All pull requests

Pull requests

• Avoid overflow in the 2x2 pivot inverse of the symmetric-indefinite factorizations (Reference-LAPACK PR 1377)

  #6175·martin-frbg opened Oct 25, 2026Collaborator·

  ·

  · 0.3.35

• RISC-V: faster DAMAX kernel for RISCV64_ZVL256B

  #6174·HiderX opened Oct 25, 2026·

  ·

• RISC-V: fix Inf in nrm2_rvv and use it for RISCV64_ZVL256B

  #6172·HiderX opened Oct 25, 2026·

  ·

• RISC-V: use the RVV DSDOT kernel for RISCV64_ZVL256B

  #6171·HiderX opened Oct 25, 2026·

  ·

• RISC-V: return the first occurrence on ties in the RVV index kernels

  #6170·HiderX opened Oct 25, 2026·

  ·

• RISC-V: fix ISMIN/IDMIN in the ZVL256B imin_vector kernel

  #6169·HiderX opened Oct 25, 2026·

  ·

• RISC-V: unroll the ZVL256B DGEMM kernel K-loop by two

  #6168·HiderX opened Oct 25, 2026·

  ·

• arm64: faster TRMM on VORTEXM4 (four-slice ZA stores, diagonal tiles, streaming copies)

  #6159·tesch1 opened Oct 22, 2026Contributor·

  ·

• arm64: SME2 complex level-3 kernels for VORTEXM4

  #6158·tesch1 opened Oct 22, 2026Contributor·

  ·

• arm64: SME GEMM kernels with packing buffers per thread (thread safety)

  #6157·tesch1 opened Oct 22, 2026Contributor·

  ·

• arm64: SME2 SGEMV and DGEMV kernels for VORTEXM4

  #6156·tesch1 opened Oct 22, 2026Contributor·

  ·

• arm64: NEON small-matrix GEMM kernels for VORTEXM4

  #6155·tesch1 opened Oct 22, 2026Contributor·

  ·
  1

• arm64: VORTEXM4 level-3 panel copies in streaming mode

  #6154·tesch1 opened Oct 21, 2026Contributor·

  ·

• arm64: read the SME2 vector length only after the FEAT_SME2 test

  #6153·tesch1 opened Oct 21, 2026Contributor·

  ·
  1

• arm64: NEON BFDOT sbgemm kernel for VORTEX (Apple M1-M3), k-pair interleaved packing

  #6145·0tomhoyt opened Oct 21, 2026·

  ·
  21

• arm64: SME2 bf16 sbgemm kernel for VORTEXM4 (Apple M4)

  #6142·0tomhoyt opened Oct 20, 2026·

  ·
  8

• Preserve config.h timestamp when contents are unchanged

  #6140·birolcapa opened Oct 20, 2026·

  ·
  2

• Enable GEMM_GEMV_FORWARD by default on x86_64

  #6128·ACSimon33 opened Oct 14, 2026Contributor·

  ·

• cmake: fix VORTEXM4 guard for DYNAMIC_ARCH

  #6117·topolarity opened Oct 9, 2026·

  ·
  1

• Restore the Windows thread server callback code from PR 4577

  #6115·martin-frbg opened Oct 9, 2026Collaborator·

  ·

  · 0.3.35
  1

• Fix ARMV9SME kernel selection in CMake cross builds

  #6114·Kiptoo-Deus opened Oct 9, 2026·

  ·

• Minor notes regarding OpenMP implementation details

  #6109·nmnobre opened Oct 8, 2026·

  ·
  3

• AArch64 NRM2: recover finite results near overflow

  #6095·BenKnill opened Oct 5, 2026Contributor·

  ·
  2

• Add L() macro for local labels and enable .subsections_via_symbols on aarch64-darwin

  #6090·topolarity opened Oct 2, 2026·

  ·

• Split gotoblas dispatch table by group / operation (2/2)

  #6088·topolarity opened Oct 2, 2026·

  ·
  3