project-base/server-skynet-source-3rd-jemalloc
2
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Simplify small size class infrastructure.

Browse Source 
Program-generate small size class tables for all valid combinations of
LG_TINY_MIN, LG_QUANTUM, and PAGE_SHIFT.  Use the appropriate table to generate
all relevant data structures, and remove the distinction between
tiny/quantum/cacheline/subpage bins.

Remove --enable-dynamic-page-shift.  This option didn't prove useful in
practice, and it prevented optimizations.

Add Tilera architecture support.
This commit is contained in:
Jason Evans
2012-02-28 16:50:47 -08:00
parent 5389146191
commit b172610317
17 changed files with 327 additions and 858 deletions
Download Patch File Download Diff File Expand all files Collapse all files

198
doc/jemalloc.xml.in
View File

@@ -458,20 +458,11 @@ for (i = 0; i < nbins; i++) {
a frontier and free list to track which regions are in use. Allocation
requests that are no more than half the quantum (8 or 16, depending on
architecture) are rounded up to the nearest power of two that is at least
<code language="C">sizeof(<type>void *</type>)</code>. Allocation requests
that are more than half the quantum, but no more than the minimum
cacheline-multiple size class (see the <link
linkend="opt.lg_qspace_max"><mallctl>opt.lg_qspace_max</mallctl></link>
option) are rounded up to the nearest multiple of the quantum. Allocation
requests that are more than the minimum cacheline-multiple size class, but
no more than the minimum subpage-multiple size class (see the <link
linkend="opt.lg_cspace_max"><mallctl>opt.lg_cspace_max</mallctl></link>
option) are rounded up to the nearest multiple of the cacheline size (64).
Allocation requests that are more than the minimum subpage-multiple size
class, but no more than the maximum subpage-multiple size class are rounded
up to the nearest multiple of the subpage size (256). Allocation requests
that are more than the maximum subpage-multiple size class, but small
enough to fit in an arena-managed chunk (see the <link
<code language="C">sizeof(<type>double</type>)</code>. All other small
object size classes are multiples of the quantum, spaced such that internal
fragmentation is limited to approximately 25% for all but the smallest size
classes. Allocation requests that are larger than the maximum small size
class, but small enough to fit in an arena-managed chunk (see the <link
linkend="opt.lg_chunk"><mallctl>opt.lg_chunk</mallctl></link> option), are
rounded up to the nearest run size. Allocation requests that are too large
to fit in an arena-managed chunk are rounded up to the nearest multiple of
@@ -507,16 +498,28 @@ for (i = 0; i < nbins; i++) {
<entry>[8]</entry>
</row>
<row>
<entry>Quantum-spaced</entry>
<entry>16-spaced</entry>
<entry>[16, 32, 48, ..., 128]</entry>
</row>
<row>
<entry>Cacheline-spaced</entry>
<entry>[192, 256, 320, ..., 512]</entry>
<entry>32-spaced</entry>
<entry>[160, 192, 224, 256]</entry>
</row>
<row>
<entry>Subpage-spaced</entry>
<entry>[768, 1024, 1280, ..., 3840]</entry>
<entry>64-spaced</entry>
<entry>[320, 384, 448, 512]</entry>
</row>
<row>
<entry>128-spaced</entry>
<entry>[640, 768, 896, 1024]</entry>
</row>
<row>
<entry>256-spaced</entry>
<entry>[1280, 1536, 1792, 2048]</entry>
</row>
<row>
<entry>512-spaced</entry>
<entry>[2560, 3072, 3584]</entry>
</row>
<row>
<entry namest="c1" nameend="c2">Large</entry>
@@ -714,30 +717,6 @@ for (i = 0; i < nbins; i++) {
</para></listitem>
</varlistentry>
<varlistentry id="opt.lg_qspace_max">
<term>
<mallctl>opt.lg_qspace_max</mallctl>
(<type>size_t</type>)
<literal>r-</literal>
</term>
<listitem><para>Size (log base 2) of the maximum size class that is a
multiple of the quantum (8 or 16 bytes, depending on architecture).
Above this size, cacheline spacing is used for size classes. The
default value is 128 bytes (2^7).</para></listitem>
</varlistentry>
<varlistentry id="opt.lg_cspace_max">
<term>
<mallctl>opt.lg_cspace_max</mallctl>
