I'm seeing huge spikes in the frame generation curve during Gears 5. Should I adjust sampling?

Here is the deal: when the Super Alloy core clocks are bouncing between 2450MHz and 2680MHz, a low sampling rate creates fake jitters. I tried bumping the refresh rate to 1ms using RivaTuner, but the curve was still a mess. I then used GPU-Z for real-time tracking and saw frame times jumping between 12-18ms, which explains the tearing. I finally used a frame limiter to lock the game at 60 FPS, and the generation curve finally flattened out into a straight line. After verifying with Unigine, the data accuracy improved massively. If you have high frequency swings, you absolutely must compress the sampling period to under 1ms, or you are just guessing at the actual stutter.