'Re: [ast-users] Avoid clearing memory from |mmap(MAP_ANON)| ? / was: Re: Severe performance regressi'

[prev in list] [next in list] [prev in thread] [next in thread] 

List: ast-users
Subject: Re: [ast-users] Avoid clearing memory from |mmap(MAP_ANON)| ? / was: Re: Severe performance regressi
From: Glenn Fowler <glenn.s.fowler () gmail ! com>
Date: 2013-12-10 9:40:00
Message-ID: CAK449vAyLMuYnBqqgEBmRCb=xHL_YKBiGyhfuM+gAGCXdHdd_g () mail ! gmail ! com
[Download RAW message or body]

[Attachment #2 (multipart/alternative)]

I don't know how you are measuring excessive
with the alpha just posted there are no timing or memory diffs on linux for
segsizes 64Ki 4Ki 8Ki 64Mi
add usage to VMALLOC_OPTIONS to get a summary of total region usage in the
last line
all the segsizes show ~250Mi total usage

$ time VMALLOC_OPTIONS=segsize=64Ki $SHELL ./t01

real    0m14.61s
user    0m14.45s
sys     0m0.14s
$ time VMALLOC_OPTIONS=segsize=4Mi $SHELL ./t01

real    0m14.51s
user    0m14.36s
sys     0m0.14s
$ time VMALLOC_OPTIONS=segsize=8Mi $SHELL ./t01

real    0m14.68s
user    0m14.48s
sys     0m0.19s
$ time VMALLOC_OPTIONS=segsize=64Mi $SHELL ./t01

real    0m14.49s
user    0m14.33s
sys     0m0.14s
$ time VMALLOC_OPTIONS=segsize=64Mi,usage $SHELL ./t01
vmalloc: 0x25c2064e6000 67108864 init
vmalloc: 0x25c2064e6000 67108864 init
vmalloc: 0x25c2064e6000 67108864 region 0x007db200 size=268435456 segs=1
packs=1 busy=58% cache=115424/1801

real    0m15.03s
user    0m14.61s
sys     0m0.40s

On Tue, Dec 10, 2013 at 4:16 AM, Lionel Cons <lionelcons1972@gmail.com>wrote:

> On 6 December 2013 16:18, Glenn Fowler <glenn.s.fowler@gmail.com> wrote:
> > On Thu, Dec 5, 2013 at 6:41 PM, Roland Mainz <roland.mainz@nrubsig.org>
> > wrote:
> >> 2. The default block size by the normal |mmap(MAP_ANON)| allocator is
> >> 1MB. This is IMHO far to small because there is IMO not enough space
> >> for the coalescing algorithm to operate and a *lot* of fragmentation
> >> occurs.
> >> IMHO a _minimum_ page size of 4MB should be picked (as a side-effect
> >> the shell would get 4MB or 2MB largepages on platforms like Solaris
> >> automagically).
> >
> >
> > default block size upped to 4Mi and pagesize=<n>[KMGP][i] can ovveride in
> > VMALLOC_OPTIONS for testing
>
> A default block size of 4Mi is not sufficient on linux/x86-64 and
> Solaris/x86-64 to prevent the excessive memory consumption. I've
> played with script to create a graph to see where the excesses stop
> and the turning point is at 8.4Mi.
>
> So, if the default block size is bumped it should be bumped to at least
> 8Mi.
>
> Lionel
>

[Attachment #5 (text/html)]

<div dir="ltr"><div>I don&#39;t know how you are measuring excessive</div><div>with \
the alpha just posted there are no timing or memory diffs on linux for segsizes 64Ki \
4Ki 8Ki 64Mi</div><div>add usage to VMALLOC_OPTIONS to get a summary of total region \
usage in the last line</div> <div>all the segsizes show ~250Mi total \
usage</div><div><br></div><div>$ time VMALLOC_OPTIONS=segsize=64Ki $SHELL ./t01      \
</div><div><br></div><div>real    0m14.61s</div><div>user    0m14.45s</div><div>sys   \
0m0.14s</div> <div>$ time VMALLOC_OPTIONS=segsize=4Mi $SHELL ./t01      \
</div><div><br></div><div>real    0m14.51s</div><div>user    0m14.36s</div><div>sys   \
0m0.14s</div><div>$ time VMALLOC_OPTIONS=segsize=8Mi $SHELL ./t01</div><div> \
<br></div><div>real    0m14.68s</div><div>user    0m14.48s</div><div>sys     \
0m0.19s</div><div>$ time VMALLOC_OPTIONS=segsize=64Mi $SHELL \
./t01</div><div><br></div><div>real    0m14.49s</div><div>user    0m14.33s</div><div> \
sys     0m0.14s</div><div>$ time VMALLOC_OPTIONS=segsize=64Mi,usage $SHELL \
./t01</div><div>vmalloc: 0x25c2064e6000 67108864 init</div><div>vmalloc: \
0x25c2064e6000 67108864 init</div><div>vmalloc: 0x25c2064e6000 67108864 region \
0x007db200 size=268435456 segs=1 packs=1 busy=58% cache=115424/1801</div> \
<div><br></div><div>real    0m15.03s</div><div>user    0m14.61s</div><div>sys     \
0m0.40s</div><div><br></div></div><div class="gmail_extra"><br><br><div \
class="gmail_quote">On Tue, Dec 10, 2013 at 4:16 AM, Lionel Cons <span \
dir="ltr">&lt;<a href="mailto:lionelcons1972@gmail.com" \
target="_blank">lionelcons1972@gmail.com</a>&gt;</span> wrote:<br> <blockquote \
class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc \
solid;padding-left:1ex"><div class="im">On 6 December 2013 16:18, Glenn Fowler &lt;<a \
href="mailto:glenn.s.fowler@gmail.com">glenn.s.fowler@gmail.com</a>&gt; wrote:<br>

&gt; On Thu, Dec 5, 2013 at 6:41 PM, Roland Mainz &lt;<a \
href="mailto:roland.mainz@nrubsig.org">roland.mainz@nrubsig.org</a>&gt; &gt; \
wrote: </div><div class="im">&gt;&gt; 2. The default block size by the normal \
|mmap(MAP_ANON)| allocator is &gt;&gt; 1MB. This is IMHO far to small because \
there is IMO not enough space &gt;&gt; for the coalescing algorithm to operate \
and a *lot* of fragmentation &gt;&gt; occurs. 
&gt;&gt; IMHO a _minimum_ page size of 4MB should be picked (as a side-effect 
&gt;&gt; the shell would get 4MB or 2MB largepages on platforms like Solaris 
&gt;&gt; automagically). 
&gt; 
&gt; 
&gt; default block size upped to 4Mi and pagesize=&lt;n&gt;[KMGP][i] can ovveride \
in &gt; VMALLOC_OPTIONS for testing 
 
</div>A default block size of 4Mi is not sufficient on linux/x86-64 and 
Solaris/x86-64 to prevent the excessive memory consumption. I&#39;ve 
played with script to create a graph to see where the excesses stop 
and the turning point is at 8.4Mi. 
 
So, if the default block size is bumped it should be bumped to at least 8Mi. 
 
Lionel 
</blockquote></div> </div>

_______________________________________________
ast-users mailing list
ast-users@lists.research.att.com
http://lists.research.att.com/mailman/listinfo/ast-users

[prev in list] [next in list] [prev in thread] [next in thread] 