Ensuring consistent performance during voltage adjustments
Ensuring consistent performance during voltage adjustments
I apologize for the previous post, but no one addressed my initial problem within 16 hours and it persists. The mods may remove it if you wish. My PC details: https://pcpartpicker.com/user/ArushM/saved/#view=XTXfpg. I was trying to overclock and suddenly crashed after watching a YouTube video—no video loaded, just a complete crash. It happened during normal tasks too, except under stress tests. Overclock settings: i5-13600K @ 5.5/4.5 GHz, default voltage (1.25 idle, 1.23 under load), LLC mode 4 (MSI), no AVX2 ratio loss, max ~250W but it throttles severely. I used Intel UHD 770 iGPU @ 21GHz (1.15V, static), ring ratio dropped from 4.7 to 4.5 (stock) and still crashes. Kingston 5200 CL40 @ 5800 CL36 at stock voltage. GPU was Linpack Extreme for ~24 hours, Memtest86+ for 13 hours (20 passes). Tested Fortnite for 2 hours. Cause? Any explanation or solution? Also, tips to check ring ratio and GPU safely next time?
It's only been 16 hours—don’t be so quick, or else hire someone to handle it. It’s just people trying to get there, not a support team.
Running a locked speed on the CPU means you don’t have to worry about power limits. It seems like memory issues are present, especially at 5200MHz—it might not be a strong setup. For better results, aim for a 6400+ MHz processor; then you can push it to the maximum of your motherboard or board (usually 7200-7600 if you're skilled and have a decent Z790). Check your memory speed, preferably using DDR5. If it's unstable, performance will drop. Use the AIDA64 cache and memory testing tools. Achieving 55ns or less is great, and below 52-53 suggests limited scaling for games. Your current setup may not reach that level.
Memtest86+ stability might still be adequate, but newer versions could offer improvements. The original version was considered reliable, especially for older DDR5 hardware, though it wasn’t the absolute top choice. Your reported 71 ns latency is reasonable for a basic tool, but newer iterations often provide better performance and reliability.
This setup would really pay off, especially if you're dealing with slow memory. A 6400 kit with OC would likely boost performance by at least 10%. Plus, it should be much simpler to find tutorials and guides for A-die.
They are the same Kingston 5200 CL40 sticks, upgraded to 5800 CL36 and tested with memtest. No BSOD occurs—just the system freezes before restarting. Since you're a minor and the PC was a gift, no GPU was installed. What additional tests are available? You recently swapped the RAM to 5600 CL36.
I work with testmem5, but as I mentioned you can roughly assess stability by the RAM speed. If it feels erratic, it will likely run much slower. Using the aida64 test can also show cache performance, helping you evaluate ring frequency consistency and speed. Keep boosting it until you notice a drop, then you'll understand your maximum capacity. The system's voltage is crucial for the ring, and memory matters too—if you can upgrade, aim for 1.15 to 1.2 volts to boost core speed.
The computer stopped working and started using a different speed setting. The test page isn’t working right, and the download button isn’t doing anything. To check how fast it’s loading, you can use AIDA64 like a simple helper to see if the files are ready. Just remember, the voltage changed from normal to 1.2V, and now it’s acting differently.