Rui's Blog
  • Rui's Blog/Paper Reading Notes - Introduction
  • Personal Blog
    • Personal Blog - Index
      • How to Create Picture-in-Picture Effect / Video Overlay for a Presentation Video
      • How to Do Your Part to Protect the Environment in Wisconsin
      • How to Get a Driver's License in Wisconsin
      • How to Travel from the U.S. to China onboard AA127 in June 2021
      • How to Transfer Credits Back to UW-Madison
      • Resources on Learning Academic Writing (for Computer Science)
    • Towards applying to CS Ph.D. programs
  • Machine Learning Systems
    • Machine Learning Systems - Index
      • MLSys Papers - Short Notes
      • [2011 NSDI] Dominant Resource Fairness: Fair Allocation of Multiple Resource Types
      • [2014 OSDI] Scaling Distributed Machine Learning with the Parameter Server
      • [2018 OSDI] Gandiva: Introspective Cluster Scheduling for Deep Learning
      • [2018 SIGCOMM] Chameleon: Scalable Adaptation of Video Analytics via Temporal and Cross-camera ...
      • [2018 NIPS] Dynamic Space-Time Scheduling for GPU Inference
      • [2019 ATC] Analysis of Large-Scale Multi-Tenant GPU Clusters for DNN Training Workloads
      • [2019 NSDI] Tiresias: A GPU Cluster Manager for Distributed Deep Learning
      • [2019 SOSP] ByteScheduler: A Generic Communication Scheduler for Distributed DNN Training ...
      • [2019 SOSP] PipeDream: Generalized Pipeline Parallelism for DNN Training
      • [2019 SOSP] Parity Models: Erasure-Coded Resilience for Prediction Serving Systems
      • [2019 NIPS] GPipe: Efficient Training of Giant Neural Networks using Pipeline Parallelism
      • [2019 SC] ZeRO: memory optimizations toward training trillion parameter models
      • [2020 OSDI] Gavel: Heterogeneity-Aware Cluster Scheduling Policies for Deep Learning Workloads
      • [2020 OSDI] AntMan: Dynamic Scaling on GPU Clusters for Deep Learning
      • [2020 OSDI] BytePS: A High Performance and Generic Framework for Distributed DNN Training
      • [2020 SIGCOMM] Reducto: On-Camera Filtering for Resource-Efficient Real-Time Video Analytics
        • [2020 MLSys] Salus: Fine-Grained GPU Sharing Primitives for Deep Learning Applications
      • [2020 EuroSys] AlloX: Compute Allocation in Hybrid Clusters
      • [2020 VLDB] PyTorch Distributed: Experiences on Accelerating Data Parallel Training
      • [2020 NetAI] Is Network the Bottleneck of Distributed Training?
      • [2020 NSDI] Themis: Fair and Efficient GPU Cluster Scheduling
      • [2021 MLSys] Accordion: Adaptive Gradient Communication via Critical Learning Regime Identification
      • [2021 VLDB] Analyzing and Mitigating Data Stalls in DNN Training
      • [2021 FAST] CheckFreq: Frequent, Fine-Grained DNN Checkpointing
      • [2021 EuroMLSys] Interference-Aware Scheduling for Inference Serving
      • [2021 OSDI] Pollux: Co-adaptive Cluster Scheduling for Goodput-Optimized Deep Learning
      • [2021 MLSys] Wavelet: Efficient DNN Training with Tick-Tock Scheduling
      • [2021 NSDI] SwitchML: Scaling Distributed Machine Learning with In-Network Aggregation
    • Big Data Systems - Index
      • Big Data Systems Papers - Short Notes
      • [2003 SOSP] The Google File System
      • [2004 OSDI] MapReduce: Simplified Data Processing on Large Clusters
      • [2010 SIGMOD] Pregel: A System for Large-Scale Graph Processing
      • [2011 NSDI] Mesos: A Platform for Fine-Grained Resource Sharing in the Data Center
      • [2012 NSDI] Resilient Distributed Datasets: A Fault-Tolerant Abstraction for In-Memory Cluster ...
      • [2012 OSDI] PowerGraph: Distributed Graph-Parallel Computation on Natural Graphs
      • [2019 FAST] DistCache: Provable Load Balancing for Large-Scale Storage Systems with Distributed...
      • [2021 HotOS] From Cloud Computing to Sky Computing
      • [2021 EuroSys] NextDoor: Accelerating graph sampling for graph machine learning using GPUs
  • Earlier Readings & Notes
    • High Performance Computing Course Notes
      • Lecture 1: Course Overview
      • Lecture 2: From Code to Instructions. The FDX Cycle. Instruction Level Parallelism.
      • Lecture 3: Superscalar architectures. Measuring Computer Performance. Memory Aspects.
