S32G IPCF communication problem

[Alex_mjl]

Hi NXP,

    I am using the S32G274A chip. I want to know what is the bandwidth and latency of core A and core M when they use IPCF for inter-core communication. These parameters do not appear to be mentioned in the user manual.

[alejandro_e]

Hello @Alex_mjl,

Thanks for contacting NXP support. About your question, we don't have public latency data related to transfers between the cores A and M. Could you tell me more information about your setup so I may share the information with the internal team, and ask if there is performance data related to the IPCF in the S32G2? please help me providing the following:

• Are you using the RDB2 or a custom board?
• Which BSP are you using?
• Which configuration are you using for the M7 core, Bare-Metal, FreeRTOS, AUTOSAR?
• could you describe a little about the scenario in which you are going to use IPCF?

Thanks in advance for the information.

[Alex_mjl]

Hi alejandro_e,

    Thank you for your reply. Pls allow me to describe the use case first: I want to use M-core to collect low speed bus data, such as CAN, LIN, FlexRay, etc. Then the collected data is sent to A-core through IPCF for calculation. That's why I want to know the bandwidth and latency of IPCF communication.

    Now let me answer your question:

• custom board
• BSP38
• bare metal

[alejandro_e]

Hello @Alex_mjl,

Thanks for the information. I will share this topic with the internal team  and let you know if we have specific data on the IPCF latency. For now I can tell you that IPCF transfers are faster than the protocols you mentioned, so it will not be a bottleneck in your application.

I will get back to you as soon as I have new information.

[alejandro_e · Accepted solution]

Hello @Alex_mjl,

Sorry for the late reply, the internal team was overloaded this past week. They provided the following information:

"

Since the IPCF is share memory based, it won't be the bottleneck for those low data volume applications.

Regarding the latency, a processing latency exists and depends on the interrupt handler on the A53 OS side. We don't have the latency result of the interrupt handler. Customer integrates the IPCF into their OS, it varies and depends on the OS they choose. 

" 

Please let me know if this information answers your question.