From: Anton Blanchard <anton@samba.org>
To: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Cc: netdev@vger.kernel.org, benh@kernel.crashing.org
Subject: [PATCH] ixgbe: Reduce memory consumption with larger page sizes
Date: Tue, 22 Oct 2013 10:37:57 +1100 [thread overview]
Message-ID: <20131022103757.162f1a79@kryten> (raw)
The ixgbe driver allocates pages for its receive rings. It currently
uses 512 pages, regardless of page size. During receive handling it
adds the unused part of the page back into the rx ring, avoiding the
need for a new allocation.
On a ppc64 box with 64 threads and 64kB pages, we end up with
512 entries * 64 rx queues * 64kB = 2GB memory used. Even more of a
concern is that we use up 2GB of IOMMU space in order to map all this
memory.
The driver makes a number of decisions based on if PAGE_SIZE is less
than 8kB, so use this as the breakpoint and only allocate 128 entries
on 8kB or larger page sizes.
Signed-off-by: Anton Blanchard <anton@samba.org>
---
Jeff: The breakpoint and the ring size I chose was pretty arbitrary,
feel free to adjust as you see fit. Our main concern is we get that 2GB
consumption down to something more reasonable :)
Index: b/drivers/net/ethernet/intel/ixgbe/ixgbe.h
===================================================================
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe.h
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe.h
@@ -67,7 +67,11 @@
#define IXGBE_MAX_TXD 4096
#define IXGBE_MIN_TXD 64
+#if (PAGE_SIZE < 8192)
#define IXGBE_DEFAULT_RXD 512
+#else
+#define IXGBE_DEFAULT_RXD 128
+#endif
#define IXGBE_MAX_RXD 4096
#define IXGBE_MIN_RXD 64
next reply other threads:[~2013-10-21 23:37 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-10-21 23:37 Anton Blanchard [this message]
2013-10-22 10:23 ` [PATCH] ixgbe: Reduce memory consumption with larger page sizes Jeff Kirsher
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20131022103757.162f1a79@kryten \
--to=anton@samba.org \
--cc=benh@kernel.crashing.org \
--cc=jeffrey.t.kirsher@intel.com \
--cc=netdev@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).