I do agree with you on the fill rate aspect now. I don't believe the bottleneck is straight computational performance either though. Note how the 1070 is more than twice as fast as the 1060 despite having less than 50% more shader resources while being an identical architecture. It seems the benchmarks works in mysterious ways.
The bottleneck is maybe somewhere else but we will have soon a game benchmark by Digitalfoundry where I expect the PS5 to perform much slower than a 2080 Ti and it will be interesting to compare to a 5700XT.
EDIT:
https://sluglibrary.com/SlugDemo.zip
There is a demo available on windows it is possible to compare GPU result.
https://sluglibrary.com/
The demo use arial but it is possible to change to any font.
Slug is distributed as a static library that has a basic C++ interface, and full source is included with every license. Slug is platform agnostic and runs on Windows, Mac, Linux, iOS, Android, and all major game consoles. It can be used with Vulkan, Direct3D 10/11/12, OpenGL 3.0–4.6, Metal, and WebGL2. The API is fully documented in the Slug User Manual.
The primary function of Slug is to take a Unicode string (encoded as UTF-8), lay out the corresponding glyphs, and generate a vertex buffer containing the data needed to draw them. When text is rendered, your application binds the vertex buffer, one of our glyph shaders, and two texture maps associated with the font. One texture holds all of the Bézier curve data, and the other texture holds spatial data structures that Slug uses for efficient rendering.
Slug can render each glyph as a single quad. This has the advantage that it’s easy to clip to an irregular shape or project onto a curved surface in a 3D environment. It also means that vertex buffer storage requirements do not depend on the complexity of the glyphs. There is also an optimization that uses a tighter polygon having between three and six vertices that can be enabled to increase speed at moderate and large font sizes.)
Last edited: