/Lab2_Q1_816027279

Primary LanguageC

ECNG3006 Lab 2

Versioning

The various versions are managed using releases/tags.

The tags are as follows:

  • rrsched: Round Robin Scheduling without sleep on idle
  • priorinherit: Priority Inheritance without sleep on idle
  • priorinherit-alt: Priority Inheritance with modified priorities and no sleep on idle
  • rrsched-sleep: Round Robin Scheduling with sleep on idle
  • priorinherit-sleep: Priority Inheritance with sleep on idle
  • priorinherit-alt-sleep: Priority Inheritance with modified priorities and sleep on idle

All binaries and output files are available under the assets for each release.

P1

  • /project/sdkconfig

    • This is generated by make menuconfig.

  • /project/build/include/sdkconfig.h

    • This is generated from sdkconfig during the build (from make).

  • /sdk/components/freertos/port/esp8266/include/freertos/FreeRTOSConfig.h

    • This specifies the behaviour of freertos on the esp8266. It has default values for all the necessary definitions.

P2

The following must be defined as 1 to get the task performance information:

  • configGENERATE_RUN_TIME_STATS
  • configUSE_STATS_FORMATTING_FUNCTIONS
  • configSUPPORT_DYNAMIC_ALLOCATION

portCONFIGURE_TIMER_FOR_RUN_TIME_STATS must be defined as a function which will configure a timer and portGET_RUN_TIME_COUNTER_VALUE must be set to a function or value that returns the current count. The timer should count at at least 10 times the tick frequency.

The function void vTaskGetRunTimeStats( char *pcWriteBuffer );, defined in freertos/task.h, writes the runtime information to the buffer.

Question 2 - Vary the assigned priorities for your tasks, and look at the difference in performance. Is it what you expected?

The changing the assigned priorities did not change the performance much. The usage by each task usually matched each other. This is likely because each task relied on the execution before it to pass the semaphore, hence the priority was usually passed down to the lowest task, which was then increased by the highest priority task. This was what I had expected, given that the amount of execution time for each was expected to be low compared to the delay of 1 second each had.

Question 3 - Does the use of the processor sleep affect the system performance?

The use of processor sleep did1 affect performance. The impact was likely due to the overhead associated with sleeping and waking up the processor, but the sleep time chosen was small. The overall effect was that the percent usage for the IDLE task increased (and the other tasks decreased). The percent usage for the IDLE task (without sleep) was normally 27%, but with sleep it increased to around 33% (matching the LED on and off tasks).