These clock speeds appear to be quite low.
These clock speeds appear to be quite low.
I recently purchased a 5900X paired with an x570 motherboard. I’m noticing the processor isn’t delivering the expected performance. The clock speeds look lower than what others report. I’ve shared images from three different setups, along with a log from hwinfo. During gaming, I rarely see it exceed 4.0 GHz; when it does, it jumps to about 4.5 GHz and then drops back quickly. Since I’m playing CPU-heavy games right now, it seems there’s still a lot of potential left. I attempted PBO and auto overclocking, but neither improved the Cinebench scores or clock speeds. Regarding the scores, mine is around 20300 (±150). The 5900X model is installed here: https://www.gigabyte.com/Motherboard/X57...-rev-10#kf. It’s equipped with a Kraken X63 AIO cooler, and the pump speed is at full capacity—even with fans running near maximum. Is there something I’m doing wrong, or did I miss out on the silicon chance? The Cinebench results for all cores and single core are attached.
For maximum high frequency, enter the BIOS and configure a fixed 4.8GHz across all cores, though you'll likely require about 1.3 to 1.35V.
Try adjusting the voltage using the curve optimizer. Which board do you have? Provide the RAM details—brand, model, clock speed, and timing specs. You're not hitting full performance. I see 4.6 per core, 4.95 for single core with a -30 step curve offset, and standard PBO settings.
Ram shared his experience with a Z790 chipset running at 3200MHz CL14. Mobo linked to a review page. He mentioned struggling with thermal throttling during Cinebench, only hitting around 4.45 GHz per core. He feels he missed out on the performance gains others are seeing. He also asked about measuring single-core performance, noting that his Cinebench tests show results across multiple cores but peak speeds stay below 4 GHz.
Mine will reach 5150 on air without any issues, though I understand your frustration. I’m limited to a maximum of 1900FCLK with mine without running into errors. My 5600X stays stable at 2K and maintains a 2100:1 scaling factor. It’s not quite as smooth as Intel’s performance. How did you set up your TIM? That matters a lot, which is why I’m asking. Things like to run cooler—the faster they cool, the quicker they go. I’m someone who loves pushing every clock speed possible; because air cooling gives me a bit of an edge, I tend to be a bit unconventional. I also avoid using windows or software for overclocking, so you’d need to adjust that in your program since it isn’t installed.
a cross spanning from side to side with a small dot between the lines appeared. Yesterday I picked up a Dark Rock Pro 4 just to check if my AI was acting strange. No major changes, only a slight increase in temperature. After a gaming session, I checked hwinfo while monitoring its performance. The peak reached 4.745 GHz, but during most of the time it stayed near 4 GHz on the top cores. All other cores barely moved. Does your CPU really maintain those 4.9 GHz+ speeds while playing?
Confirm your offset setting? You can adjust it below or above using the curve. Undervolting and thermal throttling are practically unavoidable. It's typical for cores to reach their limits around a specific point, with some exceeding and others falling short. This is normal behavior. On Cinebench single-core tests, you'll mainly observe the highest boost frequencies from one core, which isn't realistic in real-world scenarios. A 5900X should maintain stable boost clocks near 4.4 to 4.6 under sustained load.
You've restated my idea exactly. The main focus was that single-core performance spikes are rare outside synthetic tests, as most systems use multiple cores and other tasks are running simultaneously. What I meant by "top" was the highest achievable boost frequency—around 5GHz—not just peak overclocking.