check_blas.py

check_blas.py

D:\OpenCourses\ufldl\ref\Theano\theano\misc>python check_blas.py
Forcing DISTUTILS_USE_SDK=1
WARNING (theano.tensor.blas): Failed to import scipy.linalg.blas. Falling back o
n slower implementations (DLL load failed: 找不到指定的模块。)

        Some results that you can compare against. They were 10 executions
        of gemm in float64 with matrices of shape 2000x2000 (M=N=K=2000).
        All memory layout was in C order.

        CPU tested: Xeon E5345(2.33Ghz, 8M L2 cache, 1333Mhz FSB),
                    Xeon E5430(2.66Ghz, 12M L2 cache, 1333Mhz FSB),
                    Xeon E5450(3Ghz, 12M L2 cache, 1333Mhz FSB),
                    Xeon X5560(2.8Ghz, 12M L2 cache, hyper-threads?)
                    Core 2 E8500, Core i7 930(2.8Ghz, hyper-threads enabled),
                    Core i7 950(3.07GHz, hyper-threads enabled)
                    Xeon X5550(2.67GHz, 8M l2 cache?, hyper-threads enabled)


        Libraries tested:
            * numpy with ATLAS from distribution (FC9) package (1 thread)
            * manually compiled numpy and ATLAS with 2 threads
            * goto 1.26 with 1, 2, 4 and 8 threads
            * goto2 1.13 compiled with multiple threads enabled

                          Xeon   Xeon   Xeon  Core2 i7    i7     Xeon   Xeon
        lib/nb threads    E5345  E5430  E5450 E8500 930   950    X5560  X5550

        numpy 1.3.0 blas                                                775.92s
        numpy_FC9_atlas/1 39.2s  35.0s  30.7s 29.6s 21.5s 19.60s
        goto/1            18.7s  16.1s  14.2s 13.7s 16.1s 14.67s
        numpy_MAN_atlas/2 12.0s  11.6s  10.2s  9.2s  9.0s
        goto/2             9.5s   8.1s   7.1s  7.3s  8.1s  7.4s
        goto/4             4.9s   4.4s   3.7s  -     4.1s  3.8s
        goto/8             2.7s   2.4s   2.0s  -     4.1s  3.8s
        openblas/1                                        14.04s
        openblas/2                                         7.16s
        openblas/4                                         3.71s
        openblas/8                                         3.70s
        mkl 11.0.083/1            7.97s
        mkl 10.2.2.025/1                                         13.7s
        mkl 10.2.2.025/2                                          7.6s
        mkl 10.2.2.025/4                                          4.0s
        mkl 10.2.2.025/8                                          2.0s
        goto2 1.13/1                                                     14.37s
        goto2 1.13/2                                                      7.26s
        goto2 1.13/4                                                      3.70s
        goto2 1.13/8                                                      1.94s
        goto2 1.13/16                                                     3.16s

        Test time in float32

        cuda version      5.0    4.2    4.1    4.0    3.2    3.0   # note
        gpu
        K20m/ECC          0.07s
        K20/NOECC         0.07s
        M2070             0.25s         0.27s         0.32s
        M2050(Amazon)     0.25s
        C2075                    0.25s
        C1060                                         0.46s

        GTX Titan(D15U-50)0.06s  don't work
        GTX 680           0.12s  0.154s               0.218s
        GTX 580           0.16s  0.164s               0.203s
        GTX 480           0.19s  0.192s               0.237s 0.27s
        GTX 470           0.23s  0.238s               0.297s 0.34s
        GTX 660           0.20s  0.23s
        GTX 560                  0.30s
        GTX 650 Ti        0.27s
        GTX 460           0.37s                0.45s
        GTX 285                  0.452s        0.452s        0.40s # cuda 3.0 se
ems faster? driver version?
        GTX 550 Ti                             0.57s
        GT 520                   2.68s                3.06s
        520M              2.44s                       3.19s        # with bumble
bee on Ubuntu 12.04
        GT 220                                        3.80s
        GT 210                                 6.35s
        8500 GT                                              10.68s

Some Theano flags:
    blas.ldflags= -LC:\Anaconda\MinGW\lib -lopenblas
    compiledir= R:\compiledir_Windows-7-6.1.7601-SP1-Intel64_Family_6_Model_58_S
tepping_9_GenuineIntel-2.7.3-64
    floatX= float64
    device= cpu
Some environment variables:
    MKL_NUM_THREADS= None
    OMP_NUM_THREADS= None
    GOTO_NUM_THREADS= None

