Performance tricks I learned from contributing to the Azure .NET SDK - Daniel Marbach - NDC Porto

Discover performance tricks from contributing to the Azure.NET SDK, including array pooling, stack allocation, and concurrent collections for efficient memory management and parallel processing.

Key takeaways
  • Be aware of closure allocations and reuse buffers for small and local buffers.
  • Use the array pool for allocating small arrays.
  • Use stack allocation for small buffers.
  • Avoid unnecessary boxing.
  • Use concrete collection types instead of generic ones.
  • Consider using Try statements to handle exceptions.
  • Use the Stack class to allocate and manage memory on the stack.
  • Use memory-mapped files for large data sets.
  • Use the Span class to manipulate memory instead of arrays.
  • Avoid using foreach loops for large collections.
  • Use parallel loops for parallel processing.
  • Use Task.Run to run tasks asynchronously.
  • Use CancellationToken to cancel tasks.
  • Use SemaphoreSlim to limit concurrent access to resources.
  • Use ReaderWriterLockSlim to synchronize access to resources.
  • Use ConcurrentQueue<T> to synchronize access to queues.
  • Use ConcurrentDictionary<T> to synchronize access to dictionaries.
  • Use Task to run tasks asynchronously.
  • Use Semaphore to limit concurrent access to resources.
  • Use Mutex to synchronize access to resources.
  • Use Monitor to synchronize access to resources.
  • Use ReaderWriterLock to synchronize access to resources.
  • Use ConcurrentBag<T> to synchronize access to bags.
  • Use ConcurrentStack<T> to synchronize access to stacks.
  • Use ConcurrentQueue<T> to synchronize access to queues.
  • Use ConcurrentDictionary<T> to synchronize access to dictionaries.
  • Use TaskCompletionSource<T> to create tasks that can complete with data.
  • Use CancellationToken to cancel tasks.
  • Use Task cancellation token to cancel tasks.
  • Use TaskCreationOptions to specify options for creating tasks.
  • Use TaskScheduler to customize the scheduling of tasks.

More talks

Pure Java Enterprise AI/LLM Integration (EAI 2.0) by Adam Bien

Mostly Harmless Fixed Effects Regression in Python with PyFixest [PyCon DE & PyData Berlin 2024]

Asynchronous PHP In Examples #slideless | Denys Bulakh

Lessons From The Pit Lane • Marc Priestley • GOTO 2024

Project Leyden: Improving Java’s Startup Time by Per Minborg, Sébastien Deleuze

Nicolò Giso - From telemetry data to CSVs with Python, Spark and Azure Databricks

Reviewing NuGet Packages security easily using OpenSSF Scorecard - Niels Tanis - NDC Oslo 2024

What Timescale's Approach to Dropping Chunks Tells Us About Concurrent DDL: Christopher Travers