I ported FreeRTOS onto a Nucleo_H723 with a W6100 shield. One thread runs CoreMark for performance benchmarking while another serves as an Iperf server to test
COMPONENTSHardware components
STM - NUCLEO-H723ZG
x 1
wiznet - W6100 (MKR-)Ethernet Shield
x 1
PROJECT DESCRIPTION
FreeRTOS, CoreMark, and iPerf Testing on STM32CubeIDE with nucleo-h723zg Board and w6100 Ethernet Shield
Hello readers! Today, I'd like to share some fascinating insights from a test conducted using the nucleo-h723zg board combined with the w6100 Ethernet shield in the STM32CubeIDE environment.
Setup:
Hardware: nucleo-h723zg board coupled with w6100 Ethernet shield.
Software: STM32CubeIDE, FreeRTOS, I-CUBE-CoreMark program (based on EEMBC's CoreMark), and wiznet's io6Library.
Once FreeRTOS was ported to the nucleo-h723zg board, we executed the I-CUBE-CoreMark program on one thread. This program provides a benchmark for CPU performance based on EEMBC's CoreMark. Simultaneously, on another thread, we initiated the iperf server using wiznet's io6Library to concurrently measure network performance.
Important Notes:
It's essential to note that setting up the I-CUBE-CoreMark is only possible with the STM32CubeIDE version 1.9.0.
Depending on the RTOS configuration, the CoreMark scores can vary. This implies that performance tuning for the RTOS can lead to different benchmarking outcomes.
Results:
Network performance without any other tasks (no load): 52.9Mbits/sec.
Performance when CoreMark is running concurrently: A decrease to 32.0Mbits/sec. Despite the reduction, the throughput remains commendable, showcasing the board's robust capabilities.
Conclusion:
The nucleo-h723zg board, when paired with the w6100 Ethernet shield, has demonstrated its prowess in handling both high computational and network operations simultaneously. The reduction in network speed during concurrent operations with CoreMark is anticipated due to shared CPU resources. Nevertheless, the board's performance remains notable. Such findings not only highlight the board's potential but also offer valuable insights for those aiming to optimize concurrent tasks in similar environments.
