Commit Graph

13047 Commits

Author SHA1 Message Date
Antonio Sánchez
661cdb227f Fix relative paths after move.
libeigen/eigen!2234
2026-03-02 19:50:30 +00:00
Rasmus Munk Larsen
57b1de2330 Fix row-major GEMV dropping rows when n8 heuristic disables main loop
libeigen/eigen!2233

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-03-01 23:47:35 -08:00
Rasmus Munk Larsen
662d5c21ff Optimize SYMV, SYR, SYR2, and TRMV product kernels
libeigen/eigen!2228

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-03-01 19:40:11 -08:00
Rasmus Munk Larsen
c66fc52868 Add ULP accuracy measurement tool and documentation for vectorized math functions
libeigen/eigen!2153

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-03-01 13:22:16 -08:00
Rasmus Munk Larsen
c20b6f5c41 Restore EIGEN_EMPTY_STRUCT_CTOR as a no-op macro for backward compatibility
libeigen/eigen!2231

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-03-01 11:44:38 -08:00
Rasmus Munk Larsen
77d9173596 Add ca-certificates to clang-tidy CI job
libeigen/eigen!2230

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-28 09:12:13 -08:00
Rasmus Munk Larsen
444ae9761d Clamp igamma/igammac output to [0,1] for numerical stability
libeigen/eigen!2229

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-28 08:52:44 -08:00
Rasmus Munk Larsen
eddb470a09 Fix flaky array_cwise and sparse_basic tests
libeigen/eigen!2227

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-27 21:52:46 -08:00
Antonio Sánchez
25dba492e3 Use stack-constructed variable for SVD block sweep.
libeigen/eigen!2225
2026-02-28 05:04:41 +00:00
Rasmus Munk Larsen
f64d1e0acc Improve ConditionEstimator docs and tighten test bounds
libeigen/eigen!2226

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-26 22:08:13 -08:00
Rasmus Munk Larsen
8525491eb1 Add dedicated unit tests and benchmark for ConditionEstimator
libeigen/eigen!2223

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-26 18:26:38 -08:00
Antonio Sánchez
e730b1fe33 Fix mixed products GEMM.
libeigen/eigen!2224
2026-02-26 15:47:39 -08:00
Rasmus Munk Larsen
3adfa9bd37 Add const to non-mutating member functions across remaining modules
libeigen/eigen!2222

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-26 13:58:04 -08:00
Rasmus Munk Larsen
13b61529f4 Add const to non-mutating member functions in products/ and Serializer
libeigen/eigen!2221

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-26 12:35:44 -08:00
Rasmus Munk Larsen
aaca9e5856 Add missing const qualifiers in Eigen/src/Core/
libeigen/eigen!2220

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-26 11:23:53 -08:00
Rasmus Munk Larsen
1b1b7e347d Fix EIGEN_NO_AUTOMATIC_RESIZING not resizing empty destinations
libeigen/eigen!2219

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-26 07:54:27 -08:00
Rasmus Munk Larsen
064d686c57 Remove CXX11/ directory nesting for Tensor modules
libeigen/eigen!2199

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-26 07:03:38 -08:00
Rasmus Munk Larsen
11eb66e1b5 Remove pre-C++14 workarounds from unsupported/ tensor code
libeigen/eigen!2218

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-26 06:17:39 -08:00
Rasmus Munk Larsen
a95440de17 Remove obsolete bench/ and btl/ directories
libeigen/eigen!2217

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-25 20:19:45 -08:00
Rasmus Munk Larsen
6e2aff6b5d Fix ambiguous static_cast in JacobiSVD blocking threshold
libeigen/eigen!2215

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-25 19:48:20 -08:00
Rasmus Munk Larsen
d8ed4f6884 Fix GEBP half/quarter-packet loops for nr>=8 RHS packing on ARM64
libeigen/eigen!2216

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-25 19:26:49 -08:00
Rasmus Munk Larsen
6b6d0d8c8e Revert "Fix ambiguous static_cast in JacobiSVD blocking threshold computation"
This reverts commit e567151ce3.
2026-02-25 19:08:21 -08:00
Rasmus Munk Larsen
ba2fc4e775 Revert "Fix GEBP half/quarter-packet loops for nr>=8 RHS packing on ARM64"
This reverts commit 888d708dcd.
2026-02-25 19:08:21 -08:00
Rasmus Munk Larsen
888d708dcd Fix GEBP half/quarter-packet loops for nr>=8 RHS packing on ARM64
On ARM64 (and LoongArch64), the GEBP kernel uses nr=8, so the RHS is
packed in 8-column blocks. The half-packet and quarter-packet row
processing loops were iterating columns 4 at a time starting from j2=0,
misindexing into the 8-column packed RHS buffer. This produced
completely wrong results for float GEMM when the number of rows was
smaller than the SIMD packet size (e.g. 2x10 * 10x8 float).

