Blog

Check GPU shading throughput on a shared office PC

Measure a shared office PC's browser-based GPU shading and 1080p fill rate, compare repeat runs, and spot software rendering or background load.

Kevin DoyleHardware tests 5 min read

A graphics test on an office computer can help you check whether its browser is using graphics acceleration, or whether another task is affecting performance. On a shared PC, though, one result is only useful if you know what was running and how the computer was set up.

Use the Graphics Test to measure browser-based shading throughput and fill rate at 1080p. The test runs in the browser, so you can check the computer without installing software.

Run the graphics test

  1. Open the Graphics Test on the shared office computer. If the computer is managed by your workplace, follow its rules before running a performance test.
  2. Close video meetings, browser games, design apps, and other graphics-heavy work. Ask nearby users before closing anything, and do not interrupt their work.
  3. Keep the browser window open and follow the on-screen prompt to run the test. Avoid switching tabs or opening other apps while it runs.
  4. Record the shading result, fill rate, and score out of 100. Note the browser, whether the computer was plugged in, and whether the office computer was busy.
  5. After the computer has returned to its usual idle state, run the test again under the same conditions. Compare the readings rather than treating one score as a final verdict.

The test uses a 1080p graphics workload. That gives you a repeatable point of comparison, but it does not mean the computer's monitor must be set to 1080p.

What the result means

Similar scores on both runs

Results that stay close suggest the computer handled this browser graphics workload consistently. They do not prove that every graphics feature works, or predict the frame rate in a particular game or work application. For a useful comparison, test the same computer in the same browser with similar apps open and the same power conditions.

A much lower score on the second run

The difference can come from background work, a video call, another person using the computer, or heat from repeated activity. A browser graphics result is not a precise temperature reading. Let the computer return to its normal idle state, close unnecessary graphics tasks, and test once more.

Shading or fill rate is low compared with an earlier check

A lower result may reflect a browser update, a graphics driver change, a different power mode, or the browser using software rendering instead of the GPU. Software rendering means the computer's processor is drawing the graphics instead of the graphics processor. Check the browser's graphics acceleration setting and compare with a previous result from the same computer, not an unrelated machine.

The test is inconsistent or does not complete

High background load, a remote desktop session, a restricted browser, or a graphics feature problem can interfere. Retry after closing other work and reloading the page. If other browser graphics tasks are also slow, see what to check when a browser uses software graphics rendering.

The score looks good, but a work app still feels slow

The test measures one browser graphics workload, including shading and fill rate at 1080p. It does not measure every feature used by video editing, 3D design, or a particular game. The app may also be limited by the processor, memory, storage, its own settings, or the network. Treat the score as one clue, not a full diagnosis.

How to get a more reliable result

  1. Wait for shared activity to finish. Check that no one is presenting, streaming video, or using a graphics-heavy app. A busy shared computer can produce lower or less consistent results.
  2. Use the same browser and conditions. Repeating the test in a different browser can change the result. Record the browser and avoid comparing a plugged-in run with a battery-powered run.
  3. Check browser graphics acceleration. In Chrome or Edge, look under Settings › System › Use graphics acceleration when available. If you change this setting, restart the browser before testing again. On a workplace computer, an administrator may control the setting, so do not try to bypass that policy.
  4. Check the Windows graphics preference if allowed. In Windows 11, open Settings › System › Display › Graphics, select the browser, and check its available graphics preference. Options vary by computer. If it is managed, ask your IT team before changing it.
  5. Keep power conditions consistent. On a Windows 11 laptop, check Settings › System › Power & battery. On a Mac laptop, check System Settings › Battery. Power options vary by model, and managed devices may restrict them. For a before-and-after comparison, keep the same power source and mode.
  6. Restart only when you have permission. A restart can clear a stuck browser or background task, but it may disrupt other users or close unsaved work. Save what you need and follow office procedures first.

When it may be a graphics hardware problem

A low score alone is not enough to call the GPU faulty. Look for a pattern: repeated low results under quiet, consistent conditions, graphics glitches in more than one app, or crashes when normal graphics work begins. If the computer is shared or managed, send IT the test results along with the browser, date, power condition, and what was happening on the computer. Do not open the case or change drivers unless your workplace authorizes it.

If the same problem continues across browsers and graphics apps after IT checks the software and drivers, the graphics hardware or its cooling may need service. A browser test can flag an inconsistency, but it cannot identify a defective GPU component on its own.

What this test does not tell you

A browser benchmark cannot tell you the exact GPU temperature, prove that a particular graphics feature such as ray tracing works, or guarantee performance in a named application. Scores can also vary with browser versions, drivers, power limits, and other work on the computer. Use the test to compare repeat runs on this office PC, and share the context with IT if you need help interpreting a change.