Questions about DPDK, RSS configuration, and NIC-to-GPU DMA

Questions about DPDK, RSS configuration, and NIC-to-GPU DMA

[Sid ali cherrati]

[-- Attachment #1: Type: text/plain, Size: 1004 bytes --]

Hello DPDK team,

I’m currently working on setting up a server that receives UDP packets from
electronic boards integrated into detectors. The server needs to:

   -

   Listen on a specific IP address and port (provided at startup),
   -

   Efficiently receive these UDP packets,
   -

   Transfer the data directly to GPU memory, ideally avoiding memory copies
   using DMA.

I have a few questions regarding this:

   1.

   Is there already a DPDK-based software or framework (maybe a
   higher-level abstraction) that can handle RSS configuration, queue setup,
   and start listening on a specific IP and port as specified at runtime?
   2.

   Regarding NIC-to-GPU DMA: Is this only supported on NVIDIA GPUs, or are
   there any solutions or workarounds available for Intel GPUs as well?

Any guidance, examples, or references you could share would be greatly
appreciated.

Thank you for your support and for all your great work on DPDK!

Best regards,
SidAli

[-- Attachment #2: Type: text/html, Size: 1141 bytes --]