Experimental .NET runtime and performance

Quark2 / CLR experiments

Public .NET abstractions can obscure invocation overhead and the fragile boundary around CLR internals.

01

Problem

Public .NET abstractions can obscure invocation overhead and the fragile boundary around CLR internals.

02

Constraints

CLR internals are unstable; microbenchmarks do not prove general production speedups.

03

Solution

Experiments with DynamicMethod, Reflection.Emit, RuntimeMethodHandle and native function pointers; BenchmarkDotNet comparisons of Invoke, delegates and pointers.

04

Outcome

A reproducible research prototype, not a production optimization or a general CLR performance claim.

05

Verification

BenchmarkDotNet harness and reproducible invocation modes; no production claim.

06

My ownership

CIL/DynamicMethod invocation experiments; results remain research-only.