Questions about the browser texture test
What this test measures, what it cannot tell you, and where to go next.
Does this detect faulty VRAM?
No. The current workload samples textures and measures frame delivery. It does not read back memory and compare bytes. A clean run cannot rule out faulty VRAM; use a native data-verification tool for that question.
Is the tool free and does it need a download?
The browser workload is free and needs no account or installation. Exporting a report is optional and creates a local JSON file.
Is texture memory the same as my installed VRAM?
No. It is the calculated size of selected RGBA8 texture data: count × width × height × 4 bytes. It excludes other resources and does not measure installed, free or physically resident video memory.
Which graphics cards and browsers can run it?
It requires a working WebGL2 context. Support depends on the browser, driver, device and security settings, not just the GPU brand. An integrated GPU or software renderer may be selected. Check the renderer shown on the page.
Why is the result close to 60 FPS?
Animation callbacks generally follow display scheduling. A 60 Hz screen can limit this measurement to around 60 FPS even on a much faster GPU. Keep the display and browser the same when comparing runs.
Does a low score mean a damaged card?
No. The score is a site-specific blend of FPS normalized to 60 and frame consistency. Power settings, competing work, browser scheduling and software rendering can lower it. It is not a hardware-health rating.
What if the browser loses its graphics context?
The run stops rather than reporting success. Lower the workload, reload and repeat. Resource pressure or a driver reset can cause context loss; the event does not identify defective memory.
Where are results stored?
The test keeps results in page memory. Export creates a JSON download on your device. The site does not upload those results automatically. Ordinary hosting requests and page-view analytics are described in the privacy policy.
Can I compare this score with a game benchmark?
No. The workload, canvas resolution, frame scheduling and statistics differ. Use this page for repeated comparisons under the same conditions, and test the actual game or application for its performance.
What should I do about artifacts or crashes?
Record the triggering workload, return overclocks to stock and reproduce the problem. Follow the troubleshooting guide, then use native verification and a representative application. Persistent problems at stock settings need further investigation with the vendor or a qualified technician.