The 7980XE OC isn't performing as expected.
The 7980XE OC isn't performing as expected.
My 9900K tends to fail there, this model acts oddly. Cinebench scores well at 4.7Ghz. I’m seeing 4651 in CB R15. Temperatures seem normal too. Other CPUs perform similarly, achieving about 50 FPS in Physics, while mine drops to 20% at 40 FPS—roughly 7980XE at stock.
Check this out: the 4.5Ghz model fetches 15000, while the 4.7Ghz version brings in about 12200.
R15 isn't a reliable base for stability; unstable configurations still work. When it crashes at R15, it usually tries to boot anyway. Try lowering it to 4.5GHz and see if the results improve. Mesh bus CPUs appear to rely heavily on memory, which might explain why they perform better under your setup.
I'm not checking for stability. I'm focusing on scoring. Something feels off with the numbers.
In Cinebench, high CPU temperatures reduce your score. On my dual Xeon rig with Grizzly Hydronaut thermal grease, after installing the new CPUs, I achieved a score of R20 on the first day. After four months of gaming use, temperatures improved slightly. When I repeated the test, I scored R20 again—still with unchanged settings. To verify, I ran Prime95 under heavy load to raise CPU temps significantly. The score dropped sharply, below 5800 points. This suggests my CPUs are overheating, which is why Cinebench scores decline.
Also keep in mind when comparing on hwbot, particularly regarding permitted adjustments. I don't know anything about CPU tweaks in the 3DMark series except for improved RAM speed, which could be beneficial. If the stated RAM is adequate, slower RAM might not be a major issue, but tighter timing could be necessary to perform well. Running RAM at lower speeds often yields better outcomes. I've noticed Kingston HyperX 4000 sticks sometimes exhibit this behavior (B-die). The XMP4000 profile tends to lag behind the XMP3600 in real-world tests. Watch for thermal throttling during testing, even if it means using monitoring tools, which can slightly lower scores. I don't remember the final verdict, but some users reported early "phantom throttling" when the product launched. Consider relaxing current limits and check for any throttling beyond just temperature. Cinebench R15 works well as a non-AVX benchmark.
No restrictions on throttling. Other systems with similar clocks on the TimeSpy leaderboard score around 15,000 in Physics, while my score is closer to 12,200. I plan to run stock and compare results to check for variations. Lighter OC might help identify differences.