2 ====================================================================
6 * Set soname of shared library on Linux.
7 * Added LIBNAMESUFFIX flag in Makefile.rule. The user can use
8 this flag to control the library name, e.g. libopenblas.a,
9 libopenblas_ifort.a or libopenblas_omp.a.
10 * Added GEMM_MULTITHREAD_THRESHOLD flag in Makefile.rule.
11 The lib use single thread in GEMM function with small matrices.
13 * Used GEMV SSE/SSE2 kernels on x86 32-bit.
14 * Exported CBLAS functions in Windows DLL.
16 * Completed Level-3 BLAS optimization on Loongson 3A CPU.
17 * Improved GEMV performance on Loongson 3A CPU.
18 * Improved Level-3 BLAS performance on Loongson 3B CPU. (EXPERIMENT)
20 ====================================================================
24 * Fixed missing "#include <sched.h>" bug on Mac OS X.
25 Thank Mike Nolta for the patch.
26 * Upgraded LAPACK to 3.4.0 version
27 * Fixed a bug on Mac OS X. Don't require SystemStubs on OS X.
28 SystemStubs does not exist on Lion. Thank Stefan Karpinski.
29 * Improved README with using OpenMP. Check the internal threads
30 count less than or equal to omp_get_max_threads()
32 * Auto-detect Intel Core i6/i7 (Sandy Bridge) CPU with Nehalem assembly kernels
33 * Fixed some bugs on MingW 64-bit including zgemv, cdot, zdot.
35 ====================================================================
39 * Fixed a bug about installation. The header file "fblas77.h"
41 * Fixed #61 a building bug about setting TARGET and DYNAMIC_ARCH.
42 * Try to handle absolute path of shared library in OSX. (#57)
43 Thank Dr Kane O'Donnell.
44 * Changed the installation folder layout to $(PREFIX)/include and
48 * Fixed #58 zdot/xdot SEGFAULT bug with GCC-4.6 on x86. According
49 to i386 calling convention, The callee should remove the first
50 hidden parameter.Thank Mr. John for this patch.
52 ====================================================================
57 * Added DTB_ENTRIES into dynamic arch setting parameters. Now,
58 it can read DTB_ENTRIES on runtime. (Refs issue #55 on github)
60 ====================================================================
65 * Fixed a building bug when DYNAMIC_ARCH=1 & INTERFACE64=1.
66 (Refs issue #44 on github)
68 ====================================================================
73 * Stop the build and output the error message when detecting
74 fortran compiler failed. (Refs issue #42 on github)
76 ====================================================================
81 * Fixed blasint undefined bug in <cblas.h> file. Other software
82 could include this header successfully(Refs issue #13 on github)
83 * Fixed the SEGFAULT bug on 64 cores. On SMP server, the number
84 of CPUs or cores should be less than or equal to 64.(Refs issue #14
86 * Support "void goto_set_num_threads(int num_threads)" and "void
87 openblas_set_num_threads(int num_threads)" when USE_OPENMP=1
88 * Added extern "C" to support C++. Thank Tasio for the patch(Refs
90 * Provided an error message when the arch is not supported.(Refs
92 * Fixed issue #23. Fixed a bug of f_check script about generating link flags.
93 * Added openblas_set_num_threads for Fortran.
94 * Fixed #25 a wrong result of rotmg.
95 * Fixed a bug about detecting underscore prefix in c_check.
96 * Print the wall time (cycles) with enabling FUNCTION_PROFILE
97 * Fixed #35 a build bug with NO_LAPACK=1 & DYNAMIC_ARCH=1
98 * Added install target. You can use "make install". (Refs #20)
102 * Fixed #28 a wrong result of dsdot on x86_64.
103 * Fixed #32 a SEGFAULT bug of zdotc with gcc-4.6.
104 * Fixed #33 ztrmm bug on Nehalem.
105 * Walk round #27 the low performance axpy issue with small imput size & multithreads.
108 * Fixed #28 a wrong result of dsdot on Loongson3A/MIPS64.
109 * Optimized single/double precision BLAS Level3 on Loongson3A/MIPS64. (Refs #2)
110 * Optimized single/double precision axpy function on Loongson3A/MIPS64. (Refs #3)
112 ====================================================================
117 * Support "make NO_LAPACK=1" to build the library without
119 * Fixed randomly SEGFAULT when nodemask==NULL with above Linux 2.6.34.
120 Thank Mr.Ei-ji Nakama providing this patch. (Refs issue #12 on github)
121 * Added DEBUG=1 rule in Makefile.rule to build debug version.
122 * Disable compiling quad precision in reference BLAS library(netlib BLAS).
123 * Added unit testcases in utest/ subdir. Used CUnit framework.
124 * Supported OPENBLAS_* & GOTO_* environment variables (Pleas see README)
125 * Imported GotoBLAS2 1.13 BSD version
128 * On x86 32bits, fixed a bug in zdot_sse2.S line 191. This would casue
129 zdotu & zdotc failures.Instead,Walk around it. (Refs issue #8 #9 on github)
130 * Modified ?axpy functions to return same netlib BLAS results
131 when incx==0 or incy==0 (Refs issue #7 on github)
132 * Modified ?swap functions to return same netlib BLAS results
133 when incx==0 or incy==0 (Refs issue #6 on github)
134 * Modified ?rot functions to return same netlib BLAS results
135 when incx==0 or incy==0 (Refs issue #4 on github)
136 * Detect Intel Westmere,Intel Clarkdale and Intel Arrandale
137 to use Nehalem codes.
138 * Fixed a typo bug about compiling dynamic ARCH library.
140 * Improve daxpy performance on ICT Loongson 3A.
141 * Supported ICT Loongson 3A CPU (Refs issue #1 on github)
142 ====================================================================