“We have entered the age of specialization, when hardware and software must be co-designed together to achieve leading-edge capabilities, exactly as SiPanda is doing.”
- Doug Burger
Technical Fellow, Microsoft Azure Hardware
Programming and Network Datapath Acceleration (PANDA) framework. PANDA breaks the existing network datapath paradigm to provide the flexibility hyperscalers enjoy to everyone. PANDA allows network and application architects to program the capabilities for their network in C, parse it using PANDA, and run it on their existing network hardware.
Listen to a short introduction to our technology and the ways we will help you get a better network.
Existing network programming frameworks all have things they are great at, but none of them are general-purpose. By providing a low level abstraction layer, PANDA allows CSPs to write programs for their network in C/C++, and run them on their existing network hardware. PANDA complements tools like DPDK, P4, XBP/eBPF, and TC Flower to provide unmatched flexibility.
Delivering flexibility and terabit/second performance at scale. Panda-V is a domain specific architecture that tightly couples hardware and software to deliver next-gen performance across software defined, programmable network infrastructure. Panda-V is not an evolution of existing network paradigms – it is a complete rethink of how to design and implement a flexible, high-performance network.
SiPanda’s Panda-V serial data processing accelerator was built to carry programmable networks to terabit speeds and beyond. Panda-V was built to compliment the PANDA framework, and is based on a domain specific, enhanced OpenISA RISC-V architecture. It accelerates the network datapath in a similar way that GPUs accelerate graphics. Panda-V dramatically increases network performance, while reducing the I/O workload for the server CPUs.