确保numpy正在mac pro上使用MKL库

我使用的是Enthound的Canopy/EPD版本的python，它附带了与MKL链接的numpy。我目前正在我的工作电脑（Windows7，四核i5@3.33GHz，4GB内存）和我的家庭工作站（MacPro 3.1，Ubuntu 12.04，2x四核Xeon@2.8GHz，6GB内存）上并行运行一组模拟（使用PP）

但是，当我对我的模拟进行基准测试时，它们在工作计算机上运行得更快（每次迭代35秒，而在mac pro上则是60秒）。这个问题在内核之间得到了完美的平衡（令人尴尬的并行问题），因此我怀疑家中linux工作站上的MKL库存在问题。是否有一种方法可以验证MKL库是否在python中实际使用。我读过一些线程，它们说您可以检查python是否链接到它，但它不能确保它被正确构建并实际使用

>>> numpy.show_config()

您将看到类似这样的输出，表明MKL确实是链接的

lapack_opt_info:
    libraries = ['mkl_lapack95_lp64', 'mkl_intel_lp64', 'mkl_intel_thread', 'mkl_core', 'mkl_mc', 'mkl_mc3', 'pthread']
    library_dirs = ['/Users/vagrant/src/master-env/Resources/Python.app/Contents/MacOS/../../../../lib']
    define_macros = [('SCIPY_MKL_H', None)]
    include_dirs = ['/Users/vagrant/src/master-env/Resources/Python.app/Contents/MacOS/../../../../include']
blas_opt_info:
    libraries = ['mkl_intel_lp64', 'mkl_intel_thread', 'mkl_core', 'mkl_mc', 'mkl_mc3', 'pthread']
    library_dirs = ['/Users/vagrant/src/master-env/Resources/Python.app/Contents/MacOS/../../../../lib']
    define_macros = [('SCIPY_MKL_H', None)]
    include_dirs = ['/Users/vagrant/src/master-env/Resources/Python.app/Contents/MacOS/../../../../include']
openblas_info:
NOT AVAILABLE
lapack_mkl_info:
    libraries = ['mkl_lapack95_lp64', 'mkl_intel_lp64', 'mkl_intel_thread', 'mkl_core', 'mkl_mc', 'mkl_mc3', 'pthread']
    library_dirs = ['/Users/vagrant/src/master-env/Resources/Python.app/Contents/MacOS/../../../../lib']
    define_macros = [('SCIPY_MKL_H', None)]
    include_dirs = ['/Users/vagrant/src/master-env/Resources/Python.app/Contents/MacOS/../../../../include']
blas_mkl_info:
    libraries = ['mkl_intel_lp64', 'mkl_intel_thread', 'mkl_core', 'mkl_mc', 'mkl_mc3', 'pthread']
    library_dirs = ['/Users/vagrant/src/master-env/Resources/Python.app/Contents/MacOS/../../../../lib']
    define_macros = [('SCIPY_MKL_H', None)]
    include_dirs = ['/Users/vagrant/src/master-env/Resources/Python.app/Contents/MacOS/../../../../include']
mkl_info:
    libraries = ['mkl_intel_lp64', 'mkl_intel_thread', 'mkl_core', 'mkl_mc', 'mkl_mc3', 'pthread']
    library_dirs = ['/Users/vagrant/src/master-env/Resources/Python.app/Contents/MacOS/../../../../lib']
    define_macros = [('SCIPY_MKL_H', None)]
    include_dirs = ['/Users/vagrant/src/master-env/Resources/Python.app/Contents/MacOS/../../../../include']

---

scipy.show\u config（）

不是您想要的吗？我不确定这是否算作“确保”它被实际使用。当我键入时，所有的“库目录”和“包含目录”都会说：“/home/vagrant/src/master env/…”。这是否正确（我的用户帐户不是“vagrant”，当我检查该文件夹时，它不在那里）。该字段是否重要，或是否未使用？库正确地报告了“mkl_lapack95_lp64”等。关于您的潜在问题，您是否考虑了物理核心与逻辑核心？您可以尝试调整环境变量MKL_NUM_THREADS以避免可能的颠簸。我会再看一遍，但我正在执行的操作不是多线程的（很多add（）、multiply（）等，但没有dot（）。此外，我不确定为什么XEON上会出现问题，而i5上不会出现问题。这些路径是硬编码列表，您可以在config.py中找到，而不一定是numpy使用的路径。