      • Lecture 4: The memory hierarchy. Caches.
      • Lecture 5: Caches, wrap up. Virtual Memory.
      • Lecture 6: The Walls to Sequential Computing. Moore’s Law.
      • Lecture 7: Parallel Computing. Flynn's Taxonomy. Amdahl's Law.
      • Lecture 8: GPU Computing Intro. The CUDA Programming Model. CUDA Execution Configuration.
      • Lecture 9: GPU Memory Spaces
      • Lecture 10: GPU Scheduling Issues.
      • Lecture 11: Execution Divergence. Control Flow in CUDA. CUDA Shared Memory Issues.
      • Lecture 12: Global Memory Access Patterns and Implications.
      • Lecture 13: Atomic operations in CUDA. GPU ode optimization rules of thumb.
      • Lecture 14: CUDA Case Studies. (1) 1D Stencil Operation. (2) Vector Reduction in CUDA.
      • Lecture 15: CUDA Case Studies. (3) Parallel Prefix Scan on the GPU. Using Multiple Streams in CUDA.
      • Lecture 16: Streams, and overlapping data copy with execution.
      • Lecture 17: GPU Computing: Advanced Features.
      • Lecture 18: GPU Computing with thrust and cub.
      • Lecture 19: Hardware aspects relevant in multi-core, shared memory parallel computing.
      • Lecture 20: Multi-core Parallel Computing with OpenMP. Parallel Regions.
      • Lecture 21: OpenMP Work Sharing.
      • Lecture 22: OpenMP Work Sharing
      • Lecture 23: OpenMP NUMA Aspects. Caching and OpenMP.
      • Lecture 24: Critical Thinking. Code Optimization Aspects.
      • Lecture 25: Computing with Supercomputers.
      • Lecture 26: MPI Parallel Programming General Introduction. Point-to-Point Communication.
      • Lecture 27: MPI Parallel Programming Point-to-Point communication: Blocking vs. Non-blocking sends.
      • Lecture 28: MPI Parallel Programming: MPI Collectives. Overview of topics covered in the class.
    • Cloud Computing Course Notes
      • 1.1 Introduction to Clouds, MapReduce
      • 1.2 Gossip, Membership, and Grids
      • 1.3 P2P Systems
      • 1.4 Key-Value Stores, Time, and Ordering
      • 1.5 Classical Distributed Algorithms
      • 4.1 Spark, Hortonworks, HDFS, CAP
      • 4.2 Large Scale Data Storage
    • Operating Systems Papers - Index
      • CS 736 @ UW-Madison Fall 2020 Reading List
      • All File Systems Are Not Created Equal: On the Complexity of Crafting Crash-Consistent Applications
      • ARC: A Self-Tuning, Low Overhead Replacement Cache
      • A File is Not a File: Understanding the I/O Behavior of Apple Desktop Applications
      • Biscuit: The benefits and costs of writing a POSIX kernel in a high-level language
      • Data Domain: Avoiding the Disk Bottleneck in the Data Domain Deduplication File System
      • Disco: Running Commodity Operating Systems on Scalable Multiprocessors
      • FFS: A Fast File System for UNIX
      • From WiscKey to Bourbon: A Learned Index for Log-Structured Merge Trees
      • LegoOS: A Disseminated, Distributed OS for Hardware Resource Disaggregation
      • LFS: The Design and Implementation of a Log-Structured File System
      • Lottery Scheduling: Flexible Proportional-Share Resource Management
      • Memory Resource Management in VMware ESX Server
      • Monotasks: Architecting for Performance Clarity in Data Analytics Frameworks
      • NFS: Sun's Network File System
      • OptFS: Optimistic Crash Consistency
      • RAID: A Case for Redundant Arrays of Inexpensive Disks
      • RDP: Row-Diagonal Parity for Double Disk Failure Correction
      • Resource Containers: A New Facility for Resource Management in Server Systems
      • ReVirt: Enabling Intrusion Analysis through Virtual-Machine Logging and Replay
      • Scheduler Activations: Effective Kernel Support for the User-Level Management of Parallelism
      • SnapMirror: File-System-Based Asynchronous Mirroring for Disaster Recovery
      • The Linux Scheduler: a Decade of Wasted Cores
      • The Unwritten Contract of Solid State Drives
      • Venti: A New Approach to Archival Storage
    • Earlier Notes
      • How to read a paper
  • FIXME
    • Template for Paper Reading Notes
Powered by GitBook
On this page
  • One-line Summary
  • Paper Structure Outline
  • Background & Motivation
  • Design and Implementation
  • Evaluation
  • Links

