--- title: "The handheld got faster by dropping Windows." date: 2026-07-15 url: https://remiam.co.uk/notes/handheld-faster-by-dropping-windows tags: [Hardware, Gaming, Platforms] read_time_minutes: 5 description: "Why the Legion Go 2 running up to 75 percent faster on SteamOS than Windows 11, on identical hardware, is the clearest public argument for building a system around its one job." --- # The handheld got faster by dropping Windows. *Published 2026-07-15 · 5 min read · by Liam (Remiam)* The Legion Go 2 shipped a SteamOS version in June, and independent testing has the same handheld running up to 75 percent faster on SteamOS than on Windows 11. The hardware did not change. The operating system did. Lenovo shipped a SteamOS version of the Legion Go 2 in June, at 1,199 dollars (TechTimes, 2026), part of a wider move that has put Valve's handheld Linux onto machines it never built. Most of the coverage framed it as a Steam Deck rival. The number worth sitting with is a different one. On the smaller Legion Go S, independent testing had the same handheld running up to 75 percent faster in Horizon Zero Dawn Remastered on SteamOS than on Windows 11, and it never lost a test across the run (XDA, 2026). Same silicon, same battery, same screen. A different operating system. ## The gap is not magic, it is focus. Seventy five percent is the headline case, not the average. In a heavier game the lead narrows to about 11 percent, which is the difference between an uneven 54 and a steady 60 (XDA, 2026). But the direction never reverses, and that is the interesting part. SteamOS wins because it is doing less. It is not running a desktop shell, a search indexer, an update service and a dozen background processes that assume a plugged-in tower. It caches shaders ahead of time, keeps system overhead low, and sleeps and resumes like a console instead of a laptop. Windows is a general-purpose operating system being asked to behave like an appliance. SteamOS was only ever the appliance. ## Why a studio reads this as a design note. We build bespoke systems for a living, and the standing pressure on every one of them is to reach for the general-purpose thing: the off-the-shelf platform, the framework that does everything, the stack the last project used. It is the safe call and often the right one. This is the counterweight, made unusually legible because Valve ran the experiment in public on one box. When you shape the whole stack around the single job the thing has to do, you are not just tidying up. You are leaving 11 to 75 percent on the table for anyone who does not. The wins are boring and that is the point. Fewer processes competing for the same cores. A resume that is instant because nothing has to be reloaded. Work done ahead of time rather than on the frame you needed it. None of it is a clever trick. It is the compounding return of a system that only has to be good at one thing, running on hardware that was going to be identical either way. ## Where the general platform still wins. Purpose-built is not a free win, and treating it as one is how you end up rebuilding a bespoke platform that a boring, supported one would have carried for years. Windows still runs the anti-cheat that keeps some games out of SteamOS entirely, and it is the machine you reach for when the device has to do more than the one job tomorrow. The lesson is not that specific always beats general. It is that the cost of general is real, usually invisible, and worth pricing before you default to it. On a handheld that cost is a benchmark anyone can read. On the systems we ship it is quieter, but it is the same tax. ## References 1. [SteamOS up to 75 percent faster than Windows 11 on the Legion Go S, XDA Developers (2026)](https://www.xda-developers.com/steamos-destroys-windows-and-i-have-the-receipts-to-prove-it/) 2. [Legion Go 2 SteamOS launch and 1,199 dollar price, TechTimes (2026)](https://www.techtimes.com/articles/318296/20260612/lenovos-1199-legion-go-2-steamos-arrives-this-month-challenge-steam-deck.htm)