> Their sample size has dropped dramatically since the previous year, from 100k to 25k for laptops and from 186k to 48k for desktops. I suspect that's just an

> Their sample size has dropped dramatically since the previous year, from 100k to 25k for laptops and from 186k to 48k for desktops. I suspect that's just an effect of 2025 data being limited to just ~January, rather than a full 12 months. If people run a benchmark only once every 4 months on average, that would certainly explain the sample size.