Intel DAAL AI加速——支持从数据预处理到模型预测，数据源必须使用DAAL的底层封装库...

数据源加速见官方文档（必须使用DAAL自己的库）：

Data Management

• • Numeric Tables
  • Tensors
  • Data Sources
  • Data Dictionaries
  • Data Serialization andDeserialization
  • Data Compression
  • Data Model

可以看到支持的数据源：同数据类型的table（matrix），不同类型的table，以及从DB文件取数据、数据序列化、压缩等。

在这些定制的数据源上，Intel DAAL使用自己底层的CPU进行硬件加速！下面摘自其官方：

Intel DAAL addresses all stages of the data analytics pipeline: preprocessing, transformation, analysis, modeling, validation, and decision-making.

Intel DAAL is developed by the same team as the Intel® Math Kernel Library (Intel® MKL)—the leading math library in the world. This team works closely with Intel® processor architects to squeeze performance from Intel processor-based systems.

Specs at a Glance

 

Processors	Intel Atom®, Intel Core™, Intel® Xeon®, and Intel® Xeon Phi™ processors and compatible processors
Languages	Python*, C++, Java*
Development Tools and Environments	Microsoft Visual Studio* (Windows*) Eclipse* and CDT* (Linux*)
Operating Systems	Use the same API for application development on multiple operating systems: Windows, Linux, and macOS*

统计特征的计算加速例子：

 

 

# file: low_order_moms_dense_batch.py
#===============================================================================
# Copyright 2014-2018 Intel Corporation.
#
# This software and the related documents are Intel copyrighted  materials,  and
# your use of  them is  governed by the  express license  under which  they were
# provided to you (License).  Unless the License provides otherwise, you may not
# use, modify, copy, publish, distribute,  disclose or transmit this software or
# the related documents without Intel's prior written permission.
#
# This software and the related documents  are provided as  is,  with no express
# or implied  warranties,  other  than those  that are  expressly stated  in the
# License.
#===============================================================================

## <a name="DAAL-EXAMPLE-PY-LOW_ORDER_MOMENTS_DENSE_BATCH"></a>
## \example low_order_moms_dense_batch.py

import os
import sys

from daal.algorithms import low_order_moments
from daal.data_management import FileDataSource, DataSourceIface

utils_folder = os.path.realpath(os.path.abspath(os.path.dirname(os.path.dirname(__file__))))
if utils_folder not in sys.path:
    sys.path.insert(0, utils_folder)
from utils import printNumericTable

DAAL_PREFIX = os.path.join('..', 'data')

# Input data set parameters
dataFileName = os.path.join(DAAL_PREFIX, 'batch', 'covcormoments_dense.csv')


def printResults(res):
    printNumericTable(res.get(low_order_moments.minimum),              "Minimum:")
    printNumericTable(res.get(low_order_moments.maximum),              "Maximum:")
    printNumericTable(res.get(low_order_moments.sum),                  "Sum:")
    printNumericTable(res.get(low_order_moments.sumSquares),           "Sum of squares:")
    printNumericTable(res.get(low_order_moments.sumSquaresCentered),   "Sum of squared difference from the means:")
    printNumericTable(res.get(low_order_moments.mean),                 "Mean:")
    printNumericTable(res.get(low_order_moments.secondOrderRawMoment), "Second order raw moment:")
    printNumericTable(res.get(low_order_moments.variance),             "Variance:")
    printNumericTable(res.get(low_order_moments.standardDeviation),    "Standard deviation:")
    printNumericTable(res.get(low_order_moments.variation),            "Variation:")

if __name__ == "__main__":

    # Initialize FileDataSource to retrieve input data from .csv file
    dataSource = FileDataSource(
        dataFileName,
        DataSourceIface.doAllocateNumericTable,
        DataSourceIface.doDictionaryFromContext
    )

    # Retrieve the data from input file
    dataSource.loadDataBlock()

    # Create algorithm for computing low order moments in batch processing mode
    algorithm = low_order_moments.Batch()

    # Set input arguments of the algorithm
    algorithm.input.set(low_order_moments.data, dataSource.getNumericTable())

    # Get computed low order moments
    res = algorithm.compute()

    printResults(res)　　

转载于:https://www.cnblogs.com/bonelee/p/9702982.html