Query on HSE Random Number Generation Performance and Initialization Handling for MCU0/MCU1

[nitins25]

Hello NXP Team,

Based on our measurements on the S32K312 and S32S388 MCU0/MCU1 projects, we observed that the initial "Random Number Generate" HSE service — which performs the internal seeding operation — takes approximately 60 ms to complete.

From the second iteration onward, the response time reduces to around 8 ms.

We are using this service within the Security Access (0x27) UDS routine, where the HSE generates and provides the random number.

As per our earlier discussion with NXP in one of the forums (see attached snippet), we understood that this delay is related to the HSE’s seeding process.

Could you please confirm:

1. Whether any optimization is possible to reduce the initial latency?

2. If this limitation must be accepted as is, would it be considered a deviation from the GM CTRS timing requirements?

3. Is the RNG initialization (seeding) handled automatically within Crypto_HSE_Init(), or is the integrator expected to invoke it explicitly during system initialization?

Thank you for your support.

Best regards,

Nitin

[lukaszadrapa]

Hi @nitins25 

Unfortunately, there is no option to reduce this latency.
The function Crypto_43_HSE_Init does not address this delay. It performs pure software initialization of the Crypto driver and does not trigger any HSE services.
Regarding the deviation from GM CTRS timing requirements, I’m unable to provide a definitive answer. As a GM, you should have access to dedicated FAE support, either through the private GM community or via the NXP support portal: https://support.nxp.com/s/?language=en_US
I recommend using those channels to reach the appropriate team for further assistance.

Regards,
Lukas