2 ====================================================================
6 * Added BLAS extensions as following.
7 s/d/c/zaxpby, s/d/c/zimatcopy, s/d/c/zomatcopy.
8 * Added OPENBLAS_CORETYPE environment for dynamic_arch. (a86d34)
9 * Added NO_AVX2 flag for old binutils. (#401)
10 * Support outputing the CPU corename on runtime.(#407)
11 * Patched LAPACK to fix bug 114, 117, 118.
12 (http://www.netlib.org/lapack/bug_list.html)
13 * Disabled ?gemm3m for a work-around fix. (#400)
15 * Fixed lots of bugs for optimized kernels on sandybridge,Haswell,
16 bulldozer, and piledriver.
17 https://github.com/xianyi/OpenBLAS/wiki/Fixed-optimized-kernels-To-do-List
20 * Improved LAPACK testing.
22 ====================================================================
26 * Improved the result for LAPACK testing. (#372)
27 * Installed DLL to prefix/bin instead of prefix/lib. (#366)
28 * Build import library on Windows.(#374)
30 * To improve LAPACK testing, we fallback some kernels. (#372)
31 https://github.com/xianyi/OpenBLAS/wiki/Fixed-optimized-kernels-To-do-List
33 ====================================================================
37 * Added OPENBLAS_VERBOSE environment variable.(#338)
38 * Make OpenBLAS thread-pool resilient to fork via pthread_atfork.
39 (#294, Thank Olivier Grisel)
43 * Detect Intel Haswell for new Macbook.
45 ====================================================================
49 * Update LAPACK to 3.5.0 version
50 * Fixed compatiable issues with Clang and Pathscale compilers.
53 * Optimization on Intel Haswell.
54 * Enable optimization kernels on AMD Bulldozer and Piledriver.
57 * Support ARMv6 and ARMv7 ISA.
58 * Optimization on ARM Cortex-A9.
60 ====================================================================
64 * Support Open64 5.0. (#266)
65 * Add executable stack markings. (#262, Thank Sébastien Fabbro)
66 * Respect user's LDFLAGS (Thank Sébastien Fabbro)
69 * Rollback bulldozer and piledriver kernels to barcelona kernels (#263)
70 We will fix the compuational error bug in bulldozer and piledriver kernels.
72 ====================================================================
76 * Support LSB (Linux Standard Base) 4.1.
78 * Include LAPACK 3.4.2 source codes to the repo.
79 Avoid downloading at compile time.
80 * Add NO_PARALLEL_MAKE flag to disable parallel make.
81 * Create openblas_get_parallel to retrieve information which
82 parallelization model is used by OpenBLAS. (Thank grisuthedragon)
83 * Detect LLVM/Clang compiler. The default compiler is Clang on Mac OS X.
84 * Change LIBSUFFIX from .lib to .a on windows.
85 * A work-around for dtrti_U single thread bug. Replace it with LAPACK codes. (#191)
88 * Optimize c/zgemm, trsm, dgemv_n, ddot, daxpy, dcopy on
89 AMD Bulldozer. (Thank Werner Saar)
90 * Add Intel Haswell support (using Sandybridge optimizations).
92 * Add AMD Piledriver support (using Bulldozer optimizations).
93 * Fix the computational error in zgemm avx kernel on
95 * Fix the overflow bug in gemv.
96 * Fix the overflow bug in multi-threaded BLAS3, getrf when NUM_THREADS
97 is very large.(#214, #221, #246).
99 * Support loongcc (Open64 based) compiler for ICT Loongson 3A/B.
102 * Support Power7 by old Power6 kernels. (#220)
104 ====================================================================
108 * Improved OpenMP performance slightly. (d744c9)
109 * Improved cblas.h compatibility with Intel MKL.(#185)
110 * Fixed the overflowing bug in single thread cholesky factorization.
111 * Fixed the overflowing buffer bug of multithreading hbmv and sbmv.(#174)
114 * Added AMD Bulldozer x86-64 S/DGEMM AVX kernels. (Thank Werner Saar)
115 We will tune the performance in future.
116 * Auto-detect Intel Xeon E7540.
117 * Fixed the overflowing buffer bug of gemv. (#173)
118 * Fixed the bug of s/cdot about invalid reading NAN on x86_64. (#189)
122 ====================================================================
126 * Added NO_SHARED flag to disable generating the shared library.
