phynexis-deps/lapack
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
3,696 Commits 10 Branches 16 Tags
4fc8389ed9c21cfbeef1a5f3c5a2f61d38dfbc09
Commit Graph
3696 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
Simon Maertens 4fc8389ed9 The include directory in the binary directory should take precedence over the one in the source. 2026-07-21 15:31:11 +02:00
Simon Maertens 7d732a4c54 Removed lapacke_mangling.h. The header is generated on the fly and the one that was checked into the repository interfered with the generated one. 2026-07-21 15:04:28 +02:00
Simon Maertens 3b108cbd6a Fixed constant NaN propagation for NAG Fortran compiler 2026-07-21 13:22:00 +02:00
langou c1c5b726be Merge pull request #1249 from martin-frbg/issue1223 
Adjust line reflow in complex (GER/POR/SYR)FSX for ILP64 build
 2026-07-15 03:30:25 -06:00
langou 1954919f2f Merge pull request #1260 from ACSimon33/export_import_cblas_globals 
Fix CBLAS tests for shared Windows builds
 2026-07-15 03:28:26 -06:00
langou bceadf3238 Merge pull request #1278 from ACSimon33/index_64_conversion_errors 
Fix Index-64 conversion errors
 2026-07-15 03:25:51 -06:00
langou ee81e65267 Merge pull request #1250 from martin-frbg/issue1118 
Allow using gfortran even with the Cray PE_ENV set to something else
 2026-07-15 03:23:35 -06:00
langou 379b30d311 Merge pull request #1245 from nakatamaho/fix/lartg-stable-complex-s 
avoid large intermediates in complex LARTG
 2026-07-15 03:22:29 -06:00
langou 31a1d7844b Merge pull request #1325 from langou/add_path_in_ilaenv_for_tzrzf_and_unmrz 2026-07-14 14:07:54 -06:00
langou 06060fb7f6 Merge pull request #1161 from scr2016/scr2016-dgecx-draft 
adding GECXX to LAPACK
 2026-07-14 10:09:40 -06:00
langou 212bf98263 add a return to line 2026-07-14 06:51:01 -06:00
langou 5b484b2ae5 Merge branch 'master' into scr2016-dgecx-draft 2026-07-14 06:39:18 -06:00
Julien Langou 0e9dea7351 add path for TRTRZ and UNMRZ in ILAENV 2026-07-14 04:17:11 -06:00
langou 693f114224 Merge pull request #1324 from Reference-LAPACK/revert-1295-issue1277 
Revert "*evr, *stevr: use fast CSTEMR path for partial eigenvalue ranges on I…"
 2026-07-13 10:59:55 -06:00
langou e0b72f6a9f Merge pull request #1279 from ACSimon33/fix-external-declarations 
Fix external declarations
 2026-07-13 10:52:33 -06:00
Simon Lukas Märtens 1068fdbe9c Merge branch 'master' into fix-external-declarations 2026-07-13 18:19:35 +02:00
langou 7551f87f5e Merge pull request #1305 from jschueller/issue608 
TGSJA: replace HUGE intrinsic with LAMCH('O')
 2026-07-13 09:10:40 -06:00
langou d62eebccd4 Revert "*evr, *stevr: use fast CSTEMR path for partial eigenvalue ranges on I…" 2026-07-13 08:11:05 -06:00
langou 9e55d36949 Merge pull request #1297 from jschueller/issue744 
*drgev: relax eigenvalue consistency test
 2026-07-13 06:31:33 -06:00
langou c0b35f8a83 Merge pull request #1323 from langou/remove_many_IF_statements_in_TRSM 
Remove IF statement related to initial zeros in X for triangular multiplications and solves ( tbmv, tbsv, tpmv, tpsv, trmm, trmv, trsm, and trsv )
 2026-07-13 06:20:27 -06:00
langou ee358af929 Merge pull request #1322 from Reference-LAPACK/lapack__comments_in_geqrf 
comments in geqrf based on #1304 and #257
 2026-07-13 06:20:12 -06:00
langou 6bf019efd1 Merge pull request #1298 from jschueller/issue973 
*chklq: fix test failures when NB=1 with M=N
 2026-07-13 06:09:06 -06:00
langou 05a6d9f4e9 Merge pull request #1295 from jschueller/issue1277 
*evr, *stevr: use fast CSTEMR path for partial eigenvalue ranges on I…
 2026-07-13 03:30:18 -06:00
langou 9cf8bd2378 Merge pull request #1288 from jschueller/issue1108 
slaruv/dlaruv: fix seed advancement on rejected uniform deviate
 2026-07-13 03:09:45 -06:00
langou 76bb13e396 Merge pull request #1290 from jschueller/issue938 
slarfgp/dlarfgp: avoid overflow when computing 1/ALPHA
 2026-07-13 03:07:53 -06:00
Julien Langou 114140a8ba remove the IF (X(J).NE.ZERO) THEN that saves FLOPS but might prevent NaN propagation 2026-07-12 10:55:23 -06:00
Julien Langou ba20cd239b remove the IF (A(J,K).NE.ZERO) THEN that saves FLOPS but might prevent NaN propagation 2026-07-12 10:47:44 -06:00
Julien Langou acff87b05a remove the IF (X(J).NE.ZERO) THEN that saves FLOPS but might prevent NaN propagation 2026-07-12 10:37:27 -06:00