Numpy config: (used when the Theano flag "blas.ldflags" is empty)
lapack_opt_info:
    libraries = ['mkl_lapack95_lp64', 'mkl_core_dll', 'mkl_intel_lp64_dll', 'mkl
_intel_thread_dll']
    library_dirs = ['C:\\aroot\\stage\\libs']
    define_macros = [('SCIPY_MKL_H', None)]
    include_dirs = ['C:\\aroot\\stage\\include']
blas_opt_info:
    libraries = ['mkl_core_dll', 'mkl_intel_lp64_dll', 'mkl_intel_thread_dll']
    library_dirs = ['C:\\aroot\\stage\\libs']
    define_macros = [('SCIPY_MKL_H', None)]
    include_dirs = ['C:\\aroot\\stage\\include']
lapack_mkl_info:
    libraries = ['mkl_lapack95_lp64', 'mkl_core_dll', 'mkl_intel_lp64_dll', 'mkl
_intel_thread_dll']
    library_dirs = ['C:\\aroot\\stage\\libs']
    define_macros = [('SCIPY_MKL_H', None)]
    include_dirs = ['C:\\aroot\\stage\\include']
blas_mkl_info:
    libraries = ['mkl_core_dll', 'mkl_intel_lp64_dll', 'mkl_intel_thread_dll']
    library_dirs = ['C:\\aroot\\stage\\libs']
    define_macros = [('SCIPY_MKL_H', None)]
    include_dirs = ['C:\\aroot\\stage\\include']
mkl_info:
    libraries = ['mkl_core_dll', 'mkl_intel_lp64_dll', 'mkl_intel_thread_dll']
    library_dirs = ['C:\\aroot\\stage\\libs']
    define_macros = [('SCIPY_MKL_H', None)]
    include_dirs = ['C:\\aroot\\stage\\include']
Numpy dot module: numpy.core._dotblas
Numpy location: C:\Anaconda\lib\site-packages\numpy\__init__.pyc
Numpy version: 1.7.1
Traceback (most recent call last):
  File "check_blas.py", line 229, in <module>
    iters=options.iter, order=options.order)
  File "check_blas.py", line 71, in execute
    f = theano.function([], updates=[(c, 0.4 * c + .8 * T.dot(a, b))])
  File "D:\OpenCourses\ufldl\ref\Theano\theano\compile\function.py", line 222, i
n function
    profile=profile)
  File "D:\OpenCourses\ufldl\ref\Theano\theano\compile\pfunc.py", line 506, in p
func
    on_unused_input=on_unused_input)
  File "D:\OpenCourses\ufldl\ref\Theano\theano\compile\function_module.py", line
 1299, in orig_function
    defaults)
  File "D:\OpenCourses\ufldl\ref\Theano\theano\compile\function_module.py", line
 1168, in create
    _fn, _i, _o = self.linker.make_thunk(input_storage=input_storage_lists)
  File "D:\OpenCourses\ufldl\ref\Theano\theano\gof\link.py", line 382, in make_t
hunk
    output_storage = output_storage)[:3]
  File "D:\OpenCourses\ufldl\ref\Theano\theano\gof\vm.py", line 840, in make_all

    for node in order]
  File "D:\OpenCourses\ufldl\ref\Theano\theano\gof\op.py", line 589, in make_thu
nk
    output_storage=node_output_storage)
  File "D:\OpenCourses\ufldl\ref\Theano\theano\gof\cc.py", line 933, in make_thu
nk
    keep_lock=keep_lock)
  File "D:\OpenCourses\ufldl\ref\Theano\theano\gof\cc.py", line 876, in __compil
e__
    keep_lock=keep_lock)
  File "D:\OpenCourses\ufldl\ref\Theano\theano\gof\cc.py", line 1304, in cthunk_
factory
    key=key, fn=self.compile_cmodule_by_step, keep_lock=keep_lock)
  File "D:\OpenCourses\ufldl\ref\Theano\theano\gof\cmodule.py", line 992, in mod
ule_from_key
    module = next(compile_steps)
  File "D:\OpenCourses\ufldl\ref\Theano\theano\gof\cc.py", line 1221, in compile
_cmodule_by_step
    preargs=preargs)
  File "D:\OpenCourses\ufldl\ref\Theano\theano\gof\cmodule.py", line 1841, in co
mpile_str
    return dlimport(lib_filename)
  File "D:\OpenCourses\ufldl\ref\Theano\theano\gof\cmodule.py", line 280, in dli
mport
    rval = __import__(module_name, {}, {}, [module_name])
ImportError: ('DLL load failed: \xd5\xd2\xb2\xbb\xb5\xbd\xd6\xb8\xb6\xa8\xb5\xc4
\xc4\xa3\xbf\xe9\xa1\xa3', '[Gemm{inplace}(<TensorType(float64, matrix)>, Tensor
Constant{0.8}, <TensorType(float64, matrix)>, <TensorType(float64, matrix)>, Ten
sorConstant{0.4})]')