perf(gltf): optimize inner-loop getTransformMatrix call

[FishOfTheNorthStar]

There are several calls to getTransformMatrix, a fastgltf function, in our update loops that are called very often, especially for animated scenes. That function takes a 4x4 matrix and multiplies it against the local transform, to get a world transform, but in our calls to it we don't actually use that feature so it's just doing a bunch of multiplications for no good reason.

So this eliminates that unnecessary math to produce roughly 10-12% improved CPU-side performance, by my tests.

[github-actions]

Hi 👋, thank you for your PR!

We've run benchmarks in an emulated environment. Here are the results:

ARM Emulated 32b - lv_conf_perf32b

Scene Name	Avg CPU (%)	Avg FPS	Avg Time (ms)	Render Time (ms)	Flush Time (ms)
All scenes avg.	28	38	7	7	0

Detailed Results Per Scene

Scene Name	Avg CPU (%)	Avg FPS	Avg Time (ms)	Render Time (ms)	Flush Time (ms)
Empty screen	11	33	0	0	0
Moving wallpaper	2	33	1	1	0
Single rectangle	0	50	0	0	0
Multiple rectangles	0	38 (-1)	0	0	0
Multiple RGB images	0	38 (-1)	0	0	0
Multiple ARGB images	9	42	0	0	0
Rotated ARGB images	54	44 (+1)	15 (+1)	15 (+1)	0
Multiple labels	7 (+2)	35 (+2)	0	0	0
Screen sized text	97 (+1)	47	20	20	0
Multiple arcs	33 (-6)	33	7	7	0
Containers	1 (-2)	37	0	0	0
Containers with overlay	97 (+8)	21	44	44	0
Containers with opa	17	37 (-1)	0	0	0
Containers with opa_layer	18	34	6 (+1)	6 (+1)	0
Containers with scrolling	46 (-1)	46	12	12	0
Widgets demo	67	40	16	16	0
All scenes avg.	28	38	7	7	0

ARM Emulated 64b - lv_conf_perf64b

Scene Name	Avg CPU (%)	Avg FPS	Avg Time (ms)	Render Time (ms)	Flush Time (ms)
All scenes avg.	24	38	6	6	0

Detailed Results Per Scene

Scene Name	Avg CPU (%)	Avg FPS	Avg Time (ms)	Render Time (ms)	Flush Time (ms)
Empty screen	11	33	0	0	0
Moving wallpaper	1	33	0	0	0
Single rectangle	0	49 (-1)	0	0	0
Multiple rectangles	0	46	0	0	0
Multiple RGB images	0	39	0	0	0
Multiple ARGB images	1	38	0	0	0
Rotated ARGB images	29	34	9	9	0
Multiple labels	3	37 (-2)	0	0	0
Screen sized text	82 (+1)	45	17	17	0
Multiple arcs	34 (+1)	33 (-1)	6	6	0
Containers	4	37	0	0	0
Containers with overlay	88 (+1)	23	41	41	0
Containers with opa	16 (+1)	37	1 (+1)	1 (+1)	0
Containers with opa_layer	7 (-1)	39 (+2)	1	1	0
Containers with scrolling	45 (+1)	47 (+1)	12 (+1)	12 (+1)	0
Widgets demo	66 (-1)	42	15	15	0
All scenes avg.	24	38	6	6	0

Disclaimer: These benchmarks were run in an emulated environment using QEMU with instruction counting mode.
The timing values represent relative performance metrics within this specific virtualized setup and should
not be interpreted as absolute real-world performance measurements. Values are deterministic and useful for
comparing different LVGL features and configurations, but may not correlate directly with performance on
physical hardware. The measurements are intended for comparative analysis only.

---

🤖 This comment was automatically generated by a bot.

[cubic-dev-ai]

No issues found across 1 file

[AndreCostaaa]

Suggested change

	* scale(rotate(translate())) to align with fastgltf's internals, but that may change - MK
	* scale(rotate(translate())) to align with fastgltf's internals, but that may change

[FishOfTheNorthStar]

Oh thanks, I'll remember that for future. That comment is gone now.

[FishOfTheNorthStar]

Note: my last commit to this PR incorporates a revised order of operations just recently updated in fastgltf. Be sure to update submodules before building this.

[FishOfTheNorthStar]

Closing this PR to avoid confusion and merge conflicts since the order of operations it was changing has since been confirmed to be correct the original way, despite how it looks. #9273 is the confirmed correct way. I'll need to adjust two lines of 9273 to reflect the getLocalTransformMatrix call, but otherwise 9273 replaces this.