← Back

StockSharp Technical Update: .NET 10, Async APIs, and 165 GPU Indicators

StockSharp Technical Update: .NET 10, Async APIs, and 165 GPU Indicators
StockSharp Technical Update: .NET 10, Async APIs, and 165 GPU Indicators

The StockSharp public codebase has recently gained .NET 10 support, an extensive migration of its core APIs to async, a decomposed strategy engine, and a GPU module with 165 indicator calculators.

The platform's main stack now runs on .NET 10. Key interfaces for market data storage, message adapters, providers, and other subsystems have been converted to asynchronous equivalents.

The changes also affect trading strategy execution. The previously monolithic Strategy class has been divided into more independent components, while tests verify functional parity with the former engine. This architecture makes individual parts easier to test, extend, and reuse.

Order execution mechanics have been moved into a standalone MatchingEngine. It supports regular and stop orders, including order modifications, while strategies can choose between server-side and local execution of protective orders.

The platform also includes the Algo.Gpu module for large-scale analysis. It provides 165 GPU calculators for technical indicators. The calculations are built on ILGPU and can use CUDA, OpenCL, or a CPU backend, automatically selecting a suitable device.

The GPU module is primarily intended for batch workloads such as:

  • strategy parameter optimization;
  • processing large numbers of time series;
  • historical analysis;
  • bulk indicator calculation;
  • trading system optimization.

Results produced by the GPU implementations are checked against StockSharp's standard CPU indicators. Actual performance gains depend on the dataset size, selected indicator, and hardware, so there is no universal acceleration factor.

The platform's source code and change history are available in the StockSharp repository.

Comments (0)

Login or Create account, Log in or register to leave a comment

No comments yet. Be the first!