THE STATE OF GO What’s New since Go 1.20 白明 Tony Bai 《Go语言精进之路》作者 tonybai.com博主 那段时间错过的Go版本 2022.3 Go 1.18 泛型落地 2022.8 Go 1.19 修订Go内存模型文档、支持Soft memory limit 2023.3 Go 1.20 引入Profile-Guided Optimizations

Analytics Spring 2020 2/25: State Management Vasiliki (Vasia) Kalavri vkalavri@bu.edu ## State in dataflow computations Any non-trivial streaming computation maintains state: • rolling aggregations jpg) ## Streaming state ## Unmanaged • No explicit state primitives - Users define state using arbitrary types - The system is unaware of which parts of an operator constitute state ## Managed • Explicit Explicit state primitives including state types and interfaces - The system is aware of state and can transparently checkpoint it, restore it, re-scale it ![Image](/uploads/documents/b/9/f/d/b9fd594d

## State of the UI: Leveraging Kubernetes Dashboard and Shaping its Future Spencer Sugarman UX Researcher, Google Dan Romlein UX Designer, Google ## Talk Goals 1. What Dashboard is and why you should support kind/feature lifecycle/frozen #3148 opened on Jul 16 by SerialVelocity Angular Migration: State Management kind/feature lifecycle/frozen #3133 opened on Jul 7 by austbot Support metrics API good

Elasticity policies and state migration Vasiliki (Vasia) Kalavri vkalavri@bu.edu Streaming applications are long-running • Workload will change • Conditions might change • State is accumulated over time terminate processes • Adjust dataflow channels and network connections • Re-partition and migrate state in a consistent manner • Block and unblock computations to ensure result correctness ## Automatic Re-configuration requires state migration with correctness guarantees. ## State migration ## State migration strategies ## • Stop-and-restart • halt the whole computation, take a state snapshot of all operators

.) } } ### Keyed vs. non-keyed windows Window operators can be applied on a keyed or a non-keyed stream: • Window operators on keyed windows are evaluated in parallel • Non-keyed windows are processed non-keyed DataStream). • A window function is applied on a WindowedStream (or AllWindowedStream) and processes the elements assigned to a window. ### Keyed vs. non-keyed windows // define a keyed window the window assigner .reduce/aggregate/process(...) // specify the window function // define a non-keyed window-all operator stream .windowAll(...) // specify the window assigner .reduce/aggregate/process(

number of expected features in the input x • hidden_size – The number of features in the hidden state h - num_layers – Number of recurrent layers. E.g., setting num_layers=2 would mean stacking two RNNs number of expected features in the input x • hidden_size – The number of features in the hidden state h - num_layers - Number of recurrent layers. E.g., setting num_layers=2 would mean stacking two RNNs

cf30310bdd99235d1d629/p5_1.jpg) ## Passive Standby • Each primary periodically checkpoints its state and sends it to the secondary  ## Snapshotting Protocols A snapshot algorithm attempts to capture a coherent global state of a distributed system We need to retrieve a distributed cut in a system execution that yields a

Client 1 Local Game State  Client 2 Local Game State ## Deterministic simulation Initial state GameSystems::Tick() deb25e9aa12f06fc0/p14_2.jpg) GameSystems::Tick() on Desktop PC Intel Core i7-8700K New state New state With a deterministic simulation, we mostly need to send player inputs! ## Deterministic lock-step [Image](/uploads/documents/8/e/8/b/8e8bf3c37670c50deb25e9aa12f06fc0/p20_3.jpg) New input data Misprediction!!! State is dirty!  ## Doing

Side Effects Transaction inspect Model State observe ## Level of Detail – Time quanta and synchronization How often do we really care to know the state of the model? ■ Consider a Public Key Cryptography Cryptography Accelerator What happens in HW: Registers Inputs Complex State Machine Doing Things like a Montgomery Ladder Result Registers What we need to see: Registers Inputs openssl call and workloads In the modeling context, coroutines... • Provide syntactic sugar for things like state machines • Look more like RTL • Handle parallelism in a way that isn’t clunky - Scale across large