Was this helpful?

  1. Earlier Readings & Notes
  2. Operating Systems Papers - Index

Scheduler Activations: Effective Kernel Support for the User-Level Management of Parallelism

One-line Summary

The authors present a new kernel interface and user-level thread package that provide the same functionality as kernel threads without compromising the performance and flexibility advantages of user-level management of parallelism.

Paper Structure Outline

  1. Introduction

    1. The Problem

    2. The Goals of this Work

    3. The Approach

  2. User-Level Threads: Performance Advantages and Functionality Limitations

    1. The Case for User-Level Thread Management

    2. Sources of Poor Integration in User-Level Threads Built on the Traditional Kernel Interface

  3. Effective Kernel Support for the User-Level Management of Parallelism

    1. Explicit Vectoring of Kernel Events to the User-Level Thread Scheduler

    2. Notifying the Kernel of User-Level Events Affecting Processor Allocation

    3. Critical Sections

  4. Implementation

    1. Processor Allocation Policy

    2. Thread Scheduling Policy

    3. Performance Enhancements

    4. Debugging Consideration

  5. Performance

    1. Thread Performance

    2. Upcall Performance

    3. Application Performance

  6. Related Ideas

  7. Summary

Background & Motivation

Threads are built either at the user-level or kernel-level.

  • User-level

    • Advantages

      • Requires no kernel intervention, good performance: fast thread mgmt operations (context switches)

      • Flexible: Customizable for applications

    • Limitations

      • Poor integration with system services: Implemented over kernel-level threads, which block and are preempted w/o notifying user-level thread

      • Performs poorly during I/O, preemption, and page faults due to overhead of kernel trapping

      • Scheduled obliviously w.r.t. the user-level thread state

  • Kernel-level

    • Advantages

      • Each thread gets mapped to a physical processor while it is running

    • Limitations

      • Bad performance: Requires kernel intervention (switch into the kernel) for thread mgmt operations (fork, join, wait, signal, etc.)

      • Not as flexible (implemented in the kernel, so the scheduling policy cannot be changed easily later on)

The authors argue that kernel-level threads are inherently worse than user-level threads (extra kernel trap and copy operations), but IRL user-level threads many exhibit poor performance/incorrect behavior in multiprocessor systems. They then attempt to take the best of both worlds by building a new kernel interface and a variant of a user-level thread library that communicates effectively with the kernel to combine the functionality of kernel-level threads and the performance and flexibility of user-level threads.

Design and Implementation

Main contributions:

  • The kernel allocates a virtual multiprocessor to each application

  • Applications:

    • Told how many and which processors it has

    • A user-level application's own thread scheduler decides which threads to run on its allocated physical processors

  • OS:

    • The kernel is told how many threads an application would like to run so it can try to allocate that many physical processors for it

    • Complete control over which processors are given to which application

Scheduler Activations (SA) is a kernel mechanism that provides a communication structure between the kernel processor and the user-level thread system. This is a vectored event that causes the user-level thread system (via an up-call) to reconsider its scheduling decision of which threads to run on which processors when events (processor allocations and deallocations) need to take place.

The following roles are performed by SAs:

  • Notify the user-level thread system of kernel events

  • Provides the execution context for execution of user-level threads

  • Provides space for saving user-level thread context in the kernel when a thread is stopped by the kernel

Scheduler Activation Upcalls:

  • Add this processor

  • Processor has been preempted

  • Scheduler activation has blocked

  • Scheduler activation has unblocked

Application downcalls:

  • Add more processors

  • This processor is idle

Evaluation

  • Performance degrades slowly when available memory drops, and then more sharply once the application's working set does not fit in memory

  • Application performance with original FastThreads degrades more quickly: a user level thread blocks in the kernel → the application loses that physical processor for the duration of the I/O.

  • Application performance is better with modified FastThreads than with Topaz because most thread operations can be implemented without kernel involvement.

  • Topaz: Kernel-level thread

  • orig FastThrds: User-level thread

  • new FastThrds: Scheduler activation

Links

  • Thanks to Jiaxin Lin for the paper review notes!

PreviousReVirt: Enabling Intrusion Analysis through Virtual-Machine Logging and ReplayNextSnapMirror: File-System-Based Asynchronous Mirroring for Disaster Recovery

Last updated 4 years ago

Was this helpful?

Reading notes from and

Paper PDF
CS 736 reviews from Spring 2015's offering
Course slides from CS 443 @ Northwestern
U of Waterloo
Stanford
163KB
16-SchedAct+ResourceContainers.pptx
CS 736 course slides on Scheduler Activations and Resource Containers
Things that happen on an I/O request/completion. T1: Add two processors, user-level library picks two threads. T2: Thread 1 on SA A blocks in kernel, notified of that with a new SA C, library picks to run T3 on SA C. T3: Thread 1 finishes I/O, for the kernel to notify the user-level, take SA from B and use SA D to tell library both 1 and 2 can continue. T4: Use SA D to run t1.