svn commit: r47018 - head/en_US.ISO8859-1/htdocs/news/status
Warren Block
wblock at FreeBSD.org
Sun Jul 19 19:27:27 UTC 2015
Author: wblock
Date: Sun Jul 19 19:27:26 2015
New Revision: 47018
URL: https://svnweb.freebsd.org/changeset/doc/47018
Log:
Restore mysteriously lost portion of PCID report, add URL to repository
commit.
Modified:
head/en_US.ISO8859-1/htdocs/news/status/report-2015-04-2015-06.xml
Modified: head/en_US.ISO8859-1/htdocs/news/status/report-2015-04-2015-06.xml
==============================================================================
--- head/en_US.ISO8859-1/htdocs/news/status/report-2015-04-2015-06.xml Sun Jul 19 18:50:12 2015 (r47017)
+++ head/en_US.ISO8859-1/htdocs/news/status/report-2015-04-2015-06.xml Sun Jul 19 19:27:26 2015 (r47018)
@@ -1085,6 +1085,11 @@
</person>
</contact>
+ <links>
+ <url href="https://svnweb.freebsd.org/base?view=revision&revision=282684">Commit
+ r282684</url>
+ </links>
+
<body>
<p>Process-Context Identifiers (PCIDs) is a feature of the
Translation Lookaside Buffer (TLB) on Intel processors,
@@ -1128,7 +1133,28 @@
performed. As result, the new implementation issues exactly
the same amount of shootdown IPIs as pmap which does not
utilize PCID.</p>
+
+ <p>Another change included with the PCID rewrite is a move of
+ the address space switching code from assembler to C source,
+ making the algorithm easier to understand and validate.</p>
+
+ <p>Measurements done with <tt>hwpmc(4)</tt> on a Haswell machine
+ indicated that the new implementation reduced the amount of
+ data TLB misses up to 10 times, without an impact on the IPI
+ counters.</p>
+
+ <p>The rewrite was committed to HEAD at r282684.</p>
+
+ <p>Note: AMD processors do not have PCID feature for host paging
+ (AMD provides ASIDs for SVM use). But it is likely that AMD
+ processors do cache TLB translations for different address
+ spaces transparently, and snoop writes to the page tables to
+ invalidate the caches.</p>
</body>
+
+ <sponsor>
+ The FreeBSD Foundation
+ </sponsor>
</project>
<project cat='proj'>
More information about the svn-doc-all
mailing list