- Is Jetson Xavier is automotive compliant? if yes what is ASIL Level supported?
- Is real time linux support available on Jetson xavier SDK?
if yes , please provide details on real time features. - Which compilers are supported for Jetson xavier SDK? Is MISRA 2012 complaint?
Hi dattatraya.bora, Jetson AGX Xavier is not ASIL-certified. For automotive applications, please refer to the NVIDIA DRIVE platform.
Jetson AGX Xavier has Cortex-R5 core which can be used as a realtime co-processor, known as the Sensor Processing Engine (SPE). We provide a reference sample running FreeRTOS. For more info, please see these downloads:
- https://developer.nvidia.com/embedded/dlc/NVIDIA_Jetson_Sensor_Processing_Engine_DevGuide
- https://developer.nvidia.com/embedded/dlc/l4t_rt_aux_cpu_src
Also, I haven’t tried this myself, but I believe the community has been able to enable the PREEMPT_RT patch in the Linux kernel:
https://medium.com/@r7vme/real-time-kernel-for-nvidia-agx-xavier-b660e107a211
Also, Concurrent-RT has a realtime Linux kernel called RedHawk that supports Jetson:
https://www.concurrent-rt.com/redhawk-linux-nvidia-jetson-support/
The compilers are the standard ones provided by the Ubuntu apt repository packages. NVIDIA also provides the CUDA compiler (nvcc).
Thank you, Dustin :)
Yes, Mr. Dattatraya, Concurrent Real-Time does have a realtime Linux kernel running on Jetson AGX Xavier. Please find the press release at below link:
Concurrent Real-Time’s RedHawk linux kernel provides 6 “flavors” of kernel; debug (for driver development), trace (no debug features, but tracing feature for gauging/tuning RT performance) and standard (with debug or trace features). We also provide these kernel features with PREEMPT_RT patch enabled kernels.
Please note that, RedHawk Linux kernel doesn’t necessarily need PREEMPT_RT patch enabled in order to achieve low-latency realtime performance.
Hi Anish,
It nice to know that hard real time critical events can be handled using Concurrent RT. I have gone through the Press release link provided :
https://www.concurrent-rt.com/press-release/concurrent-real-time-introduces-redhawk-linux-nvidias-jetson-agx-xavier/
And it mentioned that RedHawk provides a guaranteed response time of less than 50 microseconds on the AGX Xavier. This 50uSec latency is measured in Top half or bottom half or user space.
I have a requirement like : Two dedicated interrupt will be generated from FPGA and these interrupts should be handled in two different cores like CPU affinity for GPIO Interrupts.
For Example: GPIO_INTR-1 should be handled in Core-5 and GPIO_INTR-2 in Core-6
Can Concurrent RT can handle the interrupts in two different cores individually, DO we have provision to set GPIO Interrupt affinity ?