I'll change to 1ms caching time to reduce the risk of drift. I'll do it when the benchmark is complete.

When we're calibrating the hardware clock frequency, we measure it in ticks per microsecond. The calibration is quite exact, but the value is rounded to a integer in ticks-per-microsecond. The worst case error due to this rounding is 0.5 ticks per microsecond. Commonly, a TSC clock has a frequency of ~2500 ticks per us and then we get an error of at most 1/5000. If we're caching gettimeofday for 100ms, we get a maximum error of 20us. With 1ms, we get a maximum error of 0.2us (rounded to 0).

Given ustime() is called every microsecond, we can still save a lot by calling gettimeofday() only once in a thousand.

Benchmark ran on this commit: 2ff52b9

Benchmark Comparison: unstable vs c0e16ef (averaged) - rps metrics

Run Summary:

• unstable: 80 total runs, 16 configurations (avg 5.00 runs per config)
• c0e16ef: 80 total runs, 16 configurations (avg 5.00 runs per config)

Statistical Notes:

• CI99%: 99% Confidence Interval - range where the true population mean is likely to fall
• PI99%: 99% Prediction Interval - range where a single future observation is likely to fall
• CV: Coefficient of Variation - relative variability (σ/μ × 100%)

Note: Values with (n=X, σ=Y, CV=Z%, CI99%=±W%, PI99%=±V%) indicate averages from X runs with standard deviation Y, coefficient of variation Z%, 99% confidence interval margin of error ±W% of the mean, and 99% prediction interval margin of error ±V% of the mean. CI bounds [A, B] and PI bounds [C, D] show the actual interval ranges.

Configuration:

• architecture: aarch64
• benchmark_mode: duration
• clients: 1600
• cluster_mode: False
• data_size: 16
• duration: 180
• tls: False
• valkey_benchmark_threads: 90
• warmup: 30

Configuration:

• architecture: aarch64
• benchmark_mode: duration
• clients: 1600
• cluster_mode: False
• data_size: 96
• duration: 180
• tls: False
• valkey_benchmark_threads: 90
• warmup: 30

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 74.98%. Comparing base (c2aca72) to head (24b2d14).
⚠️ Report is 8 commits behind head on unstable.

@@             Coverage Diff              @@
##           unstable    #3193      +/-   ##
============================================
+ Coverage     74.87%   74.98%   +0.10%     
============================================
  Files           129      129              
  Lines         71328    71337       +9     
============================================
+ Hits          53407    53490      +83     
+ Misses        17921    17847      -74