Discussing multicore Cinebench R23 results for the i9 9900k
Discussing multicore Cinebench R23 results for the i9 9900k
I'm using a specific RAM kit and have checked if XMP is activated. Make sure the modules fit the correct slots.
Cinebench requires relatively little memory, yet 2133 takes a long time—similar to what you see with DDR4. This could help account for the performance gap. When I first had my 5900X, I ran Cinebench at the default speed and then enabled XMP, seeing a difference of about 2000 points, roughly between 19000 and 21000.
@FranticSkunk You're correct. Your Cinebench R23 result seems much lower than expected. Did you keep track of the benchmark throughout? Which monitoring tool did you use? Check the Effective Clock data in HWiNFO—ideally it should remain consistent at full speed. For a fair comparison, I adjusted my settings using msconfig to cap my 10850K at 8 cores and 16 threads, matching your 9900K configuration. I also configured the BIOS to use DDR4-2133 memory, just like you do. During the test, I ran Cinebench R23 at a steady 4700 MHz. My score came in at 12849. It appears your CPU might be throttling under load, preventing it from hitting full performance. If the turbo power limits are set to their default, this is likely the cause. The standard 95W TDP would restrict the 9900K from reaching peak speeds. There could also be background processes affecting performance. To investigate further, I suggest using ThrottleStop during testing—it’s highly effective and won’t impact your scores. You can view your turbo limits via the TPL button in ThrottleStop; the Limit Reasons window will indicate any power constraints. When idle, the C0% should stay below 0.5%. If you’re unsure about ThrottleStop’s output, enable the Log File option before running Cinebench. After testing, close Cinebench and ThrottleStop to finalize logs, which should be saved in the ThrottleStop/Logs folder with today’s date. Feel free to share a log file for review. It’s time to restart and restore your memory speed to normal. I’ll run another test focusing solely on memory changes. Although R23 isn’t heavily memory-driven, it’s worth confirming. Edit: Switching from DDR-2133 to DDR-4000 only improved my score by 47 points. Your slow memory is probably not the main issue—something else might be at play.
Absolutely delighted with your dedication! It's impressive you're taking this seriously. After reviewing the logs, you might notice details I didn't catch. In Cinebench R23 I achieved a solid score of 10,000, though not quite 12,000. Throttlestop showed a steady 4.2GHz @ 60c. Someone mentioned it could be the motherboard limiting performance. My ASUS Prime Z390-A is decent, but I'm not sure its specs are the bottleneck. Thanks again for going the extra mile—your effort really makes a difference!
FranticSkunk your log file provides clear insights. During testing, the CPU runs at full capacity with a 47.00 multiplier as intended. Power draw rises to about 148W, then decreases. The far right section indicates throttling reasons. PL1 represents the long-term power cap; once it hits 95W, the CPU slows to stay within limits. This behavior is typical for Intel CPUs. The MULTI column shows a drop below 42, meaning the clock speed has fallen from 4700 MHz to under 4200 MHz. That explains the lower Cinebench scores. Running two tests consecutively reduces full-speed boost time and increases throttling at 4200 MHz or below. With adequate cooling, you can adjust PL1 and PL2 limits in BIOS or via ThrottleStop’s TPL button. I set both limits to their maximums (4095W) to avoid performance loss. The CPU only uses power it needs, so high limits won’t cause overheating. Use ThrottleStop to tweak limits while testing—monitor MHz and multiplier for optimal results. Keep the Limit Reasons window open during tests to track any throttling events. My Asus BIOS stores these settings under Ai Tweaker, with Long (PL1) and Short (PL2) Duration Package Power Limits.