High-Bandwidth Optical Network Interconnects for Machine Learning Training
High-Bandwidth Optical Network Interconnects for Machine Learning Training
Prio is a cryptographic technology that allows clients to submit metrics to servers in a private way, such that the servers only learn aggregated outputs.
RQS is an efficient de novo quality score compression tool based on traversing the k-mer landscape of NGS read datasets.
Quartz is an efficient de novo quality score compression tool based on traversing the k-mer landscape of NGS read datasets.
Cora is a compressive-acceleration tool for NGS read mapping methods
Ward is a Spectre and Meltdown resistant research operating system based on sv6 which was in turn based on xv6.
DeepPool is a prototype training system that achieves efficient strong scaling by implementing burst parallel training and GPU multiplexing.
Remy is an optimization tool to develop new TCP congestion-control schemes, given prior knowledge about the network it will encounter and an objective to optimize for.
Sprout is a transport protocol for interactive applications that desire high throughput and low delay.
Mahimahi is a tool for browser developers that provides accurate measurements when recording and replaying HTTP content over emulated network conditions