Pegasus: Coordinated Scheduling for Virtualized Accelerator-based Systems

100 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Pathways: Asynchronous Distributed Dataflow for ML (2022, background): “Recent work shows that ﬁner-grained sharing can improve resource efﬁciency further: virtualizing accelerators (Yu et al., 2020; Gupta et al., 2011; Vijaykumar et al., 2016) avoids dedicating a whole accelerator to a single user.”
Dandelion: a compiler and runtime for heterogeneous systems (2013, background): “A production implementation of Dandelion could benefit significantly from kernel- or hypervisor-level support for GPUs to enable fairness and/or isolation guarantees [85, 47] in the presence of cluster sharing, or to optimize data movement with demand-driven data movement to the GPU [89] or with zer”
AvA: Accelerated Virtualization of Accelerators (2020, background): “API Remoting [30, 39, 41, 43, 44, 63, 64, 79, 101, 103] inter-poses the top of the stack, between application and framework library (A in Figure 2).”