It's feasible to experience reduced performance even with a stable GPU and CPU overclock.
It's feasible to experience reduced performance even with a stable GPU and CPU overclock.
I recently experimented with some overclocks for my Ryzen 2600. I managed to reach 3.7ghz using 1.075 volts with the stock cooler. After running benchmarks and street tests, everything performed very stably. Then I upgraded to a Gigabyte GTX 1650 Super Dual Fan GPU, achieving 7000mhz (up from 6000mhz base) and memory clock at 2080mhz (base core clock 1920 on the Gigabyte model). Under voltage it dropped to 1.0 mV from 1050 base, and all benchmarks remained stable.
I decided to further overclock both the GPU and CPU since I’m planning to start streaming Warzone. However, I’m only getting 50-70 fps, which is lower than expected. I know my PC isn’t high-end, but with these overclocks I should see better performance.
For streaming at 720p 60fps, I used Nvenc New, CBR, keyframe 2, 4000bitrate, preset max quality, high profile. Both visual tuning and GPU settings were checked. The GPU ran at 0, max BF 2, Lanczos, and the stream looked very good with nice performance. Still, I’m only achieving 50-65fps in Warzone.
Could my overclocking cause the worst performance? Should I just stick to stock settings—CPU at 3.4ghz 1.14v and GPU at 1920mhz with 6000mhz memory?
PD:
Motherboard B450m gaming plus (latest bios updated)
Teamgroup Tforce Delta RGB 16gb 2x8 3200mhz.
Seasonic S12III 650w bronze
250gb Pny SSD (programs)
1TB Seagate Barracuda for games (Warzone)
HP 1080p 23 inch 60Hz monitor.
When overclocking a Ryzen, it improves multi-core performance, but the entire system is configured to operate at a lower than stock vcore/vid etc. Since Ryzen are not Intel, their single core performance tends to need higher voltage for boosts, which can be a challenge. Running a game that heavily relies on single core will noticeably drop in performance and FPS, especially during Cinebench tests with single core settings—stock or PBO. You’ll observe higher multi but lower single scores. Adjusting this might involve removing the 200MHz offset and setting it to 0MHz in the advanced OC settings. Using Clock Tuner 2 (with reduced OC) and Dram Calculator (only for timing) on my 3700x has already helped improve things.
Other factors that might lead to reduced performance after overclocking include high temperatures during operation, especially with the CPU and GPU under stress. Insufficient power from the PSU or motherboard (particularly the VRM section) can also impact results. For some CPUs, achieving 3.7GHz without OC is normal, but boosting to 3.9GHz or higher on all cores—especially when boost is enabled—can improve single-core performance. Real overclocking typically reaches 3.9GHz to 4GHz or more, which you can experiment with. Voltage isn't a major concern; modern Ryzen processors handle around 1.4V well if properly cooled. Memory speed is another key consideration—ensure your RAM operates at 3200MHz with acceptable latency.
When boosting a Ryzen processor, it improves multi-core efficiency, yet the system defaults to lower stock values for vcore and video clock. Since Ryzen CPUs differ from Intel in architecture, single-core performance tends to lag, often needing increased voltage for optimal gains. Running games that rely heavily on single cores leads to reduced frame rates and lower FPS. In particular, Cinebench results drop significantly when using single-core settings at stock or PBO levels. Adjusting the 200MHz offset to 0MHz in advanced OC settings could help. Using Clock Tuner 2 with moderate OC and Dram Calculator for timing analysis on my 3700X improved multi from 3723 to 5000, raising single by nearly 100 points while cutting temperatures by about 20°C from stock—achieving a 41.8 multiplier. A full overclock to 4.4GHz still yields lower single compared to stock, but boosts multi to 5101 and maintains temperatures nearer to original specs at 84°C. I’m comfortable with a more powerful setup that runs cooler, delivering better single and overall performance. Ryzens perform optimally when allowed to function naturally rather than being constrained to fixed settings.
You also need to recognize that increasing the CPU speed from its base of 3.4 GHz to 3.7 GHz offers only about a 10% boost in clock rate. Likewise, the GPU's clock speed increase is also around 10%. This doesn’t instantly mean better performance; it might simply result in gains close to nothing beyond certain benchmarks. To see this clearly, if you were achieving 70 frames per second before, you’d likely see only a slight improvement—perhaps 75 FPS at most. However, clock speed increases don’t automatically translate to proportional performance gains, and they may not significantly impact results outside of specific tests. Also, GPUs can sometimes handle memory overclocking, but beyond a certain point, errors begin to build up. It seems modern GPUs are becoming more cautious with memory errors, which can actually reduce performance rather than enhance it.
I couldn't reach above 3.7ghz. During the test, when CPU temperatures hit 90 degrees, they reached 3.8 cpu temps. My RAM is also operating at 3200mhz with the XMP profile active and meeting its suggested latency.
It's wiser to keep Ryzen at its current price until you find a better aftermarket cooler and consider a 4+GHz overclock.
Good cooler yes, that's always a plus, it doesn't have to be massive, just efficient, you're not forcing a lot of power with a Ryzen, so a mid-range cooler is more than enough, similar to an Arctic esports Duo.
When it comes to overclocking, that's all about personal preference. For me, my Ryzen isn't really pushed to its limits in the usual way, though I could do it. It still functions just like the original, it downclocks when not in use and improves performance through optimization. It's like a well-trained athlete who can walk to the mailbox or sprint at full speed as needed—never has to run at maximum all the time.
I scored around 5000 in Cinebench R20. I just watched a video from Hardware Unboxed comparing it with a 11600k; his best OC and CB20 score for his 3700x was 4887. My Ryzen, which behaves like a Ryzen and doesn't rely on locked core overclocking, outperformed him by more than 100 points, and his system stayed cooler, around 80°C or less.
OC is optional if you're interested in it, but since it's a Ryzen, there are alternatives that are just as effective—just not necessarily better overall.
With a Ryzen paired with a solid cooler, you'll likely reach maximum boost frequencies on multiple cores, though only full-core usage will really benefit, provided you manage temperatures and vcore effectively because boosts depend on both.