项目
Power Capping and Oversubscription is a collaboration between MSR, Azure Compute, CO+I, and AHSI to harvest stranded datacenter resources via smart performance-aware power capping and oversubscription.
成立:
Function as a Service (FaaS) is a software paradigm that is becoming increasingly popular. Multiple cloud providers offer FaaS as the interface to usage-driven, stateless (serverless) backend services. FaaS offers an intuitive, event-based interface for developing cloud-based applications. In contrast…
Mobile & Internet of Things (IoT) devices, along with other battery operated devices, are energy constrained. While hardware capabilities have increased tremendously over the last ten years, battery energy density has only doubled. In this project we are exploring several…
Project Prometheus is building faster, more efficient datacenter systems by co-designing distributed systems with new network primitives. Prometheus takes advantage of new programmable hardware to accelerate applications.
成立:
No public cloud can host large latency-sensitive services, such as search engines, in a way that is economic for those services today! Project LEAP (short for Lean, Efficient, And Predictable) addresses the research challenges in enabling cloud platforms to host…
成立:
Resource Central is a general ML and prediction-serving system deployed in Azure Compute. It trains ML models offline and uses them to produce predictions online.
成立:
Komodo is a formally-verified reference monitor for an attested, secure isolated execution environment (“enclave”) on ARM TrustZone. It illustrates an alternative approach to Intel’s SGX, achieving similar security guarantees through formal verification, and allowing enclave features to evolve independently of…
The Zissou project is exploring immersion cooling in large-scale cloud platforms. Our main motivation is that chip power has been steadily increasing since the end of Dennard scaling.
The goal of Project Fiddle is to build efficient systems infrastructure for very fast distributed DNN training. Our goal is to support 100x more efficient training. Our innovations cut across the systems stack: the memory subsystem, structuring parallel computation across…