Config «Simple»:
64Bytes Frame, 1 LRD / 1 LWR, No DC, No Mailbox (EK1100 – EL1004 – EL2004)

Config «Complex»:
1 Full frame (1514Bytes frame for PDO), LRD / LWR commands, DC, Mailbox via EVENT (EK1100-EL1004-EL2004-EL1252-EL3102-EL4102-EL6751-EL6751)

Test Platform:
CPU: Intel® Core i7-3770 CPU 3.40 GHz 3.39GHz
NIC: Intel® Gigabit CT desktop adapter
RAM: 8GB

1. 1. 比较在不同实时操作系统(RTOS)上运行的主控制栈性能结果,但在相同目标Intel Core i7上运行(所有主控制栈线程都在单个CPU核心上)。

Config «Simple»:
INtime RTX qnx vxworks" xenomai
Master cycle time, µs 100 100 100 100 100
Jitter, ±µs 5 2 7 10 1
Slave response time (avg), µs 43 32 43 33 22
CPU load (avg), % 16 21 21 21 13,3

Config «Complex»:

INtime RTX qnx vxworks" xenomai
Master cycle time, µs 200 200 200 200 200
Jitter, ±µs 4 2 3 4 1
Slave response time (avg), µs 170 150 171 166 154
CPU load (avg), % 10 12 13 13 3,8

2. 在不同实时操作系统(RTOS)上运行的最佳主控制栈性能结果,但在相同目标上(所有主控制栈线程都在单个CPU核心上)。

Config «Simple»:

INtime RTX qnx vxworks" xenomai
Master cycle time, µs 50 50 100 100 50
Jitter, ±µs 10 3 7 10 1
Slave response time (avg), µs 43 32 43 33 22
CPU load (avg), % 31 41 21 21 27,5
Config «Complex»:
INtime RTX qnx vxworks" xenomai
Master cycle time, µs 200 200 200 200 200
Jitter, ±µs 4 2 3 4 1
Slave response time (avg), µs 170 150 171 166 154
CPU load (avg), % 10 12 13 13 3,8

3. 比较在相同板上运行主控制栈时,采用不同的数据传输方式的性能
(例如,Hardware timed send).

Via Linux driver Via PRUSS driver Via PRUSS with HW timed send
Master cycle time, µs min 232 – max 1705 min 823 – max 1167 min 999 – max 1000
Slaves response time (avg), µs 348 176 7
CPU load (avg), % 47 32 29
详情请参阅

  Hardware Performance Comparison (957,9 КиБ, 99 hits)