Wrong delay when use "delay ms" ESP32

[flowjmaria]

Hello everyone, in a simple project with delay ms, when I only try to use delay between 1ms and 9ms the delay is wrong. However if I use delay us and delay s, no problem.

Only is for information of issue.

Thanks.

[LeighM]

This is due to the fact that the delay_mS resolution is limited by the RTOS tick period.
(RTOS has some advantages, some disadvantages

delay_uS does not allow other tasks to run, it just consumes CPU cycles

[flowjmaria]

Please, look this picture, there you can see what happen with delay_ms wrong between 1ms and 9 ms.

Thanks.

[BenR]

Hello,

Many thanks for reporting this. Are millisecond delays of 10ms and over ok?

I'll see if there is an easy way for us to fix this in the code.

[flowjmaria]

Hello Ben, yes over 10ms the delay is OK.

Thanks.

[BenR]

Aha thanks I've now made a change via the update system that hopefully will help.

Let us know how you get on.

[flowjmaria]

Hello Ben, Now it works OK.

Thanks a lot.

[jgu1]

Hi ben (hope it´s ok I ask now flowmaria?)

Seem that timing is ok now. But now I am even not possible to program the chip. And it´s the same program as first used.

[jgu1]

Me again!

I tryed to make a new project.............................. And yes, now Ili9341 work with ESP32

I work, but when I compare the speed of update with other Arduino project, it´s a little low, but ok. I though the ESP32 processor is very fast.

But finally, succes.

Br jorgen

[BenR]

Hi Jorgen,

Great news

I work, but when I compare the speed of update with other Arduino project, it´s a little low, but ok. I though the ESP32 processor is very fast.

The ESP processor is very fast but it seems the IO is not. I have done a bit of hacking away at the ESP IDF toolchain already to strip away some layers and this sped up the IO a fair bit but it's still a bit slow when compared to other MCUs.

Here is a similar topic on Espressif forums. We will keep looking for optimisations but it could be that raw IO bit toggling is just not very efficient on ESP devices.

https://www.esp32.com/viewtopic.php?t=1595

I have a plan to implement the RMT peripherals on the device which should allow for hardware driven fast comms. This will help to allow things like WS2812B and 1-Wire components to function but may also be useful for GLCDs. Hardware SPI might also be a good option.