127 * Compile LAPACKE with ILP64 modle when INTERFACE64=1 (#158)
128 * Export LAPACK 3.4.2 symbols in shared library. (#147)
129 * Only detect the number of physical CPU cores on Mac OSX. (#157)
130 * Fixed NetBSD build. (#155)
131 * Fixed compilation with TARGET=GENERIC. (#160)
133 * Restore the original CPU affinity when calling
134 openblas_set_num_threads(1) (#153)
135 * Fixed a SEGFAULT bug in dgemv_t when m is very large.(#154)
138 ====================================================================
142 * Upgraded LAPACK to 3.4.2 version. (#145)
143 * Provided support for passing CFLAGS, FFLAGS, PFLAGS,
144 FPFLAGS to make. (#137)
145 * f77blas.h:compatibility for compilers without C99 complex
146 number support. (#141)
148 * Added NO_AVX flag. Check OS supporting AVX on runtime. (#139)
149 * Fixed zdot incompatibility ABI issue with GCC 4.7 on
150 Windows 32-bit. (#140)
152 * Fixed the generation of shared library bug.
153 * Fixed the detection bug on the Loongson 3A server.
154 ====================================================================
158 * Fixed LAPACK unstable bug about ?laswp. (#130)
159 * Fixed the shared library bug about unloading the library on
161 * Fixed the compilation failure on BlueGene/P (TARGET=PPC440FP2)
162 Please use gcc and IBM xlf. (#134)
164 * Supported goto_set_num_threads and openblas_set_num_threads
165 APIs in Windows. They can set the number of threads on runtime.
167 ====================================================================
171 * Fixed exporting DLL functions bug on Windows/MingW
172 * Support GNU Hurd (Thank Sylvestre Ledru)
173 * Support kfreebsd kernel (Thank Sylvestre Ledru)
175 * Support Intel Sandy Bridge 22nm desktop/mobile CPU
177 * Improve the detection of SPARC (Thank Sylvestre Ledru)
179 ====================================================================
184 * Fixed the SEGFAULT bug about hyper-theading
185 * Support AMD Bulldozer by using GotoBLAS2 AMD Barcelona codes
187 ====================================================================
191 * Removed the limitation (64) of numbers of CPU cores.
192 Now, it supports 256 cores at max.
193 * Supported clang compiler.
194 * Fixed some build bugs on FreeBSD
196 * Optimized Level-3 BLAS on Intel Sandy Bridge x86-64 by AVX instructions.
197 Please use gcc >= 4.6 or clang >=3.1.
198 * Support AMD Bobcat by using GotoBLAS2 AMD Barcelona codes.
200 ====================================================================
204 * Upgraded LAPACK to 3.4.1 version. (Thank Zaheer Chothia)
205 * Supported LAPACKE, a C interface to LAPACKE. (Thank Zaheer Chothia)
206 * Fixed the build bug (MD5 and download) on Mac OSX.
207 * Auto download CUnit 2.1.2-2 from SF.net with UTEST_CHECK=1.
208 * Fxied the compatibility issue for compilers without C99 complex number
211 * Auto-detect Intel Sandy Bridge Core i7-3xxx & Xeon E7 Westmere-EX.
212 * Test alpha=Nan in dscale.
213 * Fixed a SEGFAULT bug in samax on x86 windows.
215 ====================================================================
219 * Set soname of shared library on Linux.
220 * Added LIBNAMESUFFIX flag in Makefile.rule. The user can use
221 this flag to control the library name, e.g. libopenblas.a,
222 libopenblas_ifort.a or libopenblas_omp.a.
223 * Added GEMM_MULTITHREAD_THRESHOLD flag in Makefile.rule.
224 The lib use single thread in GEMM function with small matrices.
226 * Used GEMV SSE/SSE2 kernels on x86 32-bit.
227 * Exported CBLAS functions in Windows DLL.
229 * Completed Level-3 BLAS optimization on Loongson 3A CPU.
230 * Improved GEMV performance on Loongson 3A CPU.
231 * Improved Level-3 BLAS performance on Loongson 3B CPU. (EXPERIMENT)
233 ====================================================================
237 * Fixed missing "#include <sched.h>" bug on Mac OS X.
238 Thank Mike Nolta for the patch.
239 * Upgraded LAPACK to 3.4.0 version
240 * Fixed a bug on Mac OS X. Don't require SystemStubs on OS X.
241 SystemStubs does not exist on Lion. Thank Stefan Karpinski.
242 * Improved README with using OpenMP. Check the internal threads
243 count less than or equal to omp_get_max_threads()
245 * Auto-detect Intel Core i6/i7 (Sandy Bridge) CPU with Nehalem assembly kernels
246 * Fixed some bugs on MingW 64-bit including zgemv, cdot, zdot.
