Benchmarks are just numbers until they've run a winter

I saw the Muse Spark 1.1 results. Means nothing to me until I see how it handles the real grind, not a lab test.