2 ====================================================================
6 * Support LSB (Linux Standard Base) 4.1.
8 * Include LAPACK 3.4.2 source codes to the repo.
9 Avoid downloading at compile time.
10 * Add NO_PARALLEL_MAKE flag to disable parallel make.
11 * Create openblas_get_parallel to retrieve information which
12 parallelization model is used by OpenBLAS. (Thank grisuthedragon)
13 * Detect LLVM/Clang compiler. The default compiler is Clang on Mac OS X.
14 * Change LIBSUFFIX from .lib to .a on windows.
15 * A walk round for dtrti_U single thread bug. Replace it with LAPACK codes. (#191)
18 * Optimize c/zgemm, trsm, dgemv_n, ddot, daxpy, dcopy on
19 AMD Bulldozer. (Thank Werner Saar)
20 * Add Intel Haswell support (using Sandybridge optimizations).
22 * Add AMD Piledriver support (using Bulldozer optimizations).
23 * Fix the computational error in zgemm avx kernel on
25 * Fix the overflow bug in gemv.
26 * Fix the overflow bug in multi-threaded BLAS3, getrf when NUM_THREADS
27 is very large.(#214, #221, #246).
29 * Support loongcc (Open64 based) compiler for ICT Loongson 3A/B.
32 * Support Power7 by old Power6 kernels. (#220)
34 ====================================================================
38 * Improved OpenMP performance slightly. (d744c9)
39 * Improved cblas.h compatibility with Intel MKL.(#185)
40 * Fixed the overflowing bug in single thread cholesky factorization.
41 * Fixed the overflowing buffer bug of multithreading hbmv and sbmv.(#174)
44 * Added AMD Bulldozer x86-64 S/DGEMM AVX kernels. (Thank Werner Saar)
45 We will tune the performance in future.
46 * Auto-detect Intel Xeon E7540.
47 * Fixed the overflowing buffer bug of gemv. (#173)
48 * Fixed the bug of s/cdot about invalid reading NAN on x86_64. (#189)
52 ====================================================================
56 * Added NO_SHARED flag to disable generating the shared library.
57 * Compile LAPACKE with ILP64 modle when INTERFACE64=1 (#158)
58 * Export LAPACK 3.4.2 symbols in shared library. (#147)
59 * Only detect the number of physical CPU cores on Mac OSX. (#157)
60 * Fixed NetBSD build. (#155)
61 * Fixed compilation with TARGET=GENERIC. (#160)
63 * Restore the original CPU affinity when calling
64 openblas_set_num_threads(1) (#153)
65 * Fixed a SEGFAULT bug in dgemv_t when m is very large.(#154)
68 ====================================================================
72 * Upgraded LAPACK to 3.4.2 version. (#145)
73 * Provided support for passing CFLAGS, FFLAGS, PFLAGS,
74 FPFLAGS to make. (#137)
75 * f77blas.h:compatibility for compilers without C99 complex
76 number support. (#141)
78 * Added NO_AVX flag. Check OS supporting AVX on runtime. (#139)
79 * Fixed zdot incompatibility ABI issue with GCC 4.7 on
80 Windows 32-bit. (#140)
82 * Fixed the generation of shared library bug.
83 * Fixed the detection bug on the Loongson 3A server.
84 ====================================================================
88 * Fixed LAPACK unstable bug about ?laswp. (#130)
89 * Fixed the shared library bug about unloading the library on
91 * Fixed the compilation failure on BlueGene/P (TARGET=PPC440FP2)
92 Please use gcc and IBM xlf. (#134)
94 * Supported goto_set_num_threads and openblas_set_num_threads
95 APIs in Windows. They can set the number of threads on runtime.
97 ====================================================================
101 * Fixed exporting DLL functions bug on Windows/MingW
102 * Support GNU Hurd (Thank Sylvestre Ledru)
103 * Support kfreebsd kernel (Thank Sylvestre Ledru)
105 * Support Intel Sandy Bridge 22nm desktop/mobile CPU
107 * Improve the detection of SPARC (Thank Sylvestre Ledru)
109 ====================================================================
114 * Fixed the SEGFAULT bug about hyper-theading
115 * Support AMD Bulldozer by using GotoBLAS2 AMD Barcelona codes
117 ====================================================================
121 * Removed the limitation (64) of numbers of CPU cores.
122 Now, it supports 256 cores at max.
123 * Supported clang compiler.
124 * Fixed some build bugs on FreeBSD
126 * Optimized Level-3 BLAS on Intel Sandy Bridge x86-64 by AVX instructions.
127 Please use gcc >= 4.6 or clang >=3.1.
128 * Support AMD Bobcat by using GotoBLAS2 AMD Barcelona codes.
130 ====================================================================
134 * Upgraded LAPACK to 3.4.1 version. (Thank Zaheer Chothia)
135 * Supported LAPACKE, a C interface to LAPACKE. (Thank Zaheer Chothia)
136 * Fixed the build bug (MD5 and download) on Mac OSX.
137 * Auto download CUnit 2.1.2-2 from SF.net with UTEST_CHECK=1.
138 * Fxied the compatibility issue for compilers without C99 complex number
141 * Auto-detect Intel Sandy Bridge Core i7-3xxx & Xeon E7 Westmere-EX.
142 * Test alpha=Nan in dscale.
143 * Fixed a SEGFAULT bug in samax on x86 windows.
145 ====================================================================
149 * Set soname of shared library on Linux.
150 * Added LIBNAMESUFFIX flag in Makefile.rule. The user can use
151 this flag to control the library name, e.g. libopenblas.a,
152 libopenblas_ifort.a or libopenblas_omp.a.
