Current Projects

GPU-accelerated inline packet processing using GPUDirect RDMA

Description: GPU-accelerated inline packet processing using GPUDirect RDMA and Melanox Bluefield2.

Project Repo:

• VectorFlow-GX

Blog Post Series:

• GPUDirect RDMA: Fast Packet Processing

Helpful Sources:

1. Boosting Inline Packet Processing Using DPDK and GPUdev with GPU
2. Developing Linux Kernel Module using GPUDirect RDMA
3. Translating Addresses in Kernel Space
4. NVIDIA DOCA Libraries
5. Boosting Inline Packet Processing using DOCA GPUNetIO
6. CUDA Programming Guide
7. Linux Kernel: DMA support for PCI Peer-to-Peer

Hyperconverged Infrastructure

Description: Developing a Hyperconverged Infrastructure (HCI) solution for edge deployments

Helpful Sources:

1. EVE-OS
2. Research of a Virtual Infrastructure Network with Hybrid Software-Defined Switching, Ushakov
3. spdk
4. libVirt
5. Automated Hypervisor Paper, Kim et al
6. Open vSwitch
7. io_uring
8. portable hardware locality

Reverse Engineering CPUs and Hardware

Description: Reverse Engineer CPUs to learning

Helpful Sources:

1. Apple Platform Security
2. Asahi Linux Project
3. m1n1: Bootloader
4. Linux x86 Boot Protocol
5. Linux AArch64 Booting
6. IDA Pro

Linux Network Development & Performance Engineering

Description: Learnining as much as possible about NetDev to build high performance solutions

Helpful Resources

1. eBPF
2. Sockets in Linux Kernel
3. XDP Tutorial
4. NUMA: Linux Kernel Memory Policy
5. NUMA: Weighted Interleaving
6. General-Purpose Graphics Processor Architecture
7. Kernel Network stack & tuning (series)
8. Understanding __sk_buff
9. Linux Kernel DMA Docs