Inline GPU Packet Processing with NVIDIA DOCA GPUNetIO

Originally published at:

Learn how the new NVIDIA DOCA GPUNetIO Library can overcome some of the limitations found in the previous DPDK solution, moving a step closer to GPU-centric packet processing applications.

A very interesting article. If the payloads of the packets were encrypted (e.g. they represent a TCP connection with TLS enabled) is there anything in DOCA to facilitate decryption/encryption. If not how would this be done in a CUDA kernel? Thx

Thanks for the feedback! I’m happy to answer your questions, however, this would currently require an NDA. You can register for Early Access here and I’ll connect with you then.

Hi I am facing some difficulty.

I want to use DPDK to perform burst transfer to My RTX 3090 GPU. I am unable to find any methods. I would appreciate if someone can help me with that

When compiling, I only have options for the A100 and not for the V100. Does this mean it doesn’t support the V100?