Detected GPU
-
GPU Benchmark — Test Your Graphics Online
Benchmark your graphics card or integrated GPU in the browser. The test renders an increasingly heavy shader and adapts the load to find your GPU's sustained throughput — so the score isn't capped by your monitor's refresh rate. Compare your result on the leaderboard, or run the stress test — from one minute up to 90 minutes — to check for thermal throttling.
00:00
remaining
Plug laptops in for a fair score. Fans spinning up during the test is normal — that's the point. Longer runs are the best way to catch thermal throttling, and you can stop a running stress test at any time.
Live benchmark
Ready when you are — click Run GPU Benchmark.
Frame Rate
-
Shader Load
-
Score (live)
-
Your GPU score
Total Score
-
Sustained Load
-
GPU
-
Rating
-
Stress test result
Early Performance
-
Final Performance
-
Performance Drop
-
Leaderboard — top GPU scores
16 scores submitted. Nickname, score and detected GPU only — no accounts, no tracking.
| # | Name | Score | Hardware | Date |
|---|---|---|---|---|
| 1 | bad 5080 | 4924101 | ANGLE (NVIDIA, NVIDIA GeForce RTX 5080 (0x00002C02) Direct3D11 vs_5_0 ps_5_0, D3 | 2026-09-30 |
| 2 | nebula | 3348644 | ANGLE (NVIDIA, NVIDIA GeForce RTX 5080 (0x00002C02) Direct3D11 vs_5_0 ps_5_0, D3 | 2026-10-15 |
| 3 | Emmich | 3275011 | ANGLE (NVIDIA, NVIDIA GeForce RTX 4080 (0x00002704) Direct3D11 vs_5_0 ps_5_0, D3 | 2026-10-12 |
| 4 | Nr4 | 2188724 | ANGLE (AMD, Radeon R9 200 Series Direct3D11 vs_5_0 ps_5_0), or similar | 2026-10-17 |
| 5 | PsychoBunny | 1654910 | ANGLE (AMD, Radeon R9 200 Series Direct3D11 vs_5_0 ps_5_0), or similar | 2026-10-13 |
| 6 | b580 | 1353330 | ANGLE (Intel, Intel(R) Arc(TM) B580 Graphics (0x0000E20B) Direct3D11 vs_5_0 ps_5 | 2026-09-30 |
| 7 | DirtyRyan | 934908 | ANGLE (NVIDIA, NVIDIA GeForce GT 610 (0x0000104A) Direct3D11 vs_5_0 ps_5_0, D3D1 | 2026-10-15 |
| 8 | POCO F5 | 272977 | ANGLE (Qualcomm, Adreno (TM) 725, OpenGL ES 3.2) | 2026-10-11 |
| 9 | gigi | 246653 | ANGLE (NVIDIA, NVIDIA GeForce GTX 1070 (0x00001B81) Direct3D11 vs_5_0 ps_5_0, D3 | 2026-10-16 |
| 10 | ECDX | 157376 | Brave | 2026-10-10 |
| 11 | YNXR1000 | 68504 | ANGLE (Intel, Intel(R) Graphics (0x00007D41) Direct3D11 vs_5_0 ps_5_0, D3D11) | 2026-10-08 |
| 12 | piglet | 48297 | ANGLE (AMD, AMD Radeon (TM) Graphics (0x000015E7) Direct3D11 vs_5_0 ps_5_0, D3D1 | 2026-10-17 |
| 13 | X540SAA | 4195 | ANGLE (Intel, Intel(R) HD Graphics (0x000022B1) Direct3D11 vs_5_0 ps_5_0, D3D11) | 2026-10-17 |
| 14 | ThatMCDude | 1616 | ANGLE (Intel, Intel(R) UHD Graphics 600 (0x00003185) Direct3D11 vs_5_0 ps_5_0, D | 2026-10-15 |
| 15 | ThatCrystalDude | 1064 | ANGLE (Intel, Intel(R) UHD Graphics 600 (0x00003185) Direct3D11 vs_5_0 ps_5_0, D | 2026-10-15 |
| 16 | 5090 | 185 | ANGLE (AMD Radeon HD 6310 Direct3D11 vs_5_0 ps_5_0) | 2026-10-15 |
Earlier results (before the September 2026 scoring update)
These 40 scores were submitted before we fixed a scoring limit: the old test stopped adding load once a GPU was fast enough, so fast graphics cards were capped by the screen's refresh rate. They are kept for reference but are not comparable with the leaderboard above.
