2 ====================================================================
7 * Fixed the SEGFAULT bug about hyper-theading
8 * Support AMD Bulldozer by using GotoBLAS2 AMD Barcelona codes
10 ====================================================================
14 * Removed the limitation (64) of numbers of CPU cores.
15 Now, it supports 256 cores at max.
16 * Supported clang compiler.
17 * Fixed some build bugs on FreeBSD
19 * Optimized Level-3 BLAS on Intel Sandy Bridge x86-64 by AVX instructions.
20 Please use gcc >= 4.6 or clang >=3.1.
21 * Support AMD Bobcat by using GotoBLAS2 AMD Barcelona codes.
23 ====================================================================
27 * Upgraded LAPACK to 3.4.1 version. (Thank Zaheer Chothia)
28 * Supported LAPACKE, a C interface to LAPACKE. (Thank Zaheer Chothia)
29 * Fixed the build bug (MD5 and download) on Mac OSX.
30 * Auto download CUnit 2.1.2-2 from SF.net with UTEST_CHECK=1.
31 * Fxied the compatibility issue for compilers without C99 complex number
34 * Auto-detect Intel Sandy Bridge Core i7-3xxx & Xeon E7 Westmere-EX.
35 * Test alpha=Nan in dscale.
36 * Fixed a SEGFAULT bug in samax on x86 windows.
38 ====================================================================
42 * Set soname of shared library on Linux.
43 * Added LIBNAMESUFFIX flag in Makefile.rule. The user can use
44 this flag to control the library name, e.g. libopenblas.a,
45 libopenblas_ifort.a or libopenblas_omp.a.
46 * Added GEMM_MULTITHREAD_THRESHOLD flag in Makefile.rule.
47 The lib use single thread in GEMM function with small matrices.
49 * Used GEMV SSE/SSE2 kernels on x86 32-bit.
50 * Exported CBLAS functions in Windows DLL.
52 * Completed Level-3 BLAS optimization on Loongson 3A CPU.
53 * Improved GEMV performance on Loongson 3A CPU.
54 * Improved Level-3 BLAS performance on Loongson 3B CPU. (EXPERIMENT)
56 ====================================================================
60 * Fixed missing "#include <sched.h>" bug on Mac OS X.
61 Thank Mike Nolta for the patch.
62 * Upgraded LAPACK to 3.4.0 version
63 * Fixed a bug on Mac OS X. Don't require SystemStubs on OS X.
64 SystemStubs does not exist on Lion. Thank Stefan Karpinski.
65 * Improved README with using OpenMP. Check the internal threads
66 count less than or equal to omp_get_max_threads()
68 * Auto-detect Intel Core i6/i7 (Sandy Bridge) CPU with Nehalem assembly kernels
69 * Fixed some bugs on MingW 64-bit including zgemv, cdot, zdot.
71 ====================================================================
75 * Fixed a bug about installation. The header file "fblas77.h"
77 * Fixed #61 a building bug about setting TARGET and DYNAMIC_ARCH.
78 * Try to handle absolute path of shared library in OSX. (#57)
79 Thank Dr Kane O'Donnell.
80 * Changed the installation folder layout to $(PREFIX)/include and
84 * Fixed #58 zdot/xdot SEGFAULT bug with GCC-4.6 on x86. According
85 to i386 calling convention, The callee should remove the first
86 hidden parameter.Thank Mr. John for this patch.
88 ====================================================================
93 * Added DTB_ENTRIES into dynamic arch setting parameters. Now,
94 it can read DTB_ENTRIES on runtime. (Refs issue #55 on github)
96 ====================================================================
101 * Fixed a building bug when DYNAMIC_ARCH=1 & INTERFACE64=1.
102 (Refs issue #44 on github)
104 ====================================================================
109 * Stop the build and output the error message when detecting
110 fortran compiler failed. (Refs issue #42 on github)
112 ====================================================================
117 * Fixed blasint undefined bug in <cblas.h> file. Other software
118 could include this header successfully(Refs issue #13 on github)
119 * Fixed the SEGFAULT bug on 64 cores. On SMP server, the number
120 of CPUs or cores should be less than or equal to 64.(Refs issue #14
122 * Support "void goto_set_num_threads(int num_threads)" and "void
123 openblas_set_num_threads(int num_threads)" when USE_OPENMP=1
124 * Added extern "C" to support C++. Thank Tasio for the patch(Refs
126 * Provided an error message when the arch is not supported.(Refs
128 * Fixed issue #23. Fixed a bug of f_check script about generating link flags.
129 * Added openblas_set_num_threads for Fortran.
130 * Fixed #25 a wrong result of rotmg.
131 * Fixed a bug about detecting underscore prefix in c_check.
132 * Print the wall time (cycles) with enabling FUNCTION_PROFILE
133 * Fixed #35 a build bug with NO_LAPACK=1 & DYNAMIC_ARCH=1
134 * Added install target. You can use "make install". (Refs #20)
138 * Fixed #28 a wrong result of dsdot on x86_64.
139 * Fixed #32 a SEGFAULT bug of zdotc with gcc-4.6.
140 * Fixed #33 ztrmm bug on Nehalem.
141 * Walk round #27 the low performance axpy issue with small imput size & multithreads.
144 * Fixed #28 a wrong result of dsdot on Loongson3A/MIPS64.
145 * Optimized single/double precision BLAS Level3 on Loongson3A/MIPS64. (Refs #2)
146 * Optimized single/double precision axpy function on Loongson3A/MIPS64. (Refs #3)
148 ====================================================================
153 * Support "make NO_LAPACK=1" to build the library without
155 * Fixed randomly SEGFAULT when nodemask==NULL with above Linux 2.6.34.
156 Thank Mr.Ei-ji Nakama providing this patch. (Refs issue #12 on github)
157 * Added DEBUG=1 rule in Makefile.rule to build debug version.
158 * Disable compiling quad precision in reference BLAS library(netlib BLAS).
159 * Added unit testcases in utest/ subdir. Used CUnit framework.
160 * Supported OPENBLAS_* & GOTO_* environment variables (Pleas see README)
161 * Imported GotoBLAS2 1.13 BSD version
164 * On x86 32bits, fixed a bug in zdot_sse2.S line 191. This would casue
165 zdotu & zdotc failures.Instead,Walk around it. (Refs issue #8 #9 on github)
166 * Modified ?axpy functions to return same netlib BLAS results
167 when incx==0 or incy==0 (Refs issue #7 on github)
168 * Modified ?swap functions to return same netlib BLAS results
169 when incx==0 or incy==0 (Refs issue #6 on github)
170 * Modified ?rot functions to return same netlib BLAS results
171 when incx==0 or incy==0 (Refs issue #4 on github)
172 * Detect Intel Westmere,Intel Clarkdale and Intel Arrandale
173 to use Nehalem codes.
174 * Fixed a typo bug about compiling dynamic ARCH library.
176 * Improve daxpy performance on ICT Loongson 3A.
177 * Supported ICT Loongson 3A CPU (Refs issue #1 on github)
178 ====================================================================