A chips that start up and complete certain benchmarks at 5800MHz, which means it's not out of the question. It's unlikely, but still possible. The manufacturer is Kingston, and their products typically fall into Micron or Hynix categories, so it's more probable you're dealing with a high-quality Micron unit than a subpar Hynix or Samsung one. Even the lowest Samsung kits reach speeds around 6000MT/s, so this doesn't seem feasible.
MemTest86 isn't particularly effective for stress testing, especially with modern Intel processors. Many kits I've seen pass MemTest86 but fail in other tools like TM5, Karhu, HCI Memtest, or Y-Cruncher VT3. A better approach would be to try TM5 and TM5c instead. This method is preferred not just for its error detection, but because it uses a USB interface, letting you assess the entire memory array instead of losing data in a small area like during a Windows stress test.
One benefit is that USB-based testing reduces the risk of data corruption from faulty drives, unlike Windows stress tests where a bad stick can cause issues. While this isn't perfect, it's a safer bet compared to in-Windows evaluations.
In older DDR/DDR2 eras, timing or frequency problems during manual overclocking were common, but with DDR5, such issues are rare. You'll likely need to run specific Windows tests if you want to catch problems.
For setting up TM5, refer to the Overclock.net guide at https://www.overclock.net/threads/memory...s.1751608/. Download the file from Mega, extract it, and open TM5.exe. Click "Load Config and Exit." Navigate to the config folder, choose a file like 1usmus_v3.cfg for DDR5 testing. Adjust the "Cycles" setting to at least 30 or 3 cycles with Extreme1 for better stability.
After configuring, edit the file in Notepad to modify the "Cycles" value. If needed, extend testing time by editing the file directly. When launching TM5 as an administrator, restart your system if errors occur.
I encountered a crash while trying to run 5600 CL40, now only 4800 CL40 is working without issues. The results with 5600 CL40 and a 4.5GHZ uncore, then 4800 CL40 at 4.5GHZ, are shown below. It passed Y cruncher too, but when I adjusted the clocks significantly it still crashed. I encountered a critical debug error during configuration selection and the program exited. Sorry for the delayed reply—I was with my dad for lunch (Happy Fathers Day to all!) and spent some time with him. IMG_0819.HEIC IMG_0820.HEIC
Launch the program, press the read button, then choose one of the available memory chips. Keep in mind that Thaiphoon Burner may struggle with DDR5, so it’s possible it won’t work. You might encounter three potential reasons: the memory is placed incorrectly (though booting at 5800MT/s makes that unlikely), unstable memory may have corrupted the Windows install, or a component could be faulty (likely factory damage). Verify the RAM is in slots 2 and 4. If you’re comfortable, perform a Windows reinstall, or clear the CMOS if needed. Also, check whether you performed any overclocking by manually setting everything back to stock or clearing CMOS. If it was done that way, it’s easy to overlook changes and still have the overclock active. Use the CMOS jumper to reset, as the BIOS “load defaults” feature isn’t always perfect.
It worked I mean I just didnt load XTU, should be good enough right? As of right now though seems to be ok but also it did crash once with no OC and just RAM OC, I mean wait and see IDK. Im not the expert here Kingston KF552C40-16.txt Id rather not but I could
Sometimes it's better to remove XTU and reset the CMOS just to be safe. I personally prefer using the BIOS for everything, but I understand your preference. You mentioned you have a 16GB Samsung chip, which is unusual for these Kingston boards. It should handle 6000 CL36 without trouble, so it's possible you were facing problems. Check if the system crashes and let me know for confirmation.