- Apr 30, 2018
- 8
- 0
- 510
I recently built a system with a Ryzen 3 2200G. I used Kingston HyperX Fury DDR4 2666 (2x4).
I read that memory speed has a big impact on how well the integrated Vega graphics work, so I tried overclocking the RAM speed. I started at buy just changing the multiplier to get 2933 with no changes to voltage or timings. This was stable under Prime95.
I compared Unigine Superposition results (720 low preset) between 2666 and 2933, and 2933 resulted in a significant boost.
I went back into the BIOS and cranked up the multiplier to 3200, again with voltage and timings at auto. Booted to a USB stick with memtest86, and testing was good. Booted to Windows and ran Prime95; 2 hours large FFT and 2 hours blend and no problems. I'm thinking I'm the luckiest person in the world, because I expected it to fail. I run Unigine Superposition again, and the score went DOWN from the 2933 MHz result.
Two questions: 1) how is it even possible for it to be stable at 3200 MHz without making any changes to voltage or timings? 2) How's it possible for the Unigine benchmark score to go down with the supposedly higher RAM speed?
I read that memory speed has a big impact on how well the integrated Vega graphics work, so I tried overclocking the RAM speed. I started at buy just changing the multiplier to get 2933 with no changes to voltage or timings. This was stable under Prime95.
I compared Unigine Superposition results (720 low preset) between 2666 and 2933, and 2933 resulted in a significant boost.
I went back into the BIOS and cranked up the multiplier to 3200, again with voltage and timings at auto. Booted to a USB stick with memtest86, and testing was good. Booted to Windows and ran Prime95; 2 hours large FFT and 2 hours blend and no problems. I'm thinking I'm the luckiest person in the world, because I expected it to fail. I run Unigine Superposition again, and the score went DOWN from the 2933 MHz result.
Two questions: 1) how is it even possible for it to be stable at 3200 MHz without making any changes to voltage or timings? 2) How's it possible for the Unigine benchmark score to go down with the supposedly higher RAM speed?
Facebook
Twitter
Instagram