Performance stays at 4.7GHz despite 9900K settings
Performance stays at 4.7GHz despite 9900K settings
The BIOS options let you pick the CPU multiplier that matches the number of active cores. The "all core" setting applies a single turbo multiplier across all cores. Adjusting per-core speeds requires Turbo Boost 3.0 support, which the 9900K lacks. Your Windows preferences manage this behavior. When using the High Performance profile and setting both Minimum and Maximum processor states to 100%, the CPU runs at full speed even when idle. To slow it down during inactivity, modify your Windows settings. Consider switching to the Balanced power profile. Enabling XMP could have altered the turbo multipliers; check HWiNFO for current Turbo Ratio Limits. If XMP is active, you may need to manually adjust the BIOS turbo multipliers. If previous settings worked but now fail, revert them. Turn off XMP, restart, and verify with HWiNFO to confirm the 50 multiplier is applied. After rebooting, return to BIOS, turn on XMP, and retest. The default turbo multipliers for the 9900K are 50, 50, 50, 48, 48, 47, 47, 47 when one, two, three, four, five, six, or seven cores are running.
That's why they hit 4.7. Turning off high performance mode made the speeds fluctuate again. It sometimes spiked to 4.8 briefly and reached 4.9 once, but never hit 5GHz. In reality, the boost is around 4.7 at normal settings. For most people, that's what you get for about $325 on a $3700 build for an extra 300MHz clock. Great job on the upgrade!
Many tracking tools don’t match Intel’s suggested approach. Some well-known applications struggle to accurately display Intel Turbo Boost, particularly when the processor quickly switches tasks between cores while cores move in and out of various power states. To get a true picture of your Intel CPU’s performance, consider using ThrottleStop. Its multiplier tracking is exceptional. Version 8.70.6 is available at techpowerup.com/download/techpowerup-throttlestop/. It reveals the Turbo Ratio Limits you’ve set and lets you compare them to official specs. Running the built-in TS Bench with a single thread setting can clarify whether your CPU is operating at its maximum multiplier. Disable other monitoring programs during testing, as they may keep multiple cores busy and disrupt full turbo operation. For instance, a 4th Gen laptop’s 4700MQ can leverage a 36 multiplier when only one core is active. ThrottleStop clearly shows the highest multiplier usage during a single-threaded session. If you find testing tedious at default settings, adjust all turbo multipliers to 50—AMD may still deliver solid performance at that speed.
I still feel regret over choosing the 9900k, but there were solid reasons for my decision compared to AMD. AMD hasn’t fully addressed memory latency issues with Ryzen yet, which affects performance and makes it a slight disadvantage against Intel in this generation. The 3700x-3900x models also lag behind Intel in IPC. I’ve noticed that even at 4GHz, Intel consistently outperforms them, not just due to clock speed. My buyer’s remorse stems from overspending on a more expensive motherboard than I had initially planned, after committing to Intel and buying the 9900k. Lower-end mid-range boards were limited by their flaws, so I ended up opting for a pricier high-end option. On the other hand, Ryzen offers some solid lower-mid range choices that would have worked well with a 3700x. Looking back, it wasn’t worth the investment. Lesson learned. The second point is that I might have received a subpar batch of silicon for my 9900k. Apart from boost speeds, benchmarks like Cinebench and TimesPy show much lower performance compared to what I expect, even with fast memory, good cooling, and a quality board.