| # | Name | Old score | Hardware | Date |
|---|---|---|---|---|
| 1 | I HAVE BOTH 1 AND 2 L | 121649 | ANGLE (NVIDIA, NVIDIA GeForce RTX 5080 (0x00002C02) Direct3D11 vs_5_0 ps_5_0, D3 | 2026-09-20 |
| 2 | CZQQ | 121636 | ANGLE (NVIDIA, NVIDIA GeForce RTX 5080 (0x00002C02) Direct3D11 vs_5_0 ps_5_0, D3 | 2026-09-20 |
| 3 | audrey | 91254 | ANGLE (AMD, AMD Radeon RX 5700 XT (0x0000731F) Direct3D11 vs_5_0 ps_5_0, D3D11) | 2026-09-04 |
| 4 | had to cap at 160 smh | 60805 | ANGLE (NVIDIA, NVIDIA GeForce RTX 5080 (0x00002C02) Direct3D11 vs_5_0 ps_5_0, D3 | 2026-09-21 |
| 5 | tt | 45776 | ANGLE (Apple, ANGLE Metal Renderer: Apple M2 Pro, Unspecified Version) | 2026-07-24 |
| 6 | APPLE AGAIN | 45742 | ANGLE (Apple, ANGLE Metal Renderer: Apple M5 Max, Unspecified Version) | 2026-09-07 |
| 7 | AN M2 IS BEATING M5 MAX | 45737 | ANGLE (Apple, ANGLE Metal Renderer: Apple M5 Max, Unspecified Version) | 2026-09-07 |
| 8 | WHY | 45721 | ANGLE (Apple, ANGLE Metal Renderer: Apple M5 Max, Unspecified Version) | 2026-09-07 |
| 9 | APPLE | 45682 | ANGLE (Apple, ANGLE Metal Renderer: Apple M5 Max, Unspecified Version) | 2026-09-07 |
| 10 | APPLE AGAIN AGAIN | 45674 | ANGLE (Apple, ANGLE Metal Renderer: Apple M5 Max, Unspecified Version) | 2026-09-07 |
| 11 | APPLE AGAIN AGAIN | 45654 | ANGLE (Apple, ANGLE Metal Renderer: Apple M5 Max, Unspecified Version) | 2026-09-07 |
| 12 | cekidot | 39571 | ANGLE (Intel, Intel(R) UHD Graphics (0x00004626) Direct3D11 vs_5_0 ps_5_0, D3D11 | 2026-09-18 |
| 13 | mark | 38002 | ANGLE (NVIDIA, NVIDIA GeForce GTX 1650 (0x00002188) Direct3D11 vs_5_0 ps_5_0, D3 | 2026-08-19 |
| 14 | Lewa | 34205 | ANGLE (Intel, Intel(R) Iris(R) Xe Graphics (0x000046A6) Direct3D11 vs_5_0 ps_5_0 | 2026-09-04 |
| 15 | FPS | 28497 | ANGLE (AMD, AMD Radeon RX 6600 XT (0x000073FF) Direct3D11 vs_5_0 ps_5_0, D3D11) | 2026-09-06 |
| 16 | iPad M2 GPU | 22967 | Apple GPU | 2026-09-15 |
| 17 | iPad Air 5th Gen | 22842 | Apple GPU | 2026-09-02 |
| 18 | mac mini | 22819 | ANGLE (Apple, ANGLE Metal Renderer: Apple M4, Unspecified Version) | 2026-09-01 |
| 19 | xps | 22815 | ANGLE (Intel, Intel(R) Iris(R) Xe Graphics (0x00009A49) Direct3D11 vs_5_0 ps_5_0 | 2026-07-25 |
| 20 | zenbook | 22810 | Intel(R) HD Graphics, or similar | 2026-08-12 |
About the GPU benchmark
The GPU benchmark renders a demanding WebGL shader and keeps raising the workload — first the number of shader iterations per pixel, then the number of full-screen passes per frame — until your graphics hardware can no longer hold a smooth frame rate. The score is the sustained work per second once the load has settled.
