84 lines
3.2 KiB
Plaintext
84 lines
3.2 KiB
Plaintext
OpenBLAS Readme
|
|
|
|
1.Introduction
|
|
OpenBLAS is an optimized BLAS library based on GotoBLAS2 1.13 BSD version. OpenBLAS is an open source project supported by Lab of Parallel Software and Computational Science, ISCAS.(http://www.rdcps.ac.cn)
|
|
|
|
2.Intallation
|
|
Download from project homepage. http://xianyi.github.com/OpenBLAS/
|
|
Or,
|
|
check out codes from git://github.com/xianyi/OpenBLAS.git
|
|
1)Normal compile
|
|
(a) type "make" to detect the CPU automatically.
|
|
or
|
|
(b) type "make TARGET=xxx" to set target CPU, e.g. "make TARGET=NEHALEM". The full target list is in file TargetList.txt.
|
|
|
|
2)Cross compile
|
|
Please set CC and FC with the cross toolchains. Then, set HOSTCC with your host C compiler. At last, set TARGET explicitly.
|
|
|
|
examples:
|
|
On X86 box, compile this library for loongson3a CPU.
|
|
make BINARY=64 CC=mips64el-unknown-linux-gnu-gcc FC=mips64el-unknown-linux-gnu-gfortran HOSTCC=gcc TARGET=LOONGSON3A
|
|
|
|
3)Debug version
|
|
make DEBUG=1
|
|
|
|
4)Intall to the directory (Optional)
|
|
e.g.
|
|
make install PREFIX=your_installation_directory
|
|
The default directory is /opt/OpenBLAS
|
|
|
|
3.Support CPU & OS
|
|
Please read GotoBLAS_01Readme.txt
|
|
|
|
Additional support CPU:
|
|
x86_64:
|
|
Intel Xeon 56xx (Westmere) //Used GotoBLAS2 Nehalem codes.
|
|
MIPS64:
|
|
ICT Loongson 3A //The initial version used GotoBLAS2 MIPS64 kernels. Thus, the performance is not good.
|
|
|
|
4.Usages
|
|
Link with libopenblas.a or -lopenblas for shared library.
|
|
|
|
4.1 Set the number of threads with environment variables. for example,
|
|
export OPENBLAS_NUM_THREADS=4
|
|
or
|
|
export GOTO_NUM_THREADS=4
|
|
or
|
|
export OMP_NUM_THREADS=4
|
|
|
|
The priorities are OPENBLAS_NUM_THREADS > GOTO_NUM_THREADS > OMP_NUM_THREADS.
|
|
|
|
If you compile this lib with USE_OPENMP=1, you should only set OMP_NUM_THREADS environment variable.
|
|
|
|
4.2 Set the number of threads with calling functions. for example,
|
|
void goto_set_num_threads(int num_threads);
|
|
or
|
|
void openblas_set_num_threads(int num_threads);
|
|
|
|
If you compile this lib with USE_OPENMP=1, you should use the above functions, too.
|
|
|
|
5.Report Bugs
|
|
Please add a issue in https://github.com/xianyi/OpenBLAS/issues
|
|
|
|
6.To-Do List:
|
|
Optimization on ICT Loongson 3A CPU
|
|
|
|
7.Contact
|
|
OpenBLAS users mailing list: http://list.rdcps.ac.cn/mailman/listinfo/openblas
|
|
|
|
8.ChangeLog
|
|
Please see Changelog.txt to obtain the differences between GotoBLAS2 1.13 BSD version.
|
|
|
|
9.Known Issues
|
|
* The number of CPUs/Cores should less than or equal to 8*sizeof(unsigned long). On 64 bits, the limit
|
|
is 64. On 32 bits, it is 32.
|
|
* This library is not compatible with EKOPath Compiler Suite 4.0.10 (http://www.pathscale.com/ekopath-compiler-suite). However, Path64 (https://github.com/path64/compiler) could compile the codes successfully.
|
|
|
|
10. Specification of Git Branches
|
|
We used the git branching model in this article (http://nvie.com/posts/a-successful-git-branching-model/).
|
|
Now, there are 4 branches in github.com.
|
|
* The master branch. This a main branch to reflect a production-ready state.
|
|
* The develop branch. This a main branch to reflect a state with the latest delivered development changes for the next release.
|
|
* The loongson3a branch. This is a feature branch. We develop Loongson3A codes on this branch. We will merge this feature to develop branch in future.
|
|
* The gh-pages branch. This is for web pages
|