248 ====================================================================
252 * Fixed a bug about installation. The header file "fblas77.h"
254 * Fixed #61 a building bug about setting TARGET and DYNAMIC_ARCH.
255 * Try to handle absolute path of shared library in OSX. (#57)
256 Thank Dr Kane O'Donnell.
257 * Changed the installation folder layout to $(PREFIX)/include and
261 * Fixed #58 zdot/xdot SEGFAULT bug with GCC-4.6 on x86. According
262 to i386 calling convention, The callee should remove the first
263 hidden parameter.Thank Mr. John for this patch.
265 ====================================================================
270 * Added DTB_ENTRIES into dynamic arch setting parameters. Now,
271 it can read DTB_ENTRIES on runtime. (Refs issue #55 on github)
273 ====================================================================
278 * Fixed a building bug when DYNAMIC_ARCH=1 & INTERFACE64=1.
279 (Refs issue #44 on github)
281 ====================================================================
286 * Stop the build and output the error message when detecting
287 fortran compiler failed. (Refs issue #42 on github)
289 ====================================================================
294 * Fixed blasint undefined bug in <cblas.h> file. Other software
295 could include this header successfully(Refs issue #13 on github)
296 * Fixed the SEGFAULT bug on 64 cores. On SMP server, the number
297 of CPUs or cores should be less than or equal to 64.(Refs issue #14
299 * Support "void goto_set_num_threads(int num_threads)" and "void
300 openblas_set_num_threads(int num_threads)" when USE_OPENMP=1
301 * Added extern "C" to support C++. Thank Tasio for the patch(Refs
303 * Provided an error message when the arch is not supported.(Refs
305 * Fixed issue #23. Fixed a bug of f_check script about generating link flags.
306 * Added openblas_set_num_threads for Fortran.
307 * Fixed #25 a wrong result of rotmg.
308 * Fixed a bug about detecting underscore prefix in c_check.
309 * Print the wall time (cycles) with enabling FUNCTION_PROFILE
310 * Fixed #35 a build bug with NO_LAPACK=1 & DYNAMIC_ARCH=1
311 * Added install target. You can use "make install". (Refs #20)
315 * Fixed #28 a wrong result of dsdot on x86_64.
316 * Fixed #32 a SEGFAULT bug of zdotc with gcc-4.6.
317 * Fixed #33 ztrmm bug on Nehalem.
318 * Work-around #27 the low performance axpy issue with small imput size & multithreads.
321 * Fixed #28 a wrong result of dsdot on Loongson3A/MIPS64.
322 * Optimized single/double precision BLAS Level3 on Loongson3A/MIPS64. (Refs #2)
323 * Optimized single/double precision axpy function on Loongson3A/MIPS64. (Refs #3)
325 ====================================================================
330 * Support "make NO_LAPACK=1" to build the library without
332 * Fixed randomly SEGFAULT when nodemask==NULL with above Linux 2.6.34.
333 Thank Mr.Ei-ji Nakama providing this patch. (Refs issue #12 on github)
334 * Added DEBUG=1 rule in Makefile.rule to build debug version.
335 * Disable compiling quad precision in reference BLAS library(netlib BLAS).
336 * Added unit testcases in utest/ subdir. Used CUnit framework.
337 * Supported OPENBLAS_* & GOTO_* environment variables (Pleas see README)
338 * Imported GotoBLAS2 1.13 BSD version
341 * On x86 32bits, fixed a bug in zdot_sse2.S line 191. This would casue
342 zdotu & zdotc failures. Instead, work-around it. (Refs issue #8 #9 on github)
343 * Modified ?axpy functions to return same netlib BLAS results
344 when incx==0 or incy==0 (Refs issue #7 on github)
345 * Modified ?swap functions to return same netlib BLAS results
346 when incx==0 or incy==0 (Refs issue #6 on github)
347 * Modified ?rot functions to return same netlib BLAS results
348 when incx==0 or incy==0 (Refs issue #4 on github)
349 * Detect Intel Westmere,Intel Clarkdale and Intel Arrandale
350 to use Nehalem codes.
351 * Fixed a typo bug about compiling dynamic ARCH library.
353 * Improve daxpy performance on ICT Loongson 3A.
354 * Supported ICT Loongson 3A CPU (Refs issue #1 on github)
355 ====================================================================