.NET 11 Numerics
Tuesday, November 10
– PST
Tanner Gooding
.NET 11 expands the numerics toolbox with IEEE 754 decimal floating-point types, `BFloat16`, generic complex numbers, and new generic random-number APIs. It also brings performance and correctness improvements to `System.Numerics.Tensors`. We'll explore how the new types fit into generic math, choose the right precision and range, and see how layout-aware tensor operations improve performance while handling shapes, overlapping memory, and special values consistently. Along the way, we'll cover rounding and practical parsing and formatting improvements. We'll finish with a brief "and more is coming" look at .NET 12 tensor work, clearly distinguishing what's in .NET 11 from what comes next.
Resources
- https://learn.microsoft.com/dotnet/standard/generics/math
- https://github.com/dotnet/runtime/issues/81376
- https://github.com/dotnet/runtime/pull/98643
- https://github.com/dotnet/runtime/pull/130414
- https://github.com/dotnet/runtime/pull/127462
- https://github.com/dotnet/runtime/pull/135281
- .NET 12 lookahead: further tensor optimizations
