xhluca
diff --git a/‎doc/developers/advanced_installation.rst
Lines changed: 63 additions & 7 deletions b/‎doc/developers/advanced_installation.rst
Lines changed: 63 additions & 7 deletions
diff --git a/‎doc/modules/computing.rst
Lines changed: 18 additions & 0 deletions b/‎doc/modules/computing.rst
Lines changed: 18 additions & 0 deletions
@@ -35,8 +35,8 @@ Building from source
 Scikit-learn requires:
 
 - Python (>= 3.5),
-- NumPy (>= 1.11),
-- SciPy (>= 0.17).
+- NumPy (>= 1.8.2),
+- SciPy (>= 0.13.3).
 
 .. note::
 
@@ -46,7 +46,7 @@ Scikit-learn requires:
 
 Building Scikit-learn also requires
 
-- Cython >=0.28.5 
+- Cython >=0.23 
 
 Running tests requires
 
@@ -165,16 +165,25 @@ Windows
 To build scikit-learn on Windows you need a working C/C++ compiler in
 addition to numpy, scipy and setuptools.
 
-The building command depends on the architecture of the Python interpreter,
-32-bit or 64-bit. You can check the architecture by running the following in
-``cmd`` or ``powershell`` console::
+Picking the right compiler depends on the version of Python (2 or 3)
+and the architecture of the Python interpreter, 32-bit or 64-bit.
+You can check the Python version by running the following in ``cmd`` or
+``powershell`` console::
+
+    python --version
+
+and the architecture with::
 
     python -c "import struct; print(struct.calcsize('P') * 8)"
 
 The above commands assume that you have the Python installation folder in your
 PATH environment variable.
 
-You will need `Build Tools for Visual Studio 2017
+
+Python >= 3.5
+-------------
+
+For Python versions as of 3.5, you need `Build Tools for Visual Studio 2017
 <https://visualstudio.microsoft.com/de/downloads/>`_.
 
 For 64-bit 
10000
Python, configure the build environment with::
@@ -189,6 +198,53 @@ And build scikit-learn from this environment::
 Replace ``x64`` by ``x86`` to build for 32-bit Python.
 
 
+32-bit Python (<= 3.4)
+----------------------
+
+For 32-bit Python versions up to 3.4 use Microsoft Visual C++ Express 2010.
+
+Once installed you should be able to build scikit-learn without any
+particular configuration by running the following command in the scikit-learn
+folder::
+
+   python setup.py install
+
+
+64-bit Python (<= 3.4)
+----------------------
+
+For 64-bit Python versions up to 3.4, you either need the full Visual Studio or
+the free Windows SDKs that can be downloaded from the links below.
+
+The Windows SDKs include the MSVC compilers both for 32 and 64-bit
+architectures. They come as a ``GRMSDKX_EN_DVD.iso`` file that can be mounted
+as a new drive with a ``setup.exe`` installer in it.
+
+- For Python  you need SDK **v7.1**: `MS Windows SDK for Windows 7 and .NET
+  Framework 4
+  <https://www.microsoft.com/en-us/download/details.aspx?id=8442>`_
+
+Both SDKs can be installed in parallel on the same host. To use the Windows
+SDKs, you need to setup the environment of a ``cmd`` console launched with the
+following flags ::
+
+    cmd /E:ON /V:ON /K
+
+Then configure the build environment with::
+
+    SET DISTUTILS_USE_SDK=1
+    SET MSSdk=1
+    "C:\Program Files\Microsoft SDKs\Windows\v7.1\Setup\WindowsSdkVer.exe" -q -version:v7.1
+    "C:\Program Files\Microsoft SDKs\Windows\v7.1\Bin\SetEnv.cmd" /x64 /release
+
+Finally you can build scikit-learn in the same ``cmd`` console::
+
+    python setup.py install
+
+Replace ``/x64`` by ``/x86`` to build for 32-bit Python instead of 64-bit
+Python.
+
+
 Building binary packages and installers
 ---------------------------------------
 
 
@@ -430,6 +430,24 @@ and in this
 from Daniel Nouri which has some nice step by step install instructions for
 Debian / Ubuntu.
 
+.. warning::
+
+    Multithreaded BLAS libraries sometimes conflict with Python's
+    ``multiprocessing`` module, which is used by e.g. ``GridSearchCV`` and
+    most other estimators that take an ``n_jobs`` argument (with the exception
+    of ``SGDClassifier``, ``SGDRegressor``, ``Perceptron``,
+    ``PassiveAggressiveClassifier`` and tree-based methods such as random
+    forests). This is true of Apple's Accelerate and OpenBLAS when built with
+    OpenMP support.
+
+    Besides scikit-learn, NumPy and SciPy also use BLAS internally, as
+    explained earlier.
+
+    If you experience hanging subprocesses with ``n_jobs>1`` or ``n_jobs=-1``,
+    make sure you have a single-threaded BLAS library, or set ``n_jobs=1``,
+    or upgrade to Python 3.4 which has a new version of ``multiprocessing``
+    that should be immune to this problem.
+
 .. _working_memory:
 
 Limiting Working Memory