Adding extra passes means fast graphics cards are not limited by your monitor’s refresh rate: a high-end desktop GPU keeps climbing where an integrated chip levels off. Because the scoring changed in September 2026 to allow this, the leaderboard was reset so every entry is measured the same way.
Browsers expose the GPU through WebGL rather than full native access, so treat the result as a quick comparison and a health check — for example confirming a laptop is really using its dedicated graphics — rather than a replacement for game benchmarks.
Understanding your GPU score
The benchmark renders a heavy WebGL shader and adapts the workload to your graphics hardware, measuring sustained shader throughput. Because the score is work multiplied by frame rate, fast GPUs are not capped by monitor refresh rate.
What this test checks
- Sustained shader throughput under an adaptive load that targets your GPU’s limit.
- Detected GPU model via the browser’s renderer information, shown on the leaderboard.
- Live frame rate and total shader load while the test runs.
- An optional stress test (1 to 90 minutes) comparing early and late performance for throttle detection.
How to understand your results
- Integrated graphics typically score around 15,000–40,000; dedicated gaming cards score well above that because the test keeps adding load until they are fully busy.
- On battery, or with the browser's hardware acceleration disabled, scores collapse - check both before judging hardware.
- A blank or generic GPU name means the browser is hiding renderer details for privacy - the score is still valid.
- Stress-test drops beyond about 15 percent indicate heat build-up in the GPU or shared laptop cooling.
Common problems and fixes
Score is tiny on a gaming PC
Check the browser is using the dedicated GPU, not the integrated one: Windows Settings > System > Display > Graphics lets you force high performance for your browser.
WebGL not available
Enable hardware acceleration in your browser settings and update graphics drivers, then restart the browser.
Laptop scores lower than yesterday
Almost always power state: battery saver, unplugged, or a hot room. Benchmarks are only comparable under the same conditions.
Screen flickers during the test
Heavy GPU load can trigger brief flicker on marginal drivers. Update the graphics driver; if flicker appears in games too, investigate cooling and the PSU on desktops.
FAQ
Is this the same as a native GPU benchmark?
No - it measures browser-accessible WebGL throughput, which is ideal for quick comparisons and leaderboards but less exhaustive than native tools.
Why is my GPU name not shown?
Some browsers mask the renderer string for privacy. The benchmark still runs and scores normally.
Does the test use my dedicated GPU or integrated graphics?
Whichever the browser is assigned. On dual-GPU laptops you can force the browser onto the dedicated GPU in your OS graphics settings.
Is the GPU stress test safe?
Yes — it is ordinary rendering load, the same as a game. Modern GPUs throttle themselves long before any risk, and you can stop the test at any time.
Why was the GPU leaderboard reset?
Earlier versions stopped adding load once shader iterations maxed out, so fast GPUs were capped by monitor refresh rate — an RTX 4090 on a 60 Hz screen could score the same as integrated graphics. The current version adds extra passes, and old scores are not comparable.
Does my monitor refresh rate affect the score?
Not meaningfully any more. The load keeps rising until frame rate falls below about 50 fps, so the score reflects how much work the GPU does per second rather than how many frames the monitor shows.
Browser-based test disclaimer
TestMyPC results are browser-based and may vary depending on your device, browser, operating system, permissions, drivers, connection, power settings and connected accessories. Use these tools as practical first checks, then confirm important faults with operating-system diagnostics, manufacturer tools or a qualified technician where needed.