software <hugin-ptx@googlegroups.com>
Date: Tue, 1 May 2012 01:48:58 -0700 (PDT)
From: Bart van Andel <bavanan...@gmail.com>
To: hugin-ptx@googlegroups.com
Message-ID: <15918111.967.1335862139020.JavaMail.geo-discussion-forums@ynep10>
In-Reply-To: <13163966.101.1335770022621.JavaMail.geo-discussion-forums@ynei5>
References: <adcea9f3-890f-4515-8e85-8da6571bf377@v1g2000yqm.googlegroups.com> <20120428210351.GB8420@postle.net> <4F9C8AA8.6060902@gmail.com> <6057150.1887.1335692484330.JavaMail.geo-discussion-forums@ynjw9>
 <4F9E3020.9000504@gmail.com>
 <13163966.101.1335770022621.JavaMail.geo-discussion-forums@ynei5>
Subject: Re: [hugin-ptx] merge enblend and multiblend code?
MIME-Version: 1.0
Content-Type: multipart/mixed; 
	boundary="----=_Part_965_21062827.1335862139018"

------=_Part_965_21062827.1335862139018
Content-Type: multipart/alternative; 
	boundary="----=_Part_966_3690397.1335862139018"

------=_Part_966_3690397.1335862139018
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit



On Monday, April 30, 2012 9:13:42 AM UTC+2, Monkey wrote:
>
> I did try multi-threading it in a couple of places, but maybe it was doing 
> it wrong because it didn't make it any quicker. That was with native 
> Windows multithreading - unfortunately Microsoft don't want you to use 
> OpenMP with Visual C++ Express, which is the only option now multiblend is 
> multi-platform. 


Fortunately, cross-compiling using MXE is pretty easy as I've shown in 
another thread [0, 1]. I haven't actually tried compiling anything with 
OpenMP enabled using this approach, but I don't think it should be too much 
of a hassle. And there's a great support community there so if I need help 
all I need to do is ask.

[0] https://groups.google.com/forum/?fromgroups#!topic/hugin-ptx/-3-i9ynuCig
[1] https://groups.google.com/d/msg/hugin-ptx/JPiViZQ-Ycw/4ygfvPVq4hgJ

 

> Besides which, the seaming algorithm can't be multithreaded (at least not 
> easily), and that and image loading are usually the two slowest parts.
>
> David
>
> On Monday, April 30, 2012 7:24:32 AM UTC+1, GnomeNomad wrote:
>>
>> Thanks, I thought it was multi-threaded. 
>>
>> On 04/28/2012 11:41 PM, Monkey wrote: 
>> > The same benefit it has on any system - it's quicker (but not better) 
>> > than Enblend :). The "multi" in multiblend doesn't have anything to do 
>> > with multicore processing - it's (currently) single-threaded. 
>> > 
>> > On Sunday, April 29, 2012 1:26:16 AM UTC+1, GnomeNomad wrote: 
>> > 
>> >     What benefit does multiblend have on uniprocessor systems? 
>>
>> -- 
>> Gnome Nomad 
>> gnomeno...@gmail.com 
>> wandering the landscape of god 
>> http://www.clanjones.org/david/ 
>> http://dancing-treefrog.deviantart.com/ 
>> http://www.cafepress.com/otherend/ 
>>
>
------=_Part_966_3690397.1335862139018
Content-Type: text/html; charset=utf-8
Content-Transfer-Encoding: quoted-printable

<br><br>On Monday, April 30, 2012 9:13:42 AM UTC+2, Monkey wrote:<blockquot=
e class=3D"gmail_quote" style=3D"margin: 0;margin-left: 0.8ex;border-left: =
1px #ccc solid;padding-left: 1ex;">I did try multi-threading it in a couple=
 of places, but maybe it was doing it wrong because it didn't make it any q=
uicker. That was with native Windows multithreading - unfortunately Microso=
ft don't want you to use OpenMP with Visual C++ Express, which is the only =
option now multiblend is multi-platform. </blockquote><div><br></div><div>F=
ortunately, cross-compiling using MXE is pretty easy as I've shown in anoth=
er thread [0, 1]. I haven't actually tried compiling anything with OpenMP e=
nabled using this approach, but I don't think it should be too much of a ha=
ssle. And there's a great support community there so if I need help all I n=
eed to do is ask.</div><div><br></div><div>[0]&nbsp;<a href=3D"https://grou=
ps.google.com/forum/?fromgroups#!topic/hugin-ptx/-3-i9ynuCig">https://group=
s.google.com/forum/?fromgroups#!topic/hugin-ptx/-3-i9ynuCig</a></div><div>[=
1]&nbsp;<a href=3D"https://groups.google.com/d/msg/hugin-ptx/JPiViZQ-Ycw/4y=
gfvPVq4hgJ" target=3D"_blank" style=3D"cursor: pointer; ">https://groups.go=
ogle.com/<wbr>d/msg/hugin-ptx/JPiViZQ-Ycw/<wbr>4ygfvPVq4hgJ</a></div><div><=
br></div><div>&nbsp;</div><blockquote class=3D"gmail_quote" style=3D"margin=
: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;">Besi=
des which, the seaming algorithm can't be multithreaded (at least not easil=
y), and that and image loading are usually the two slowest parts.<br><br>Da=
vid<br><br>On Monday, April 30, 2012 7:24:32 AM UTC+1, GnomeNomad wrote:<bl=
ockquote class=3D"gmail_quote" style=3D"margin:0;margin-left:0.8ex;border-l=
eft:1px #ccc solid;padding-left:1ex">Thanks, I thought it was multi-threade=
d.
<br>
<br>On 04/28/2012 11:41 PM, Monkey wrote:
<br>&gt; The same benefit it has on any system - it's quicker (but not bett=
er)
<br>&gt; than Enblend :). The "multi" in multiblend doesn't have anything t=
o do
<br>&gt; with multicore processing - it's (currently) single-threaded.
<br>&gt;
<br>&gt; On Sunday, April 29, 2012 1:26:16 AM UTC+1, GnomeNomad wrote:
<br>&gt;
<br>&gt; &nbsp; &nbsp; What benefit does multiblend have on uniprocessor sy=
stems?
<br>
<br>--=20
<br>Gnome Nomad
<br><a href=3D"mailto:gnomeno...@gmail.com" target=3D"_blank">gnomenomad@gm=
ail.com</a>
<br>wandering the landscape of god
<br><a href=3D"http://www.clanjones.org/david/" target=3D"_blank">http://ww=
w.clanjones.org/<wbr>david/</a>
<br><a href=3D"http://dancing-treefrog.deviantart.com/" target=3D"_blank">h=
ttp://dancing-treefrog.<wbr>deviantart.com/</a>
<br><a href=3D"http://www.cafepress.com/otherend/" target=3D"_blank">http:/=
/www.cafepress.com/<wbr>otherend/</a>
<br></blockquote></blockquote>
------=_Part_966_3690397.1335862139018--

------=_Part_965_21062827.1335862139018--