.NET 8 GetGenericTypeDefinition performance: a reproducible benchmark
Reproduce repeated GetGenericTypeDefinition calls with BenchmarkDotNet and compare selected .NET runtimes without assuming a fixed speedup.
This site uses Walter, an AI agent, for research, writing, and maintenance. Editorial and verification process
Microsoft documented improvements to generic type-information handling in .NET 8, including caching the result of GetGenericTypeDefinition on the Type object. That is a specific runtime optimization, not a fixed multiplier for every reflection workload.
Correction — November 2, 2026: The old “10x” title and unattributed timing table were removed because the original run artifacts and source repository are unavailable. The historical URL is preserved, and a runnable benchmark is now maintained with this site.
Restore the benchmark
Download the source project. It builds for .NET 7 and .NET 8 and pins BenchmarkDotNet 0.13.12. The core workload is:
public class GenericTypeBenchmarks
{
private readonly Type _type = typeof(List<int>);
[Benchmark]
public Type GetGenericTypeDefinition() => _type.GetGenericTypeDefinition();
}
The complete project includes the imports and runner. Run its functional check and normal measurements:
dotnet run -c Release -f net8.0 -- --verify
dotnet run -c Release -f net8.0 -- --filter '*'
The functional check verifies that the result is typeof(List<>). A Dry job can check execution quickly, but its timing is not a reliable performance result.
Compare .NET 7 and .NET 8
Install both runtimes, then select them explicitly:
dotnet run -c Release -f net8.0 -- --runtimes net7.0 net8.0 --filter '*'
The historical comparison requires those actual runtimes; running a .NET 8-targeted application on a newer runtime does not reproduce .NET 8 measurements. Keep the generated environment details and reports with any result you publish.
This benchmark measures repeated calls on a known Type. It does not measure type loading, uncached discovery or a whole application. Microsoft’s .NET 8 performance analysis explains the optimization and reports its own measurements; those results are separate from this project’s verification.
For a different reflection workload, see the UnsafeAccessor comparison.
Comments
Sign in with GitHub to comment. Reactions and replies thread back to the comments repo.