From Functions to Coroutines 40 Years Evolution Rainer Grimm Training, Coaching, and Technology Consulting www.ModernesCpp.netEvolution of Callable template T invoke(Func

2/55 Deciphering Coroutines - Part 2 Mastering Asynchronous Control Flow Andreas Weis CppCon 20243/55 About me - Andreas Weis (he/him) ComicSansMS cpp@andreas-weis.net Co-organizer of the Munich best-practice talk.16/55 A mental model for coroutines: Cooperative Threads void spawn_task () { // ... Result r = outer_function (); }16/55 A mental model for coroutines: Cooperative Threads void spawn_task mental model for coroutines: Cooperative Threads PartialResult middle_function () { auto r = inner_function (); return PartialResult :: from_io_result (r); }16/55 A mental model for coroutines: Cooperative

Challenges when onboarding coroutines A lot of existing code mostly callback-based A custom-built I/O event loop predates Asio by a decade fine-tuned for specific use cases Somewhat conservative conservative audience real benefit needs to be demonstrated to justify switchingCallbacks vs coroutines // read length auto len = make_unique(); async_read(socket, buffer(len.get(), sizeof(int32_t)) use_awaitable); // ...handle data... Local variable Manual lifetime managementCallbacks vs coroutines // read length auto len = make_unique(); async_read(socket, buffer(len.get(), sizeof(int32_t))

performance • Easiser operation • Cloud native • Curve is composed of two parts • Curve Block Storage (CurveBS) • CurveBS: a high performance cloud native distributed block storage • Curve File File System (CurveFS) • CurveFS: a high performance cloud native file system (Based on CurveBS / S3 compatible Storage)Operator capability Level OPERATOR CAPABILITY LEVEL CURVE COMMENT BASIC INSTALL Y Support horizontal/vertical scaling, auto config tuning, abnormal detection, schedule tuningCloud native Feature list • Features for Cluster • Features for CurveBS • Features for CurveFS • Features

Jeffrey E. Erickson cppcon 2023 1cppcon 2023 Behavioral Modeling in HW/SW Co-design using C++ coroutines Jeffrey E Erickson, Ph.D. Sebastian Schönberg, Ph.D.© 2023 Intel Corporation and Jeffrey E. Erickson Jeffrey E. Erickson cppcon 2023 8 Coroutines are an enabling language feature • To date, coroutines have been about execution and workloads • In the modeling context, coroutines… • Provide syntactic sugar Handle parallelism in a way that isn’t clunky • Scale across large and complex HW interactions But, coroutines lack comprehensive standard library support • We used concurrencppIntel Confidential Department

knowledge Francesco Zoffoli - Coroutine Patterns and How to Use Them - CppCon 2023 5Motivation Coroutines introduce new paradigm Francesco Zoffoli - Coroutine Patterns and How to Use Them - CppCon 2023 CppCon 2023 11Overview – Executor • Executes synchronous functions • Could be multi-threaded • Coroutines are split in synchronous functions Francesco Zoffoli - Coroutine Patterns and How to Use Them CppCon 2023 19 [1] https://ericniebler.com/2020/11/08/structured-concurrency/Lifetime Member coroutines implicitly capture this Francesco Zoffoli - Coroutine Patterns and How to Use Them - CppCon

networking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 6.5 Coroutines and concurrency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 http_client.fetch(url) return response.body And here is the same function rewritten asynchronously as a native coroutine: from tornado.httpclient import AsyncHTTPClient async def asynchronous_fetch(url): http_client http_client = AsyncHTTPClient() response = yield http_client.fetch(url) raise gen.Return(response.body) Coroutines are a little magical, but what they do internally is something like this: from tornado.concurrent

org/projects/tornado/downloads/]. User’s guide Introduction Asynchronous and non-Blocking I/O Coroutines Queue example - a concurrent web spider Structure of a Tornado web application Templates and UI IOStream-based TCP server Coroutines and concurrency tornado.gen — Generator-based coroutines tornado.locks – Synchronization primitives tornado.queues – Queues for coroutines tornado.process — Utilities Introduction Asynchronous and non-Blocking I/O Blocking Asynchronous Examples Coroutines Native vs decorated coroutines How it works How to call a coroutine Coroutine patterns Calling blocking functions

fminimum_num, fmaximum_num: Treat NaNs as missing data, -0 < +0. Hardware Support ● Many GPUs have native atomic float min/max ● CPU support growing (e.g. ARM v8.1) ● Hardware treats -0 < +0 and NaN as fetch_fmaximum_num(float, memory_order = memory_order::seq_cst) noexcept;Performance Considerations Native atomic operations vs. CAS loops. Significant performance difference in highly concurrent systems systems Careful consideration of corner cases: NaNs and signed zeros Performance implications: Native atomic operations vs. CAS loops New atomic float min/max operations Semantics close to hardware