# Performance and the benchmark > Measure how a pack runs on your PC, what you get, what is sent and how to take it back. Source: https://stygion.eu/docs/launcher/vykon Updated: 2026-10-16 The launcher can measure how a specific pack runs **on your computer**, and show an estimate from the measurements of similar PCs before you even download it. It is all your call: you start the test yourself, you see the result at once and without a network, and sending it is a second, voluntary step. ## Measure on my PC Open **Performance** in the instance detail and press **Measure on my PC**. - The launcher makes a **copy of the instance** and runs the test in it. Your instance, mods and worlds are not touched. - The game starts on a **fixed scene**: the pack generates its own world from a fixed seed and the camera flies a fixed path. The first pass generates terrain (it is not measured), the second is **measured for 90 seconds**. That is what makes two PCs comparable. - The window is **1920×1080**, or **1280×720** when the larger one does not fit. - **Keep the game in the foreground during the test and do not touch it.** Losing focus, minimising, an open menu or a changed window size and the run **does not count**; the launcher tells you why. - You can stop it **any time** with the button in the launcher; nothing is saved. > [!NOTE] > The test runs on **Minecraft 1.20 to 1.21.x**. On **26.x** we are still finishing the scene, so the test is coming soon. Versions older than 1.20 cannot be measured. The pack must come from CurseForge or Modrinth for the result to be sendable; other instances can be measured, but not sent. ### What you get The result is shown at once and **without a network**: - **average FPS** and **1 % low** (= 1000 / p99 frame time, that is how the worst hundredth of frames runs), - a **frame-time chart** per second, - **memory after garbage collection** and the **load times** of the game and the world, - **a sentence on what ran out first**: the graphics card or the processor. The launcher reads graphics load only where the system allows it; where it cannot, it says so instead of showing zero. Useful even without anybody else: change a setting, measure again and see whether it helped. ## Sending a result After a test you choose **Send**, or **Keep it only for me**. The first time we ask for your consent, and there is an **Exactly what is sent** view next to the button. - **Sent:** the pack and its version, the game and loader versions, the processor and graphics models, the driver version (major and minor only), RAM, the system, the window resolution, graphics and Java settings, the measured frames (a histogram), memory and load times. - **Not sent:** your name, IP address, serial numbers or anything from your worlds. - **Under your STYGION ID** (higher limits, withdrawal possible) or **anonymously** (lower limits). Who sent a run is never shown publicly. - The server wants **at least about 90 seconds since sending started**; the launcher shows a countdown. - The legal basis is your consent, which you withdraw in **Settings → Benchmark**. **Taking it back.** Sent runs are listed in **Settings → Benchmark**. **Withdraw and unlink** a run from your account: then it is not counted or shown anywhere. We do not delete runs, we pull them out of the counting and cut them from you. Runs kept only on your PC you delete there as well. ## Estimates on modpacks In **Discover** a pack card carries a chip like **"~70 FPS · 6 GB"**: an estimate from the measurements of **similar PCs** (the same class of graphics, processor, RAM and resolution). Hover to see how many runs it comes from; dots show the confidence. An estimate from other machines is labelled as such. - When there is not enough data the chip says **"not enough data"**. Never zero. - There is little data so far; it grows as people measure. **The measuring prompt.** Before the first **Play** of a pack that hardly anyone has measured on a similar PC, the launcher asks **once per pack** whether to measure it (about 3 minutes, the first time). You can measure and play, play right away, or stop asking. Turn it off for good in **Settings → Benchmark**. ## Passive measurement **Off** by default and switched on only with a separate consent in **Settings → Benchmark**. During normal play it measures **one 5-minute window per session** (after 10 minutes of play) and sends it separately, as "in normal play, with your own settings". It is never mixed with the test. ## Settings → Benchmark Everything above in one place: the prompts before Play, passive measurement, consent to sending (give and withdraw), sending under STYGION ID, your runs on this PC and the sent runs you can take back. > [!NOTE] > A badge for measuring (Benchmarker) does not exist yet; it comes later. If a pack still runs badly after measuring, try [a preset from the pack's author](/docs/launcher/presety), or [Potato Mode](/docs/launcher/presety#potato-mode).