Add inline GEMM optimizations and general performance improvements#226

Merged

sdatkinson merged 16 commits intosdatkinson:mainfrom

jfsantos:feature/inline-gemm

Feb 23, 2026

Commits on Feb 6, 2026

Add inline GEMM optimizations and general performance improvements

João Felipe Santos
and
claude
committed

Commits on Feb 12, 2026

Fixed minor bug when building without inline GEMM
João Felipe Santos
committed

Commits on Feb 13, 2026

Added NAM_INLINE_GEMM build/test task to CI
João Felipe Santos
committed

Commits on Feb 16, 2026

Remove redundant manual loop unrolling from activations and element-wise ops
João Felipe Santos
committed
Added option to disable fasttanh to benchmodel and benchmodel_bufsize
João Felipe Santos
committed

Commits on Feb 18, 2026

Fixed bugs and added tests to ensure operations work on non-contiguous matrices
João Felipe Santos
committed

Commits on Feb 19, 2026

Merge branch 'main' into feature/inline-gemm
jfsantos
authored

Commits on Feb 23, 2026

Merging to main, updated CI build to test both Eigen and inline paths
João Felipe Santos
committed
Fixed minor bug when building without inline GEMM
João Felipe Santos
committed
Remove redundant manual loop unrolling from activations and element-wise ops
João Felipe Santos
committed
Added option to disable fasttanh to benchmodel and benchmodel_bufsize
João Felipe Santos
committed
Fixed bugs and added tests to ensure operations work on non-contiguous matrices
João Felipe Santos
committed
Merge branch 'feature/inline-gemm' of github.com:jfsantos/NeuralAmpModelerCore into feature/inline-gemm
João Felipe Santos
committed
Fixed build for inline path
João Felipe Santos
committed
Fixed build for inline path
João Felipe Santos
committed
Merge branch 'feature/inline-gemm' of github.com:jfsantos/NeuralAmpModelerCore into feature/inline-gemm
João Felipe Santos
committed