Computer Laboratory

pcie-bench: A tool to understand PCIe implementations

Running pcie-bench

There are two implementations currently available of this tool; one is based on the NetFPGA SUME interface card, the other works on the Netronome NFP-6000 card.

Repository along with instructions is available at

Public results dataset

Since we have found PCIe performance to be a complex, multi-variate problem, and because we believe that having an open corpus of performance data will be useful to guide the development of hardware, adapter, interface, network, Silicon, along with development of hypervisors, kernels and applications, we provide a database of aggregated pcie-bench datasets. We generated some of these datasets ourselves over the course of our research, while others were contributed by the community.

» Results database

Please support this community-driven public database by running pcie-bench on your own machines and contributing the results!

Submitting Results by emailing the output file to

N.B. Any output file email that is not in the correct format is treated as spam


If referencing the PCIe-bench tookit, please cite the Sigcomm 2018 paper, thank you.

Rolf Neugebauer, Gianni Antichi, José Fernando Zazo, Yury Audzevich, Sergio López-Buedo, and Andrew W. Moore. 2018. Understanding PCIe performance for end host networking. In Proceedings of the 2018 Conference of the ACM Special Interest Group on Data Communication (SIGCOMM '18). ACM, New York, NY, USA, 327-341. DOI:

Official ACM Digital Library link ACM DL PDF backup PDF SIGCOMM2018 slides

This research is done by colleagues and collaborators from several organizations including several members of the NetOS group at the University of Cambridge Computer Laboratory. If you have any questions or would like to learn more, please contact us directly:


Postal address:
NetOS group
University of Cambridge Computer Laboratory
Willam Gates Building
15 JJ Thomson Avenue
Cambridge CB3 0FD, UK
Tel.: +44 1223 763683