Commit c0f96bdc authored by dterei's avatar dterei
Browse files

Document -fvectorise and -favoid-vect. (#5801)

parent d0b823c0
......@@ -185,7 +185,7 @@ nfib n | n <= 1 = 1
parallel <literal>map</literal>.</para>
</sect2>
<sect2><title>Data Parallel Haskell</title>
<sect2 id="dph"><title>Data Parallel Haskell</title>
<para>GHC includes experimental support for Data Parallel Haskell (DPH). This code
is highly unstable and is only provided as a technology preview. More
information can be found on the corresponding <ulink
......
......@@ -2278,9 +2278,15 @@ last (x : xs) = last' x xs
<indexterm><primary><option></option></primary></indexterm>
</term>
<listitem>
<para>Data Parallel Haskell.
</para>
TODO: Document optimisation
<para>Part of <link linkend="dph">Data Parallel Haskell
(DPH)</link>.</para>
<para><emphasis>Off by default.</emphasis> Enable the
<emphasis>vectorisation</emphasis> optimisation transformation. This
optimisation transforms the nested data parallelism code of programs
using DPH into flat data parallelism. Flat data parallel programs
should have better load balancing, enable SIMD parallelism and
friendlier cache behaviour.</para>
</listitem>
</varlistentry>
......@@ -2290,9 +2296,19 @@ last (x : xs) = last' x xs
<indexterm><primary><option></option></primary></indexterm>
</term>
<listitem>
<para>Data Parallel Haskell.
</para>
TODO: Document optimisation
<para>Part of <link linkend="dph">Data Parallel Haskell
(DPH)</link>.</para>
<para><emphasis>Off by default.</emphasis> Enable the
<emphasis>vectorisation</emphasis> avoidance optimisation. This
optimisation only works when used in combination with the
<option>-fvectorise</option> transformation.</para>
<para>While vectorisation of code using DPH is often a big win, it
can also produce worse results for some kinds of code. This
optimisation modifies the vectorisation transformation to try to
determine if a function would be better of unvectorised and if
so, do just that.</para>
</listitem>
</varlistentry>
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment