LcdWriteDataMultiple refresh line by line

pcu · Jan 25th 2022, 2:05pm

Hello,

I'm porting a project from 16-bit parallel driving to the SPI bus.
The screen is 240 x 320 x 2 bytes.

I'm at the maximum SPI: 32MHz
I calculated that the refresh of full screen is around 40ms, but in reality it takes around 100ms.

I think the bottle neck is LcdWriteDataMultiple function which send data by line (640 bytes).
I use this init:

Source Code

GUIDRV_FlexColor_SetFunc(pDevice, &PortAPI, GUIDRV_FLEXCOLOR_F66709, GUIDRV_FLEXCOLOR_M16C1B8);

Can you tell me if it is possible to transmit the entire framebuffer by only one call of LcdWriteDataMultiple()?

Thank you,
Pierre

SEGGER - Schoenen · Jan 31st 2022, 11:47am

Hi,

Unfortunately, this is not possible.

Just imagine the rectangle to be send to the display has not the width of the display. Although we have a pointer to the start address we can not send it as one complete block as the memory is interrupted by pixel data not belonging to the rectangle (see attached).

You might try to use the window create flag WM_C_MEMDEV when creating a window. This will result in automatic usage of memory devices. Every drawing operation into a window will be performed into a memory device first. Once the drawing operation is finished the memory device will be send as an entire block (almost the same as a bitmap) to the display.

You might want to experiment with it and find out what fits better, using WM_CF_MEMDEV or a display cache.

If you like to use the WM_CF_MEMDEV flag for all windows just call WM_SetCreateFlags(WM_CF_MEMDEV); before GUI_Init(). All window created after the function call will use the flag, including the desktop window.

Regards,
Sven

pcu · Feb 3rd 2022, 2:22pm

Hi Sven,
Thank you for your explanation.

Finally, sending data by SPI line by line is quite fast for the application even if it is a little slower than in parallel mode.

Kind regards,

Source Code

Share