Issues

• team::with_team: pool size, participant wording, and one-off use in the docs

  tensor4all/faer#149 · Filter by author shinaoka opened on Oct 29, 2026
  

• sparse LU: tolerance-based numeric singularity detection (follow-up to #132)

  tensor4all/faer#146 · Filter by author shinaoka opened on Oct 29, 2026
  

• Small-matrix fast paths (n <= 4) selected by size inside the existing API

  tensor4all/faer#142 · Filter by author shinaoka opened on Oct 29, 2026
  1

  

• Unitary / normal eigendecomposition with a unitary eigenvector matrix, robust to clustered eigenvalues

  tensor4all/faer#141 · Filter by author shinaoka opened on Oct 29, 2026
  

• Deterministic simultaneous diagonalization of a commuting real symmetric pair (complex symmetric unitary), for degenerate spectra

  tensor4all/faer#140 · Filter by author shinaoka opened on Oct 29, 2026
  

• bicgstab: a breakdown-specific error variant (upstream #334, feature)

  tensor4all/faer#135 · Filter by author shinaoka opened on Oct 29, 2026
  

• Port primal factorizations from MatrixFactorizations.jl (QL/RQ/LQ, polar)

  enhancement
  tensor4all/faer#130 · Filter by author shinaoka opened on Oct 29, 2026
  2

• 部分ピボット LU が 8T 以上で伸びない: 現行構造のボトルネックと並列向き手法の文献調査

  tensor4all/faer#123 · Filter by author shinaoka opened on Oct 29, 2026
  

• team route: some processes run every repetition at the 1-thread time (16T, broccoli)

  tensor4all/faer#107 · Filter by author shinaoka opened on Oct 29, 2026
  13

  

• faer-ffi: redesign the C API that returns Par by value, on threadteam/with_team

  tensor4all/faer#98 · Filter by author shinaoka opened on Oct 28, 2026
  

• fork の並列化の高並列 (32〜64 スレッド) スケーラビリティ監査

  tensor4all/faer#95 · Filter by author shinaoka opened on Oct 28, 2026
  6

  

• cost ゲートなしで全幅を渡す経路: 三角積の x86 カーネル、K==1 の rank_update、GEMM のセクション数過少計上

  tensor4all/faer#94 · Filter by author shinaoka opened on Oct 28, 2026
  

• p 個の部分バッファを逐次合算する reduction (tridiag, matvec_colmajor, 疎×疎, compact-WY upgrade) の高並列スケーリング

  tensor4all/faer#93 · Filter by author shinaoka opened on Oct 28, 2026
  

• Block Householder apply re-reads the whole basis V once per column task; traffic grows with the thread count

  tensor4all/faer#90 · Filter by author shinaoka opened on Oct 28, 2026
  2

  

• Select AVX-512 at runtime without the nightly feature (pulp/gemm x86-v4)

  tensor4all/faer#86 · Filter by author shinaoka opened on Oct 28, 2026
  5

  

• perf(matmul): decide the unmerged tall-skinny column-slab branch against 0.25.0

  tensor4all/faer#71 · Filter by author shinaoka opened on Oct 27, 2026

• perf: f64 small-size residuals at 4T/8T after 0.25.0 (QR 64, solve 64, SVD 64xN)

  tensor4all/faer#70 · Filter by author shinaoka opened on Oct 27, 2026

• evd/svd: complex self_adjoint_evd and bidiagonal_svd (n >= 128) allocate on the heap despite caller-provided scratch

  tensor4all/faer#58 · Filter by author Ryo-wtnb11 opened on Oct 25, 2026
  1

• Run the matrix product through spindle on non-x86_64 targets (aarch64)

  tensor4all/faer#56 · Filter by author shinaoka opened on Oct 25, 2026
  2

• Plan nightly numerical validation from bundled LAPACK test suites

  tensor4all/faer#31 · Filter by author shinaoka opened on Oct 22, 2026

• Revert to upstream private-gemm-x86 when 0.1.23+ lands with the masked-store clamp

  tensor4all/faer#20 · Filter by author shinaoka opened on Oct 19, 2026
  1