GPU acceleration with SPI displays (TinyDRM)

[Noralf Trønnes]

Hi,

At what speed are you running the SPI bus?

Theoretical fps for a 176x220 RGB565 display at 32MHZ:
32*1024*1024/176/220/16 = 54fps

Try modetest to see max fps:
https://github.com/notro/panel-mipi-dbi/wiki/Debugging#modetest

I don't remember the command line switch to have it run continously
showing the fps.

Noralf.


Den 04.08.2025 09:54, skrev Josef Luštický:
> Hello,
> is it possible to use GPU acceleration with TinyDRM drivers?
> I am testing on STM32MP157c SoC with Vivante GC400 GPU and I use
> Mesa3D 24.0.9 (rootfs generated with Buildroot 2025.02.3) and kernel
> 5.10.176.
>
> To compare, I also have the same system but with a MIPI DSI display.
> While kmscube renders at 60 fps with the MIPI DSI display, I get only
> 11 fps with the SPI display (TinyDRM driver).
>
> Is there some architectural limitation, either in Mesa, kernel, or HW-related?
> I suppose both setups use kernel's KMS as renderer only (kmsro), thus
> it should render at the same framerate into framebuffer.
> The difference in HW is the peripheral the displays use - MIPI DSI
> display utilizes LTDC peripheral (kmscube loads
> /usr/lib/dri/stm_dri.so) vs. SPI display uses just SPI (ksmcube loads
> /usr/lib/dri/ili9225_dri.so). Both .so files are hard links.
> Both open /dev/dri/renderD128 (GPU).
>
> The MIPI DSI display is connected to the DSI Host peripheral which
> just transcodes the parallel RGB pixel stream from the LTDC (LCD
> Timing Display Controller) peripheral of the STM SoC.
> The display driver is drivers/gpu/drm/panel/panel-sitronix-st7703.c (DRM panel).
>
> The SPI display driver uses the kernel's TinyDRM API and its driver is
> drivers/gpu/drm/tiny/ili9225.c , but I can change to any other driver
> if needed (eventually to the generic panel-mipi-dbi-spi).
> Could rewriting the TinyDRM driver to a panel driver in
> drivers/gpu/drm/panel/ help?
>
> I also tried software rendering "softpipe" (disabled GPU in
> device-tree) and it renders 7 fps on both systems. Thus, I suppose,
> that some form of GPU acceleration is already used.
>
> Mesa reports the same loaded GL-extensions on both systems and apart
> from loaded .so files everything seems to be the same.
>
> Best regards
> Josef Lusticky