wenjieshen

Results 26 comments of wenjieshen

### VM coverage at/above OpenGL 3.3 (who lacks GL4?) **Summary:** The primary environments that are ≥GL 3.3 but **do not** expose GL4 to guests are those using **virgl** (virtio-gpu). Commercial...

This issue is similar to #4033. The input anti-aliased texture [^1] already exhibits a halo effect. When MSAA resolves the render target, it averages the samples again as follows. As...

On native machines, we can have direct FBO injection now. However, it's just a POC waiting for fine-tuning, such as MSAA, color space, and so on. ### Comparison In the...

> [@wenjieshen](https://github.com/wenjieshen) what do you think at performance? In the edge case, for example, a 4K resolution canvas with simple animation can experience a 3x speedup through direct rendering. It...

@hermet @SergeyLebedkin Thanks for reviewing. I’d like to confirm two points: Binary shader support is an OpenGL 4.1+ feature, while our requirement is OpenGL 3. I added a helper function...

> @wenjieshen Could you please hold this? I think we can discuss the opengl dependency further. Maybe we can bump up the opengl dependency to v4 as the minimum requirement......

@hermet Thank you. I have updated the commit message.

# Shader Binary‐Cache Experiment Results ## Overview This analysis examines shader creation and caching performance across 30 repeated executions, each compiling 16 shaders. Four distinct timing datasets were analyzed: -...

> @wenjieshen I think it's good to add `common/tvgEnv.h` (For now, code in the header could be better for size optimization), which provides the base thorvg config path based on...

Thank you for the suggestion. I’ve added a validation header to the binary file. After corrupting a file (as shown in the screenshot above), the cache successfully detected the issue...