linux-numa.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Andreas Hollmann <hollmann@in.tum.de>
To: linux-numa <linux-numa@vger.kernel.org>
Subject: System uses numa Interleave mode as default policy?
Date: Sat, 4 Jan 2014 23:32:30 +0100	[thread overview]
Message-ID: <CAGz0_-3V4TSBn0juiLVfu4yFta8aNsqJYN7jBbi_jP_h2fAxpw@mail.gmail.com> (raw)

Hi,

it seems that my system (4-socket system) is
interleaving pages across all numa nodes.

The default should be local allocation?

I tried this out using the stream benchmark and pinned
the single thread to a single cpu with taskset.

numa_balancing is disabled.

Can I configure the default memory policy or is there
something wrong?

Configuration:

$ cat /proc/cmdline
BOOT_IMAGE=/boot/vmlinuz-linux
root=UUID=294db3db-d278-4b20-9199-f5ac6926efaf rw console=tty0
console=ttyS0,115200 numa_balancing=disable nomodeset quiet

$ uname -a
Linux inwest 3.12.6-1-ARCH #1 SMP PREEMPT Fri Dec 20 19:39:00 CET 2013
x86_64 GNU/Linux

Here is how i produced the output:

wget www.cs.virginia.edu/stream/FTP/Code/stream.c
gcc -mcmodel=medium -O -DSTREAM_ARRAY_SIZE=100000000 stream.c -o stream.100M

taskset -c 0 ./stream.100M 1> /dev/null & watch -n 1 numastat -cp $!

Per-node process memory usage (in MBs) for PID 2055 (stream.100M)
         Node 0 Node 1 Node 2 Node 3 Total
         ------ ------ ------ ------ -----
Huge          0      0      0      0     0
Heap        576    576    569    568  2289
Stack         0      0      0      0     0
Private       0      0      0      0     0
-------  ------ ------ ------ ------ -----
Total       576    576    569    568  2289

numactl -l taskset -c 0 ./stream.100M 1> /dev/null & watch -n 1 numastat -cp $!

Per-node process memory usage (in MBs) for PID 2333 (stream.100M)
         Node 0 Node 1 Node 2 Node 3 Total
         ------ ------ ------ ------ -----
Huge          0      0      0      0     0
Heap        576    573    568    572  2289
Stack         0      0      0      0     0
Private       0      0      0      0     0
-------  ------ ------ ------ ------ -----
Total       576    573    568    572  2289

numactl -m 0 taskset -c 0 ./stream.100M 1> /dev/null & watch -n 1
numastat -cp $!

Per-node process memory usage (in MBs) for PID 2394 (stream.100M)
         Node 0 Node 1 Node 2 Node 3 Total
         ------ ------ ------ ------ -----
Huge          0      0      0      0     0
Heap       2289      0      0      0  2289
Stack         0      0      0      0     0
Private       0      0      0      0     0
-------  ------ ------ ------ ------ -----
Total      2289      0      0      0  2289

---

$ numactl -s 0 taskset -c 0 ./stream.100M
policy: default
preferred node: current
physcpubind: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44
45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67
68 69 70 71 72 73 74 75 76 77 78 79
cpubind: 0 1 2 3
nodebind: 0 1 2 3
membind: 0 1 2 3

Thanks,
Andreas

                 reply	other threads:[~2014-01-04 22:32 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

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=CAGz0_-3V4TSBn0juiLVfu4yFta8aNsqJYN7jBbi_jP_h2fAxpw@mail.gmail.com \
    --to=hollmann@in.tum.de \
    --cc=linux-numa@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).