Hlavacs – Combining Co-Routines and Functions into a Job System - CppCon 2021 1 / 39Helmut Hlavacs – Combining Co-Routines and Functions into a Job System - CppCon 2021 2 / 39 About Myself • Professor ComputingHelmut Hlavacs – Combining Co-Routines and Functions into a Job System - CppCon 2021 3 / 39 Creating Game Engines with C++ • Vienna Game Job System + • Graphics API Abstraction Layer + • Vienna Entity Engine 2.0 https://github.com/hlavacs 20Helmut Hlavacs – Combining Co-Routines and Functions into a Job System - CppCon 2021 4 / 39 The Game Loop auto prev = high_resolution_clock::now(); while( !finished()

Tianyi(Tanki) Zhang tankiistanki tankijong COROUTINE JOB SYSTEM WITHOUT STANDARD LIBRARY source code of the system: https://github.com/tankiJong/cpp-coroutine-job-system Hi Everyone, Tanki here. Thanks for coming We will see them many times in the rest of the talk, and they will have different names in the job system. 1314 14 template struct task; task sum(int a, int b) { int result = a customization. 4344 44 JOB SYSTEM • Scheduler + User defined workload • Optimize for CPU throughput Okay, finally, we are ready to talk about the job system! What is a job system? Job system is a kind of

. . . . . . . . . . . . . . . 5 4.2 Develop Job . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 4.3 Configure Job . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 4.4 Schedule Job . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 5 Concepts & Features 7 5.1 Schedule Model . . . . . . . . . . . . . . . . . . . . . . . . . . 17 5.5 Job Open Ecosystem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 5.5.1 Job interface . . . . . . . . . . . . . . . . . .

Resource Management: p Queue p Hierarchical queue p Fair-share p Preempt/Reclaim p … Ø Scheduler p Job preemption p Fair-share scheduling p Queue scheduling p Resource reservation p Binpack p Task pod executor pod apiVersion：v1 kind: Pod metadata: annotations: scheduling.k8s.io/group-name: job-1574739729783- podgroup volcano.sh/task-spec: spark-driver createTimestamp: “2019-11-27T09:33:19Z” 6cc54577d7254b2d84924500375112f7 spark-role: driver name: job-1574739729783-driver namespace: default resourceVersion: “12093805” selfLink: /api/v1/namespaces/default/pods/job- 1574739729783-driver uid: f26a81f3-10f

elasticjob.kernel.ui.Bootstrap，如何解决？ . . . . . . . . . 91 10.11 运行 Cloud Scheduler 持续输出日志“Elastic job: IP:PORT has leadership”，不能正常运行 91 10.12 在多网卡的情况下无法获取到合适的 IP . . . . . . . . . . . . . . . . . . CoordinatorRegistryCenter regCenter = new ZookeeperRegistryCenter(new ZookeeperConfiguration("zk_host:2181", "my-job")); regCenter.init(); return regCenter; } private static JobConfiguration createJobConfiguration() nter regCenter = new ZookeeperRegistryCenter(new ZookeeperConfiguration("zk_host:2181", "elastic-job-demo")); regCenter.init(); return regCenter; } private static JobConfiguration createJobConfiguration()

data for global illumination ▶ Process everything in parallel Speaker notesASSUMPTIONS • Built on Job System, with Async IO • Run on single PC with many processor cores • Fast & sizeable memory & storage Techniques for running on single computer - many threads - not about clusters Built for async processing - job system and async IO PC used for examples is from 4-5 years ago 16 physical cores - 32 logical cores features What GameDev is currently using Not all vendors support latest C++ Speaker notesTERMINOLOGY • Job: indivisible unit of work • Task: larger process converting input to output • SpinLock: lock that busy

'999') OR VALUE IS NULL) JOB_CODE (JOBCODE) VARCHAR(5) Not Null Chapter 17. Isql Commands 33 CHECK (VALUE > '99999') JOB_GRADE (JOBGRADE) SMALLINT Not Null CHECK (VALUE BETWEEN 0 AND 6) JOB_COUNTRY (COUNTRYNAME) VARCHAR(15) Not Null SALARY (SALARY) References DEPARTMENT (DEPT_NO) CONSTRAINT INTEG_29: Foreign key (JOB_CODE, JOB_GRADE, JOB_COUNTRY) References JOB (JOB_CODE, JOB_GRADE, JOB_COUNTRY) CONSTRAINT INTEG_27: Primary key (EMP_NO) CONSTRAINT

Volcano-controller Volcano-scheduler Kubernetes Volcano-controller Volcano-scheduler … Kube-apiserver Job/JobFlow Queue vc-controller vc-scheduler vsub kubectl Node NUMA GPU Node NUMA GPU … VolcanoGlobal VolcanoGlobal 架构 多中心 低成本 无绑定 VG Scheduler ETCD Karmada Controllers K8s API Server Queue Controller Job Controller VG Controller VG API Server Cluster A K8s API Server Cluster B Agent Other Clusters sh/v1alpha1 kind: Job metadata: name: mpi-job labels: "volcano.sh/job-type": "MPI" spec: # minimum number of pods need to be started minAvailable: 3 schedulerName: volcano plugins: # job level ssh trust

Resources Processes Human Resources Management can be divided into several distinct processes. • Job Planning and Definition • Recruitment • Candidate Selection and Hiring • Employee Training and Development Employee Evaluation and Performance Management • Employee Salary and Benefits Job Planning and Definition Generally job positions are authorised by a budget and fulfilled by people. The person fulfilling defined by a type of work. For example there may be 20 job positions in an organization for a administrative clerk. Each position is has a job definition and is authorised by a budget request from a

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185 12.3.1. Organization, Job Position and Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Run The JobShop is used to manage and track all the manufacturing or assembling work. The work (or job) carried out in the JobShop is called a Production Run. Production Runs can also be called Work Orders “Production Runs” are the to do or work lists of the manufacturing department. Each manufacturing job, batch, or run is done based on a production run. In OFBiz Jobshop you can • Create a new ProductionRun