(<type>size_t</type>)
<literal>r-</literal>
</term>
<listitem><para>Size (log base 2) of the maximum size class that is a
multiple of the cacheline size (64). Above this size, subpage spacing
(256 bytes) is used for size classes. The default value is 512 bytes
(2^9).</para></listitem>
</varlistentry>
<varlistentry id="opt.lg_chunk">
<term>
<mallctl>opt.lg_chunk</mallctl>
@@ -1178,24 +1157,6 @@ malloc_conf = "xmalloc:true";]]></programlisting>
<listitem><para>Quantum size.</para></listitem>
</varlistentry>
<varlistentry>
<term>
<mallctl>arenas.cacheline</mallctl>
(<type>size_t</type>)
<literal>r-</literal>
</term>
<listitem><para>Assumed cacheline size.</para></listitem>
</varlistentry>
<varlistentry>
<term>
<mallctl>arenas.subpage</mallctl>
(<type>size_t</type>)
<literal>r-</literal>
</term>
<listitem><para>Subpage size class interval.</para></listitem>
</varlistentry>
<varlistentry>
<term>
<mallctl>arenas.pagesize</mallctl>
@@ -1214,80 +1175,6 @@ malloc_conf = "xmalloc:true";]]></programlisting>
<listitem><para>Chunk size.</para></listitem>
</varlistentry>
<varlistentry>
<term>
<mallctl>arenas.tspace_min</mallctl>
(<type>size_t</type>)
<literal>r-</literal>
</term>
<listitem><para>Minimum tiny size class. Tiny size classes are powers
of two.</para></listitem>
</varlistentry>
<varlistentry>
<term>
<mallctl>arenas.tspace_max</mallctl>
(<type>size_t</type>)
<literal>r-</literal>
</term>
<listitem><para>Maximum tiny size class. Tiny size classes are powers
of two.</para></listitem>
</varlistentry>
<varlistentry>
<term>
<mallctl>arenas.qspace_min</mallctl>
(<type>size_t</type>)
<literal>r-</literal>
</term>
<listitem><para>Minimum quantum-spaced size class.</para></listitem>
</varlistentry>
<varlistentry>
<term>
<mallctl>arenas.qspace_max</mallctl>
(<type>size_t</type>)
<literal>r-</literal>
</term>
<listitem><para>Maximum quantum-spaced size class.</para></listitem>
</varlistentry>
<varlistentry>
<term>
<mallctl>arenas.cspace_min</mallctl>
(<type>size_t</type>)
<literal>r-</literal>
</term>
<listitem><para>Minimum cacheline-spaced size class.</para></listitem>
</varlistentry>
<varlistentry>
<term>
<mallctl>arenas.cspace_max</mallctl>
(<type>size_t</type>)
<literal>r-</literal>
</term>
<listitem><para>Maximum cacheline-spaced size class.</para></listitem>
</varlistentry>
<varlistentry>
<term>
<mallctl>arenas.sspace_min</mallctl>
(<type>size_t</type>)
<literal>r-</literal>
</term>
<listitem><para>Minimum subpage-spaced size class.</para></listitem>
</varlistentry>
<varlistentry>
<term>
<mallctl>arenas.sspace_max</mallctl>
(<type>size_t</type>)
<literal>r-</literal>
</term>
<listitem><para>Maximum subpage-spaced size class.</para></listitem>
</varlistentry>
<varlistentry>
<term>
<mallctl>arenas.tcache_max</mallctl>
@@ -1298,52 +1185,13 @@ malloc_conf = "xmalloc:true";]]></programlisting>
<listitem><para>Maximum thread-cached size class.</para></listitem>
</varlistentry>
<varlistentry>
<term>
<mallctl>arenas.ntbins</mallctl>
(<type>unsigned</type>)
<literal>r-</literal>
</term>
<listitem><para>Number of tiny bin size classes.</para></listitem>
</varlistentry>
<varlistentry>
<term>
<mallctl>arenas.nqbins</mallctl>
(<type>unsigned</type>)
<literal>r-</literal>
</term>
<listitem><para>Number of quantum-spaced bin size
classes.</para></listitem>
</varlistentry>
<varlistentry>
<term>
<mallctl>arenas.ncbins</mallctl>
(<type>unsigned</type>)
<literal>r-</literal>
</term>
<listitem><para>Number of cacheline-spaced bin size
classes.</para></listitem>
</varlistentry>
<varlistentry>
<term>
<mallctl>arenas.nsbins</mallctl>
(<type>unsigned</type>)
<literal>r-</literal>
</term>
<listitem><para>Number of subpage-spaced bin size
classes.</para></listitem>
</varlistentry>
<varlistentry>
<term>
<mallctl>arenas.nbins</mallctl>
(<type>unsigned</type>)
<literal>r-</literal>
</term>
<listitem><para>Total number of bin size classes.</para></listitem>
<listitem><para>Number of bin size classes.</para></listitem>
</varlistentry>
<varlistentry>