Solve Garbage Collection in C# for HFT

Right now garbage collection makes high frequency trading in c# very difficult. Find a way to fix it please

A thorough written survey of why .NET garbage collection causes latency spikes in HFT contexts: generational GC mechanics, LOH behavior, Server vs Workstation GC, background/concurrent GC tradeoffs, sustained-low-latency mode, GC regions in .NET 7+, and real measured pause characteristics. Includes a catalog of allocation sources in typical trading code (LINQ, closures, boxing, string handling, async state machines, collection resizing) and a comparison of mitigation strategies used in production low-latency .NET shops. Delivered as a structured technical report with annotated references and decision matrices.

A complete engineering design document for a zero-allocation hot path: memory ownership model, object pooling strategy, struct-based message types, Span<T>/Memory<T> usage rules, pre-allocated ring buffers for market data and order flow, unmanaged memory arenas, thread/core pinning, and a 'no allocations after warmup' contract with enforcement strategy (EventPipe allocation tracking, CI gates). Includes API specifications, threading model diagrams (described textually), failure-mode analysis, and a migration guide for retrofitting existing C# trading codebases.

Working C# source code for the core library: lock-free SPSC/MPSC ring buffers, object pools with thread-local caching, a bump-pointer unmanaged arena allocator with reset semantics, allocation-free fixed-point decimal type for prices, zero-alloc UTF-8 message parsing/formatting (FIX-style tag/value codec), pooled buffer writer, and allocation-free logging facade. Fully documented with XML comments, unit tests using NUnit, and an analyzer-style allocation audit checklist. Delivered as a complete buildable solution.

A BenchmarkDotNet-based benchmark suite measuring allocation rates, p50/p99/p99.9 latency, and GC pause counts for the ZeroAlloc library versus idiomatic C# equivalents; an EventPipe/ETW-based runtime monitor that asserts zero Gen0 collections after warmup; plus a written GC tuning playbook covering runtimeconfig.json settings (TieredPGO, ReadyToRun, GC modes, heap limits), OS-level tuning (core isolation, huge pages on Windows/Linux), and a methodology for validating latency budgets in CI. Delivered as runnable code plus a detailed report template with example interpreted results.

An end-to-end demo: a simulated market data feed handler, allocation-free order book (price-level ladder over pre-allocated arrays), simple market-making strategy, and order gateway, all built on the ZeroAlloc library and demonstrating sustained zero-GC operation under load with a replayable synthetic tick dataset generator. Accompanied by a polished adoption guide: step-by-step tutorial, architecture walkthrough, common pitfalls, and a roadmap for advanced topics (NativeAOT, kernel-bypass networking integration points). Delivered as buildable code plus long-form documentation suitable for publication.