153 * Added GEMM_MULTITHREAD_THRESHOLD flag in Makefile.rule.
154 The lib use single thread in GEMM function with small matrices.
156 * Used GEMV SSE/SSE2 kernels on x86 32-bit.
157 * Exported CBLAS functions in Windows DLL.
159 * Completed Level-3 BLAS optimization on Loongson 3A CPU.
160 * Improved GEMV performance on Loongson 3A CPU.
161 * Improved Level-3 BLAS performance on Loongson 3B CPU. (EXPERIMENT)
163 ====================================================================
167 * Fixed missing "#include <sched.h>" bug on Mac OS X.
168 Thank Mike Nolta for the patch.
169 * Upgraded LAPACK to 3.4.0 version
170 * Fixed a bug on Mac OS X. Don't require SystemStubs on OS X.
171 SystemStubs does not exist on Lion. Thank Stefan Karpinski.
172 * Improved README with using OpenMP. Check the internal threads
173 count less than or equal to omp_get_max_threads()
175 * Auto-detect Intel Core i6/i7 (Sandy Bridge) CPU with Nehalem assembly kernels
176 * Fixed some bugs on MingW 64-bit including zgemv, cdot, zdot.
178 ====================================================================
182 * Fixed a bug about installation. The header file "fblas77.h"
184 * Fixed #61 a building bug about setting TARGET and DYNAMIC_ARCH.
185 * Try to handle absolute path of shared library in OSX. (#57)
186 Thank Dr Kane O'Donnell.
187 * Changed the installation folder layout to $(PREFIX)/include and
191 * Fixed #58 zdot/xdot SEGFAULT bug with GCC-4.6 on x86. According
192 to i386 calling convention, The callee should remove the first
193 hidden parameter.Thank Mr. John for this patch.
195 ====================================================================
200 * Added DTB_ENTRIES into dynamic arch setting parameters. Now,
201 it can read DTB_ENTRIES on runtime. (Refs issue #55 on github)
203 ====================================================================
208 * Fixed a building bug when DYNAMIC_ARCH=1 & INTERFACE64=1.
209 (Refs issue #44 on github)
211 ====================================================================
216 * Stop the build and output the error message when detecting
217 fortran compiler failed. (Refs issue #42 on github)
219 ====================================================================
224 * Fixed blasint undefined bug in <cblas.h> file. Other software
225 could include this header successfully(Refs issue #13 on github)
226 * Fixed the SEGFAULT bug on 64 cores. On SMP server, the number
227 of CPUs or cores should be less than or equal to 64.(Refs issue #14
229 * Support "void goto_set_num_threads(int num_threads)" and "void
230 openblas_set_num_threads(int num_threads)" when USE_OPENMP=1
231 * Added extern "C" to support C++. Thank Tasio for the patch(Refs
233 * Provided an error message when the arch is not supported.(Refs
235 * Fixed issue #23. Fixed a bug of f_check script about generating link flags.
236 * Added openblas_set_num_threads for Fortran.
237 * Fixed #25 a wrong result of rotmg.
238 * Fixed a bug about detecting underscore prefix in c_check.
239 * Print the wall time (cycles) with enabling FUNCTION_PROFILE
240 * Fixed #35 a build bug with NO_LAPACK=1 & DYNAMIC_ARCH=1
241 * Added install target. You can use "make install". (Refs #20)
245 * Fixed #28 a wrong result of dsdot on x86_64.
246 * Fixed #32 a SEGFAULT bug of zdotc with gcc-4.6.
247 * Fixed #33 ztrmm bug on Nehalem.
248 * Walk round #27 the low performance axpy issue with small imput size & multithreads.
251 * Fixed #28 a wrong result of dsdot on Loongson3A/MIPS64.
252 * Optimized single/double precision BLAS Level3 on Loongson3A/MIPS64. (Refs #2)
253 * Optimized single/double precision axpy function on Loongson3A/MIPS64. (Refs #3)
255 ====================================================================
260 * Support "make NO_LAPACK=1" to build the library without
262 * Fixed randomly SEGFAULT when nodemask==NULL with above Linux 2.6.34.
263 Thank Mr.Ei-ji Nakama providing this patch. (Refs issue #12 on github)
264 * Added DEBUG=1 rule in Makefile.rule to build debug version.
265 * Disable compiling quad precision in reference BLAS library(netlib BLAS).
266 * Added unit testcases in utest/ subdir. Used CUnit framework.
267 * Supported OPENBLAS_* & GOTO_* environment variables (Pleas see README)
268 * Imported GotoBLAS2 1.13 BSD version
271 * On x86 32bits, fixed a bug in zdot_sse2.S line 191. This would casue
272 zdotu & zdotc failures.Instead,Walk around it. (Refs issue #8 #9 on github)
273 * Modified ?axpy functions to return same netlib BLAS results
274 when incx==0 or incy==0 (Refs issue #7 on github)
275 * Modified ?swap functions to return same netlib BLAS results
276 when incx==0 or incy==0 (Refs issue #6 on github)
277 * Modified ?rot functions to return same netlib BLAS results
278 when incx==0 or incy==0 (Refs issue #4 on github)
279 * Detect Intel Westmere,Intel Clarkdale and Intel Arrandale
280 to use Nehalem codes.
281 * Fixed a typo bug about compiling dynamic ARCH library.
283 * Improve daxpy performance on ICT Loongson 3A.
284 * Supported ICT Loongson 3A CPU (Refs issue #1 on github)
285 ====================================================================