Add the missing nr>=8 column iteration blocks to both loops, matching
the pattern already present in the 3x, 2x, 1x, and scalar remainder
sections.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-02-25 19:03:11 -08:00
Rasmus Munk Larsen
e567151ce3 Fix ambiguous static_cast in JacobiSVD blocking threshold computation
The L2 cache size threshold computation used numext::sqrt with a
static_cast<RealScalar>, which fails to compile when RealScalar is
AnnoyingScalar (a test-only type with multiple conversion constructors).
Since this is a pure cache-size computation unrelated to the matrix
scalar type, use std::sqrt(double) instead.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-02-25 19:03:01 -08:00
Rasmus Munk Larsen
a31de4778d Blocked Jacobi SVD sweep with L2-cache-adaptive threshold
libeigen/eigen!2206

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
Co-authored-by: Rasmus Munk Larsen <rmlarsen@google.com>
2026-02-25 10:03:05 -08:00
Rasmus Munk Larsen
647e0009ba Refactor BDCSVD D&C code to reduce compilation time and memory footprint
libeigen/eigen!2211

Closes #3048

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-25 09:11:38 -08:00
Rasmus Munk Larsen
4fab38d798 Make clang generic vector backend support 16, 32, and 64-byte vectors
libeigen/eigen!2213

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-25 08:50:47 -08:00
Rasmus Munk Larsen
ea25ea52bb Revert accidental changes from !2212 squash merge
libeigen/eigen!2214

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-25 08:31:41 -08:00
Rasmus Munk Larsen
38f0f42755 Update rmlarsen email address from @google.com to @gmail.com
libeigen/eigen!2212

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-25 07:45:02 -08:00
Rasmus Munk Larsen
d0d70a9527 Consolidate complex math function boilerplate with shared macros
libeigen/eigen!2201

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-25 07:21:20 -08:00
Rasmus Munk Larsen
c4c704e5dd Install libclang-rt-19-dev for asan-ubsan CI job
libeigen/eigen!2210

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-24 20:19:42 -08:00
Rasmus Munk Larsen
61895c5978 Selectively add constexpr to Core expression template scaffolding
libeigen/eigen!2184

Closes #3041

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-24 19:59:10 -08:00
Rasmus Munk Larsen
34092d2788 Fix flaky tests: add iteration guards, yield in busy-waits, cap thread count
libeigen/eigen!2208

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-24 18:29:07 -08:00
Rasmus Munk Larsen
28d090a49c Refactor GenericPacketMathFunctions.h into smaller focused headers
libeigen/eigen!2200

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-24 17:46:12 -08:00
Rasmus Munk Larsen
16da0279f1 Add benchmarks for unsupported modules and extend supported benchmarks
libeigen/eigen!2179

Closes #3036

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-24 17:12:33 -08:00
Rasmus Munk Larsen
fa567f6bcd Add CUDA CI jobs with NVHPC (nvc++) as host and device compiler
libeigen/eigen!2204

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-24 16:54:08 -08:00
Antonio Sánchez
2cd9bb7380 Fix sparse product with entities that do not have direct access.
libeigen/eigen!2205
2026-02-24 16:27:06 -08:00
Rasmus Munk Larsen
00cc497d32 Add clang-tidy, codespell, and sanitizer checks to CI pipeline
libeigen/eigen!2178

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-23 19:43:45 -08:00
Rasmus Munk Larsen
241af1c0ba Add NVHPC (nvc++) compiler support and CI build/test jobs
libeigen/eigen!2186

Closes #3032

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-23 18:51:15 -08:00
Antonio Sánchez
f3f2c676b5 Fix direct access for sparse blocks.
libeigen/eigen!2202
2026-02-23 12:00:52 -08:00
Rasmus Munk Larsen
d537b51ede Fix ComplexEigenSolver NaN with flush-to-zero arithmetic
libeigen/eigen!2196

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-23 11:15:31 -08:00
Rasmus Munk Larsen
667cabe3aa Clean up comments in unsupported module
libeigen/eigen!2198

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-22 22:04:23 -08:00
Rasmus Munk Larsen
78b76986b7 Comment cleanup v3: trailing ??, informal language, FIXME/TODO colons
libeigen/eigen!2197

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-22 21:20:08 -08:00
Rasmus Munk Larsen
112c2324bd Consolidate BF16/F16 wrapper macros and simplify arch math functions
libeigen/eigen!2195

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-22 20:17:43 -08:00
Rasmus Munk Larsen
d5e67adbe7 Clean up informal language, vague TODOs, and dead code in comments
libeigen/eigen!2191

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-22 18:32:10 -08:00
Rasmus Munk Larsen
7d727d26bc Refactor GenericPacketMathFunctions.h into smaller focused headers
libeigen/eigen!2190

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-22 16:30:57 -08:00
Rasmus Munk Larsen
9810969c0f Suppress false-positive GCC and clang warnings in test builds
libeigen/eigen!2187

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-22 14:54:15 -08:00
Rasmus Munk Larsen
ad7f1fe70e Improve clang vector extension backend
libeigen/eigen!2183

Closes #3042

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-22 13:31:21 -08:00
Rasmus Munk Larsen
1f49bf96cf Add new benchmarks for Core, LU, and QR operations
libeigen/eigen!2177

Closes #3035

Co-authored-by: Rasmus Munk Larsen <rmlarsen@gmail.com>
2026-02-22 12:19:37 -08:00