Hello,
I have a question about integrated graphics and WebGL itself. I have two laptops - with 11th Gen Intel(R) Core™ i5-1135G7 and 6800h. Both are plugged in, without energy saving, but the results are shocking. I did tests on https://web.basemark.com/ and I got something like this
| Test |
i5-1135G7 |
Ryzen 7 6800H |
Difference |
| Result |
573 |
1691 |
~3x |
| Array Test |
1372.48 |
5426.81 |
~4x |
| WebGL 1.0.2 Test |
507.00 |
3243.38 |
~6.4x |
| WebGL 2.0 Test |
396.22 |
2355.14 |
~5.9x |
| Shader Pipeline Test |
227.30 |
2185.91 |
~9.6x |
| Draw-call Stress |
179.66 |
1574.85 |
~8.8x |
| Geometry Stress |
233.60 |
2023.29 |
~8.6x |
| Canvas Test |
676.26 |
1264.34 |
~1.9x |
| AngularJS Test |
1371.01 |
2004.02 |
~1.5x |
| DOM Create Source Test |
762.03 |
2408.00 |
~3.2x |
The problem is GPU usage—Intel has 90%, which results in throttling and terrible performance. Has anyone tried to deal with such a situation? Is this a problem with a driver? How can I detect that, and how can I react to a low-performance device? 30fps and lower-boundaries scenes don’t help
Are you using a Texture Atlas / how many total textures are you using and what is the size? Ideally you will want to use 1 texture atlas of 2048 x 2048 or less for low end devices.
Hello.
I don’t know how to help you, but I can describe similar problems I’ve encountered.
A computer may have multiple video cards. When using a web browser, one of them may be used. The question is: who decides which video card is selected for use? This decision may be made by the driver or the operating system. Since web browsers are not designed for gaming, selecting a specific GPU for the browser can be a difficult task — or even impossible in some cases. JavaScript doesn’t have direct access to the hardware.
This issue is partially covered by WASM, but it still runs through the browser and within its limitations.
How can you find out which video card or processor is being used (especially since the main load may sometimes be on the CPU rather than the GPU)? It depends on the operating system and driver capabilities. Typically, this is done using specialized profiling tools.
You can also check “chrome://gpu” in the Chrome browser — but this does not solve the hardware selection issue.
Why do I consider hardware selection a problem? Because many users or players often don’t understand the technical details. It’s not a technical limitation — it’s a user experience issue that creates a high entry barrier for players.
Are you using a Texture Atlas / how many total textures are you using and what is the size? Ideally you will want to use 1 texture atlas of 2048 x 2048 or less for low end devices.
blank phaser canvas, tests were made on 10000x7000px on max zoom. 60fps, nothing being rendered(except the color)
Why do I consider hardware selection a problem? Because many users or players often don’t understand the technical details. It’s not a technical limitation — it’s a user experience issue that creates a high entry barrier for players.
There are a lot of Intel CPUs with integrated graphics, so switching hardware doesn’t solve anything. As a flagship example, GTA 5 FHD on medium settings can get 40fps with 95% GPU usage, while blank WebGL has similar results with just a blank canvas zoomed out. I can’t determine if the WebGL is limited, 70 million pixels to handle is too much(as the benchmark shows, it is probably a problematic, but not the main reason), or if this is a browser/driver/hardware/os issue using Intel’s iGPU.
I’m trying to figure out if this is a common problem, and have you ever encountered something like that?