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) - Complex

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 Model

CROSS-PLATFORM PITFALLS AND HOW TO AVOID THEM Erika Sweet (she/her) Microsoft C++ TeamWelcome to CppCon 2020! Visit our table at the Expo Hall https://aka.ms/cppcon/expo • Meet the Microsoft C++ Abdicating responsibility to CMake Reimplementing functionality that CMake has built-in to abstract platform and compiler specificsUse built-in CMake command line tools…. execute_process(COMMAND ${CMAKE_COMMAND} rget PUBLIC cxx_std_11) …instead of manually setting flags via CMAKE_CXX_FLAGSKeep your paths platform independent…. target_include_directories(myTarget PUBLIC $

CppCon 2024 sixit:: 5 ✅ Same Executable Cross Platform Different Contexts ❌ ❌ 1 (Sort of) 2 3 General Sanity, Testability Distributed Simulations, Games, especially RTS Algorithm correctness using the identically compiled files and run on systems that adhere to the IEEE standards. Cross platform synchronized network simulations or replays will not possible [sic].” - Most upvoted answer on

career in video games and embedded software • Started using C++ in 1995 • First cross-platform project in 1994Cross-Platform Architecture Goals • Take advantage of all platforms • Focus on the compiler • Minimize classes, an illustrative example from a larger project • Project build issues • Inclusion of platform-specific header files • Concept hierarchies • Class and Function DesignOCP: The Open–Closed Principle Adding new revisions to a feature has no effect on previously-implemented revisions.What is a Platform? • A specific set of features • A feature is an abstract unit of functionality requiring implementations

full-text, and vector data. Built-in indexing and search to make data instantly searchable across distributed systems. High-performance querying for analytics, search, and AI workloads at scale. SQL simplicity simplicity to unify access across divers data types, reducing complexity in querying distributed datasets. Horizontal scalability across hybrid environments, supporting cloud, on- prem, and edge deployments resilience as business needs evolve. By unifying diverse data workloads into a single, scalable platform, a Real-Time UDL helps businesses increase efficiency, enhance decision-making, and power AI-driven

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327 26 Multi-processing and Distributed Computing 329 26.1 Code Availability and Loading Packages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1301 68 Delimited Files 1332 69 Distributed Computing 1337 69.1 Cluster Manager Interface . . . . . . . . . . . . . . . . . . . . . . . . need to vectorize code for performance; devectorized code is fast • Designed for parallelism and distributed computation • Lightweight "green" threading (coroutines) • Unobtrusive yet powerful type system

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327 26 Multi-processing and Distributed Computing 329 26.1 Code Availability and Loading Packages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1301 68 Delimited Files 1332 69 Distributed Computing 1337 69.1 Cluster Manager Interface . . . . . . . . . . . . . . . . . . . . . . . . need to vectorize code for performance; devectorized code is fast • Designed for parallelism and distributed computation • Lightweight "green" threading (coroutines) • Unobtrusive yet powerful type system

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327 26 Multi-processing and Distributed Computing 329 26.1 Code Availability and Loading Packages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1301 68 Delimited Files 1332 69 Distributed Computing 1337 69.1 Cluster Manager Interface . . . . . . . . . . . . . . . . . . . . . . . . need to vectorize code for performance; devectorized code is fast • Designed for parallelism and distributed computation • Lightweight "green" threading (coroutines) • Unobtrusive yet powerful type system

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331 26 Multi-processing and Distributed Computing 332 26.1 Code Availability and Loading Packages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1354 69 Delimited Files 1385 70 Distributed Computing 1390 70.1 Cluster Manager Interface . . . . . . . . . . . . . . . . . . . . . . . . need to vectorize code for performance; devectorized code is fast • Designed for parallelism and distributed computation • Lightweight "green" threading (coroutines) • Unobtrusive yet powerful type system