aboutsummaryrefslogtreecommitdiffstats
path: root/erts/doc
diff options
context:
space:
mode:
authorSverker Eriksson <[email protected]>2013-06-19 18:34:56 +0200
committerSverker Eriksson <[email protected]>2013-06-19 18:51:06 +0200
commit5d67cc7f4c176ea6ab4d2538443d5fe264152861 (patch)
tree02629c584461df3b6f80d09a1de6298af4c3440c /erts/doc
parentf91c381afa2ebed6e37b49b4cc3f81bb9ca9e3eb (diff)
downloadotp-5d67cc7f4c176ea6ab4d2538443d5fe264152861.tar.gz
otp-5d67cc7f4c176ea6ab4d2538443d5fe264152861.tar.bz2
otp-5d67cc7f4c176ea6ab4d2538443d5fe264152861.zip
erts: Add new allocator strategy aoffcbf
with better performance than aoffcaobf as we don't have to rearrange the search tree when there are blocks of equal size.
Diffstat (limited to 'erts/doc')
-rw-r--r--erts/doc/src/erts_alloc.xml14
1 files changed, 12 insertions, 2 deletions
diff --git a/erts/doc/src/erts_alloc.xml b/erts/doc/src/erts_alloc.xml
index 2ffb55c6ab..5e008493b9 100644
--- a/erts/doc/src/erts_alloc.xml
+++ b/erts/doc/src/erts_alloc.xml
@@ -170,6 +170,15 @@
used. The time complexity is proportional to log N, where
N is the number of free blocks.</p>
</item>
+ <tag>Address order first fit carrier best fit</tag>
+ <item>
+ <p>Strategy: Find the <em>carrier</em> with the lowest address that
+ can satisfy the requested block size, then find a block within
+ that carrier using the "best fit" strategy.</p>
+ <p>Implementation: Balanced binary search trees are
+ used. The time complexity is proportional to log N, where
+ N is the number of free blocks.</p>
+ </item>
<tag>Address order first fit carrier address order best fit</tag>
<item>
<p>Strategy: Find the <em>carrier</em> with the lowest address that
@@ -330,7 +339,7 @@
fetched it will function as an ordinary carrier. This feature has
special requirements on the
<seealso marker="#M_as">allocation strategy</seealso> used. Currently
- only the <c>aoff</c> and the <c>aoffcaobf</c> strategies support
+ only the strategies <c>aoff</c>, <c>aoffcbf</c> and <c>aoffcaobf</c> support
abandoned carriers. This feature also requires
<seealso marker="#M_t">multiple thread specific instances</seealso>
to be enabled. When enabling this feature, multiple thread specific
@@ -340,10 +349,11 @@
allocators based on the <c>alloc_util</c> framework with the
exception of <c>temp_alloc</c> (which would be pointless).
</item>
- <tag><marker id="M_as"><c><![CDATA[+M<S>as bf|aobf|aoff|aoffcaobf|gf|af]]></c></marker></tag>
+ <tag><marker id="M_as"><c><![CDATA[+M<S>as bf|aobf|aoff|aoffcbf|aoffcaobf|gf|af]]></c></marker></tag>
<item>
Allocation strategy. Valid strategies are <c>bf</c> (best fit),
<c>aobf</c> (address order best fit), <c>aoff</c> (address order first fit),
+ <c>aoffcbf</c> (address order first fit carrier best fit),
<c>aoffcaobf</c> (address order first fit carrier address order best fit),
<c>gf</c> (good fit), and <c>af</c> (a fit). See
<seealso marker="#strategy">the description of allocation strategies</seealso> in "the <c>alloc_util</c> framework" section.</item>