> Markdown version of [/videos/100138-is-it-still-c-practical-systems-programming-with-net-war-stories-included](https://www.wearedevelopers.com/videos/100138-is-it-still-c-practical-systems-programming-with-net-war-stories-included). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Is it still C#? Practical systems programming with .NET (war stories included)

Think C# is strictly for enterprise applications? Discover how pushing modern .NET to its limits with extreme memory manipulation and vectorization rivals traditional low-level systems programming.

- **Speakers:** [Szymon Kulec](https://www.wearedevelopers.com/@szymon-kulec)
- **Event:** World Congress 2026 Europe
- **Published:** July 9, 2026
- **Duration:** 24:17
- **URL:** https://www.wearedevelopers.com/videos/100138-is-it-still-c-practical-systems-programming-with-net-war-stories-included

## Summary

Modern C# has evolved far beyond its enterprise application roots, unlocking practical systems programming capabilities that rival traditionally low-level languages. Drawing from RavenDB’s experience building a high-performance database entirely in .NET, the technical narrative reveals how pushing the runtime to its limits demands a blend of deliberate memory manipulation and advanced compilation techniques. By exploring these raw edges, developers can optimize demanding services using features previously unavailable in the managed ecosystem, ensuring critical infrastructure scales efficiently.

Extracting maximum performance fundamentally relies on preventing runtime overhead, adhering to the database engineering mantra that "fast folks do not allocate." Leveraging C#'s static virtual members within interfaces allows the JIT compiler to aggressively inline logic—yielding a minimal assembly footprint, particularly when paired with structs. Similarly, bridging synchronous and asynchronous API boundaries requires taming the runtime's state machine. Utilizing `ValueTask` components allows applications to short-circuit entirely to zero-allocation synchronous returns, bypassing the heavy async state machinery when requested data already resides in memory. For sheer computational workloads, vectorization eliminates cross-lane data dependencies. By employing the intrinsic .NET Base Class Library (BCL) vector APIs, compute-heavy operations like XXHash64 validation or cosine similarity math inject CPU-level instruction sets directly into the unrolled execution path to yield double-digit performance gains.

The push for system-level control extends downward to how the framework treats unmanaged memory constraints. Recognizing `ref` locals as managed pointers gives developers the power to traverse memory spans and arrays directly without triggering garbage collection penalties. This low-level memory choreography directly empowers innovations like 'blittable JSON', a custom storage architecture that replaces repetitive string properties with an integer-based dictionary mapped in a single traversal pass. Supported by strict memory pooling and footer-based lookup tables, this technique grants O(1) property access and dictionary compression—proving developers can bypass costly serialization trees and build strictly high-performance .NET tools.

**Keywords:** practical systems programming, .NET runtime performance, c# static virtual members, jit compiler inlining, valuetask struct optimization, zero-allocation execution, simd vectorization, intrinsic bcl vector apis, xxhash64 algorithm optimization, cosine similarity algorithms, ref keyword memory semantics, managed pointer traversal, blittable json architecture, json dictionary compression, ravendb memory pooling

## Chapters

1. **Introduction to systems programming and performance in .NET** (00:02) — Setting a baseline for performance measurements and low-level code expectations in C#.
1. **Leveraging static virtual members to maximize JIT inlining** (01:58) — Using static abstract interface members on structs enables the JIT compiler to eliminate runtime abstractions.
1. **Optimizing mixed asynchronous execution paths with value tasks** (06:53) — Wrapping code with value tasks avoids state machine overhead for operations that complete synchronously.
1. **Implementing high-speed vector operations for independent data hashing** (12:14) — Eliminating cross-lane dependencies lets developers natively unroll SIMD loops to dramatically accelerate non-cryptographic hashes.
1. **Modifying high-performance data directly via local ref returns** (16:45) — Utilizing explicit memory references and span structures reduces allocations when interacting with intensive data arrays.
1. **Building robust blittable schema formats for fast serialization** (19:49) — Encoding dictionary compression and property lookup footers avoids full document traversal during database index updates.

## Related Moments

- [Discussion on JIT memory parameters and cross-language benchmarking](https://www.wearedevelopers.com/videos/240-just-in-time-compilation-in-jvm) (from "Just-in-time Compilation in JVM")
- [Addressing technical nuances during the audience Q&A](https://www.wearedevelopers.com/videos/244-microservices-with-micronaut) (from "Microservices with Micronaut")
- [Answering audience questions on container functionality and local workflows](https://www.wearedevelopers.com/videos/166-local-development-techniques-with-kubernetes) (from "Local Development Techniques with Kubernetes")
- [Final performance results and key architectural takeaways](https://www.wearedevelopers.com/videos/733-accelerating-authentication-architecture-taking-passwordless-to-the-next-level) (from "Accelerating Authentication Architecture: Taking Passwordless to the Next Level")
- [Analyzing web component structures and network performance profiles](https://www.wearedevelopers.com/videos/699-blazor-the-new-spa-framework-for-everyone-who-hates-javascript) (from "Blazor – the new SPA Framework for everyone who hates JavaScript")
- [Optimizing performance across JSON methodologies and backend stacks](https://www.wearedevelopers.com/videos/33-api-design-getting-started) (from "API Design - Getting Started")

## Related Articles

- [Dev Digest 138 - Are you secure about this?](https://www.wearedevelopers.com/magazine/486-dev-digest-138-are-you-secure-about-this)
- [Dev Digest 136 - No JS(on) of mine](https://www.wearedevelopers.com/magazine/482-dev-digest-136-no-js-on-of-mine)
- [Dev Digest 132 - Binging WADFlix?](https://www.wearedevelopers.com/magazine/473-dev-digest-132-binging-wadflix)
- [Dev Digest 120 - Apple and peers](https://www.wearedevelopers.com/magazine/455-dev-digest-120-apple-and-peers)

## Related Jobs

- [Principal Software Engineer, Database Infrastructure](https://www.wearedevelopers.com/jobs/ext/1465908-principal-software-engineer-database-infrastructure) at **GitHub**
- [Senior Software Engineer,Billing](https://www.wearedevelopers.com/jobs/ext/1991843-senior-software-engineer-billing) at **GitHub**
- [Principal Engineer - AI Search & Vector Infrastructure](https://www.wearedevelopers.com/jobs/ext/381484-principal-engineer-ai-search-vector-infrastructure) at **Redis**
- [Senior Software Engineer](https://www.wearedevelopers.com/jobs/ext/15942-senior-software-engineer) at **GitHub**
- [Principal Software Engineer, Enterprise AI Platform](https://www.wearedevelopers.com/jobs/ext/1467292-principal-software-engineer-enterprise-ai-platform) at **GitHub**
- [Principal Engineer - AI Search & Vector Infrastructure](https://www.wearedevelopers.com/jobs/ext/353953-principal-engineer-ai-search-vector-infrastructure) at **Redis**