Julien Langou c9e94a06c5 remove the IF (X(J).NE.ZERO) THEN that saves FLOPS but might prevent NaN propagation 2026-07-12 10:35:24 -06:00
Julien Langou 6e7b86cda4 remove the IF (X(J).NE.ZERO) THEN that saves FLOPS but might prevent NaN propagation 2026-07-12 10:19:57 -06:00
Julien Langou ce673d064d remove the IF (X(J).NE.ZERO) THEN that saves FLOPS but might prevent NaN propagation 2026-07-12 10:15:11 -06:00
Julien Langou b8607fe70f remove the IF (X(J).NE.ZERO) THEN that saves FLOPS but might prevent NaN propagation 2026-07-12 10:09:50 -06:00
Julien Langou 301097afd8 remove a TEMP = that is good to make the real arithmetic code looks like the complex arithmetic one but weird otherwise 2026-07-12 10:08:38 -06:00
Julien Langou e8785e4e1e remove all TEMP = ONE/A(J,J) and force explicit division like B(I,J)/A(J,J) instead of TEMP*B(I,J) 2026-07-12 09:54:04 -06:00
Julien Langou 38ccf80859 remove all IF (A(K,J).NE.ZERO) 2026-07-12 09:26:24 -06:00
Julien Langou f0aaa58e14 remove all (ALPHA.NE.ONE) in TRSM family 2026-07-12 09:09:04 -06:00
Julien Langou b7c0841b09 comments in geqrf based on #1304 and #257 2026-07-09 06:43:06 -06:00
langou 552873f0a9 Merge pull request #1306 from jschueller/issue977 
{d/s}sytrf: fix documentation error in UPLO='U' factorization
 2026-07-09 06:18:26 -06:00
langou 36c931cac2 Merge pull request #1304 from jschueller/issue257 
{s/c/d/z}geqrf: document workspace query for empty matrices
 2026-07-09 05:34:06 -06:00
langou d1e64796ed Merge pull request #1317 from jschueller/issue255_bis 
{d/s}laed2/8: Increase deflation tolerance to match {d/s}lasd2/7
 2026-07-09 05:32:29 -06:00
langou 6840252146 Merge pull request #1301 from jschueller/issue636 
*trsm: remove zero-RHS skip for consistent NaN on singular matrix
 2026-07-09 04:40:03 -06:00
langou fc09e73853 Merge pull request #1319 from langou/tmp_remove_extra_comments_introduced_in_PR1318 
remove extra comments introduced in PR #1318
 2026-07-08 08:48:00 -06:00
Julien Langou 46b06d80b6 remove extra comments introduced in PR #1318 2026-07-08 07:22:30 -06:00
langou fff42b8aeb Merge pull request #1318 from meng004/fix-lapacke-lacpy-rowmajor-729 
Fix LAPACKE_?lacpy_work row-major triangular copy corrupting the untouched triangle (#729)
 2026-07-08 07:18:02 -06:00
meng004 3a30b715ff Fix LAPACKE_?lacpy_work row-major triangular copy (#729) 
The row-major branch transposed the full m-by-n matrix through
temporaries and copied the never-written complementary region of the
destination temporary back over B, corrupting the triangle uplo requires
to be preserved and reading uninitialized memory. Call the Fortran kernel
directly with uplo swapped (U<->L) and m/n swapped, exactly as the
correct column-major path does. Applies identically to s/d/c/z.
 2026-07-06 15:03:58 +00:00
Julien Schueller 9e64af6288 {d/s}laed2/8: Increase deflation tolerance to match {d/s}lasd2/7 
The eigenvalue divide-and-conquer helper routines DLAED2/SLAED2 and
DLAED8/SLAED8 used a deflation tolerance of 8*EPS, while the SVD
routines DLASD2/SLASD2 and DLASD7/SLASD7 now use 64*EPS (since
PR #1286). The downstream eigenvector computation in DLAED3 divides
by eigenvalue differences (DLAMBDA(I) - DLABMDA(J)), causing the same
catastrophic cancellation when close eigenvalues escape deflation with
the tighter tolerance.

Raise all four routines to 64*EPS (EIGHT*EIGHT*EPS), matching the SVD
path, so close eigenvalues are deflated and the eigenvector computation
remains stable.

Follow-up for #255
 2026-06-30 22:50:07 +02:00
langou 98e5d7a237 Merge pull request #1315 from sameer-amd-in/issue1314 
{s/c/d/z}larf{/1f/1l}: fix negative increment vector handling
 2026-06-24 02:20:05 -06:00
samahmad 77abd25204 {s/c/d/z}larf{/1f/1l}: fix negative increment vector handling 
Handle negative increment values in vector offsetting.

Change-Id: I2a0993e1ec22f8029d49d30d18be0152db480ded
 2026-06-23 22:06:06 -07:00
Julien Schueller 683f28cd41 {s/c/d/z}geqrf: document workspace query for empty matrices 
Clarify that the workspace query returns a valid LWORK (>= 1) even
when MIN(M,N) = 0. The code already handles this correctly; this
is purely a documentation update.
 2026-06-15 20:37:06 +02:00
langou 122f8924ed Merge pull request #1309 from amilsted/issue_995 2026-06-15 12:23:09 -06:00