APENet: LQCD clusters a la APE
R. Ammendola (1), M. Guagnelli (1), G. Mazza (1), F. Palombi (2), R., Petronzio (1, 3), D. Rossetti (4), A. Salamon (1), P. Vicini (4) ((1), INFN RM2, (2) E. Fermi Research Center, (3) Tor Vergata University, (4) INFN, RM1)
TL;DR
APENet is a high-speed, low-latency 3D interconnect architecture designed for PC clusters running LQCD applications, achieving performance comparable or superior to commercial networks.
Contribution
It introduces a novel PCI-X based interconnect with six channels optimized for LQCD, demonstrating promising benchmark results.
Findings
Peak bandwidth of 676 MB/s per channel
Performance comparable or better than high-end commercial networks
Hardware implementation based on a single PCI-X card
Abstract
Developed by the APE group, APENet is a new high speed, low latency, 3-dimensional interconnect architecture optimized for PC clusters running LQCD-like numerical applications. The hardware implementation is based on a single PCI-X 133MHz network interface card hosting six indipendent bi-directional channels with a peak bandwidth of 676 MB/s each direction. We discuss preliminary benchmark results showing exciting performances similar or better than those found in high-end commercial network systems.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.