Transforming Next-Generation Wireless with 5T for 5G and the NVIDIA Aerial SDK

Technical Blogs & Events Technical Blog
1

Originally published at: Transforming Next-Generation Wireless with 5T for 5G and the NVIDIA Aerial SDK | NVIDIA Technical Blog

Figure 1. 5G wireless uses an open and cloud-native radio area network (CloudRAN). NVIDIA Mellanox 5T for 5G technology provides a real-time and high-performance solution for building an efficient, time-synchronized CloudRAN infrastructure. Time synchronization and achieving high time accuracy for network traffic between O-RAN 7.2x compliant front-haul, mid-haul, and back-haul components in a cloud-native RAN…

1 Like
2

Dear Jen,

Aerial SDK seems very promising. My research group at ELTE University works on ngRAN and UPF use cases intensively using DPDK and programmable switches (Tofino). For me, one of the most exciting parts of Aerial SDK is the GPU-accelerated DPDK which would be intresting to involve in our research. What is the official way for a university research group to get access to the early access program of Aerial SDK?

Many thanks in advance,
Sandor Laki

3

Dear Sandor,

Thanks for your interest. We will be glad to discuss further your research goals and how we can support you.

To kickstart the process, sign up for NVIDIA’s Aerial developer program here: https://developer.nvidia.com/aerial-sdk
We are accepting all research applications at this time.

If you are interested in early access to the Aerial developer kit, you can request it here: https://developer.nvidia.com/nvidia-aerial-devkit

All the best and we look forward to working with your team.

4

Dear Ash,

I applied for the early access program about 3 weeks ago. How long does it take to get the approval? We see several opportunities for exploiting the benefits of Aerial: 1) We work on an open source P4 compiler for DPDK target. It is called T4P4S and available on GitHub: GitHub - P4ELTE/t4p4s: Retargetable compiler for the P4 language 2) We also work on a hybrid RAN solution where Intel/Barefoot Tofino switches are extended with external computational units to implement the required features that are not supported by the Tofnio ASIC (a short paper will soon be available in the proc. of EUROP4 WS). In our lab, we have Mellanox ConnectX-5 cards and Quadro cards supporting GPUDirect RDMA, so I hope they are enough for carrying out the first tests with GPU-DPDK.

Thank you in advance,
Sandor

5

Hi Sandor,

You are approved to get access to Aerial SDK. You should have received an email about Approval. Please follow the instructions on https://developer.nvidia.com/nvidia-aerial-get-started to download the files. Let us know if you have more queries.

Thank you,
Chaitrali

6

Dear Chaitrali,

In the FAQ of the Aerial developer kit link, an answer says that NVIDIA Quadro GV100 and NVIDIA T4 Tensor Core GPUs are supported to run the Aerial SDK. Does Tesla V100 GPU support to run the Aerial SDK?

7

Dear Jen,

I have some questions with regard to 5T for 5G technology.
What’s the requirement to support the feature on a general linux based server with ConnectX-6 DX ?

I’m developing an application (with a server equipped with a ConnectX-6 DX) which transmits a user traffic from 10 to 200 Gbps and having difficulties in shaping a traffic at a specific rate.
It looks like “Packet Pacing” feature conforms to our requirement. What’s the requirement to support the feature on a general linux based server with ConnectX-6 DX ?

Thank you and Best regards,

Kevin