![best benchmark to test gpu stability undervolt best benchmark to test gpu stability undervolt](http://www.geeks3d.com/public/jegx/200912/furmak_msi_afterburner_gtx295_02_400.jpg)
![best benchmark to test gpu stability undervolt best benchmark to test gpu stability undervolt](https://ambermd.org/gpus14/images/dhfr_nve_4fs_amber14.0.0_cumulative.png)
When I do get around to it, I'll probably run this again with the Auto OC option and see what results I get then.įor now, an extra 6-7% performance for very little effort seems pretty good, and my CPU idle temps are down around 5-10 degrees C.
![best benchmark to test gpu stability undervolt best benchmark to test gpu stability undervolt](https://i.ytimg.com/vi/t5opmpS7Pds/maxresdefault.jpg)
While I do need to upgrade my U9S CPU Cooler to a 240mm AIO at some point, I'm snowed under with work at the moment, and I can't do the installation, and all the tweaking I would want to right now. I'll be playing some games later today, and running longer duration stress tests to check the stability further. In addition to the Cinebench runs, I have also run the AIDA64 Stress test with CPU, FPU, and cache options selected for over 30 minutes and everything seems stable. Undervolted settings increased to 14880 with an average core clock of 4328MHz. Stock settings resulted in a score of 13945 with an average core clock of 4066MHz. The stock and undervolted results below are both averages of two runs. I reset the HWiNFO Stats 1 minute into the benchmark, and then took a screenshot 9 minutes later (with Cinebench still running - My 10 minute test takes about 10 min 30 seconds to complete the 12th pass before it ends). I used Cinebench R23 - 10 minute Multi Core and HWiNFO to test before and after. It yielded the same -30 setting for each core. I was a little concerned about the fairly short test duration, so I increased the test duration to 240 seconds, and ran the Per Core optimization again over night. In my testing both "All Cores" and "Per Core" yielded the same -30 setting for each core. The Per Core option took a little over an hour using the same 30 second test duration, but tests each core. The default test duration is 30 seconds, but you can change this from the Settings button at the bottom. I did need to reboot afterwards before finally applying the optimized profile. The process took around 43 minutes, and although it warned that reboots would occur, I didn't have any during the process. I then selected Control Mode as Default and Curve Optimizer Mode as All Cores and clicked "Start Optimizing" I opened Ryzen Master from the desktop shortcut and agreed to the warning. I don't remember if I was required to reboot after the installation, but I did anyway. Installation is a simple next, next, finish affair. You can download Ryzen Master from here: AMD Ryzen™ Master Utility for Overclocking Control | AMD The process for the undervolting isn't exactly one click, but it's pretty close. The U9S simply doesn't provide enough cooling for the Ryzen 7 5800X, and while the undervolt doesn't change that, it does improve things.
![best benchmark to test gpu stability undervolt best benchmark to test gpu stability undervolt](https://cdn.arstechnica.net/wp-content/uploads/2013/03/gw2-99th.gif)
This has given me the best results after trying many configurations with both single and dual fans (attached to the U9S) set to both intake and exhaust. I have the U9S configured with a single fan as intake from the rear of the case with a Phanteks T30 120mm mounted on the side bracket opposite the PSU as exhaust. MSI MPG B550I Gaming Edge WiFi (BIOS 1.9/AGESA 1.2.0.6c) After reading that there was a new release of Ryzen Master this week, which allows for automatic undervolting from within Windows, I decided to give it a try yesterday.