Distributed Ranges: A Model for Building Distributed Data Structures, Algorithms, and Views
GPU Tile 1 Tile 0 Xe LinkProject Goals - Offer high-level, standard C++ distributed data structures - Support distributed algorithms - Achieve high performance for both multi-GPU, NUMA, and multi-node reduce(par_unseq, z, 0, std::plus()); }Outline - Background (Ranges, Parallelism, Distributed Data Structures) - Distributed Ranges (Concepts) - Implementation (Algorithms and views) - Complex sparse matrices) - Lessons learnedOutline - Background (Ranges, Parallelism, Distributed Data Structures) - Distributed Ranges (Concepts) - Implementation (Algorithms and views) - Complex0 码力 | 127 页 | 2.06 MB | 5 月前3A Simple Rollback System in C++
BeyondWe still need to wait for all the player inputs to process the confirm frame…What is a rollback system?Input predictionMispredictionDoing a rollbackRollback Implementation DetailsPhysics vs Graphics game update. This allows to sequence your game frame.A Fixed TickRollback Manager InputsRollbackGame System Architecture Game systems need a function to rollback (revert to a previous state). Typically everywhere Player Character as a gameplay, physics, graphics and audio object…ECS-like architecture System and array of Components (C-struct without functionality, except operators maybe) You want to have0 码力 | 58 页 | 6.61 MB | 5 月前3Combining Co-Routines and Functions into a Job System
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 for Functions into a Job System - CppCon 2021 3 / 39 Creating Game Engines with C++ • Vienna Game Job System + • Graphics API Abstraction Layer + • Vienna Entity Component System + Vienna Type List Library 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()0 码力 | 39 页 | 1.23 MB | 5 月前3POCOAS in C++: A Portable Abstraction for Distributed Data Structures
program for a supercomputer? Introduce PGAS Model, RDMA Building Remote Pointer Types Building Distributed Data Structures Extending to GPUsThis Talk Background: how do we write a program for a supercomputer supercomputer? Introduce PGAS Model, RDMA Building Remote Pointer Types Building Distributed Data Structures Extending to GPUsThis Talk Background: how do we write a program for a supercomputer? Introduce Introduce PGAS Model, RDMA Building Remote Pointer Types Building Distributed Data Structures Extending to GPUsThis Talk Background: how do we write a program for a supercomputer? Introduce PGAS Model0 码力 | 128 页 | 2.03 MB | 5 月前3Building a Coroutine-Based Job System Without Standard Library
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 resources: https://gist.github.com/MattPD/9b55db49537a90545a90447392ad3aeb Before we dive into the system, I want to take 10 mins to have a quick recap about C++ coroutine and share some gotcha moments pretty overwhelmed by all different terms, not to mention to put them up together to form a knowledge system. I hope the following information can be helpful and I attached some links in the slides, or pointers0 码力 | 120 页 | 2.20 MB | 5 月前3Behavior-driven Tests for Microservices-based Algo Trading System
from Columbia University. Title : Behavior-driven Tests for Microservices-based Algo Trading System Summary : Introduction : This poster will introduce a test framework we use at Bloomberg to to enable behavior-based tests in natural language style for a microservices-based algo trading system that our engineering team is responsible for. Relevance : For our electronic trading business Although microservice components can easily be unit tested, it is often non-trivial to automate the system-level tests with external dependencies. Controlling external dependencies are especially important0 码力 | 1 页 | 65.24 KB | 5 月前3Make Successor Build Systems: World Tour of Build Systems
build systemsLearning from other build-systemsParallelism How many independent targets can a build system run concurrently at most?Parallelism Level 1 - Gradle . ├── app │ ... │ └── build.gradle ├── robust and easy to configure toolchains out of the box. This way builds do not normally rely on any system-wide installed tools/libraries that can change between different workstationsReproducibility Level buildsGit + Ninja + cache packs ✅ __FILE__ problem solved 🫡Caching How many targets can a build system avoid rebuilding?Caching Level 1 - Gradle Once again the low level of granularity means that even0 码力 | 115 页 | 7.02 MB | 5 月前3Six Impossible Things
overrun occurred — crashing the entire network and causing the ship to lose control of its propulsion system. wired.com/1998/07/sunk-by-windows-ntThis is the monstrosity in love, lady, that the will is infinite shattered the dream, showing that for any consistent axiomatic system there will always be theorems that cannot be proven within the system. Adrian Colyer blog.acolyer.org/2020/02/03/measure-mismeasure-fairness/All within the system – i.e. from properties of the model itself and perhaps the data it is trained on. To show that a machine learning model is fair, you need information from outside of the system. Adrian0 码力 | 144 页 | 6.47 MB | 5 月前3From Eager Futures/Promises to Lazy Continuations: Evolving an Actor Library Based on Lessons Learned from Large-Scale Deployments
type erasurekey challenges we’ve faced building distributed systems with good performance and correctnesskey challenges we’ve faced building distributed systems with good performance and correctnesschallenges “messages” to other actors is non-blocking (no waiting!) ● same programming model whether local or distributed“actors” in C++ many actor libraries are based on low-level message-passing “send/receive” struct will often need to be shared and/or moved between execution resources (i.e., cores) but for distributed/network services, the data is often only shared with other machines and bouncing the data unnecessarily0 码力 | 264 页 | 588.96 KB | 5 月前3Design and Implementation of Highly Scalable Quantifiable Data Structures in C++
(Safety) Conditions Motivating Examples, Drivers for Change Quantifiability Definition Vector Space System Model Entropy Measurement Correctness and Performance Engineering Case Study: k-FIFO Queue Design conserved ▶ Timing is critical to correctness yet can only be estimated by applications ▶ Intermediate system state is opaque (i.e. denotational, not small step operational) ▶ Correctness is binary, lacking intractable O(n!) complexity of concurrent system models prompting the search for reductions ▶ Growth of distributed software applications: blockchain, distributed file systems, network apps Design and Implementation0 码力 | 51 页 | 4.08 MB | 5 月前3
共 244 条
- 1
- 2
- 3
- 4
- 5
- 6
- 25
相关搜索词
DistributedRangesModelforBuildingDataStructuresAlgorithmsandViewsSimpleRollbackSysteminC++CombiningCoRoutinesFunctionsintoJobPOCOASPortableAbstractionCoroutineBasedWithoutStandardLibraryBehaviordrivenTestsMicroservicesbasedAlgoTradingMakeSuccessorBuildSystemsWorldTourofSixImpossibleThingsFromEagerFuturesPromisestoLazyContinuationsEvolvinganActoronLessonsLearnedfromLargeScaleDeploymentsDesignImplementationHighlyScalableQuantifiable