Re: newbie curious of general relative performance

--Boundary_(ID_/+qQlVWiepkTsmtXvZtMRg)
Content-type: text/plain; charset=utf-8
Content-transfer-encoding: quoted-printable
Content-disposition: inline

Io is a slow language. The implementation isn't tuned for performance, it's=
 tuned for flexibility. There are a few facts which make Io slow:

* Almost no optimizations above a micro level
* Primitive numbers in the range of -10 ~ +250 (iirc) are cached at interpr=
eter startup, but any numbers out of that range have to be allocated each t=
ime they're used. This is a potential source for more caching if you can ju=
stify more memory usage in the naive case. More advanced optimization techn=
iques here can encompass things like making numbers not real objects as far=
 as the interpreter is concerned internally (read: no allocations), but kee=
p actual type information separate, and allow us to just reference this one=
 allocated group of behaviours for all objects of that type. It may surpris=
e people to know that Io's type system is bonkers (almost doesn't exist unl=
ess you interface with C bindings).
* Changing the slot table layout to be backed by a dense array, and not a h=
ash table. There's no reason that has to make the core vm more complex eith=
er, in this respect. Slow lookup could then be a simple array offset lookup=
 instead of calculating the hash of some message name and then accessing th=
e element at that index.
* JIT techniques can be applied to Io quite easily, but that's a big discus=
sion that has major implementation design consequences.
* Copy on write object creation would be a net win in terms of alleviating =
some of the load on the allocator and garbage collector when a user clones =
an object. Coupled with the type system suggestion above, this could yield =
the biggest performance increase Io has ever seen, or ever will w/o some ma=
jor overhaul work.
* Garbage collection =E2=80=94 For many cases, we can tell when we build up=
 parse trees at a method level, what items will escape based on known keywo=
rds (yes Io has no real keywords, but we have things that are effectively k=
eywords based on naming conventions, think "call" and friends inside method=
s). Making assumptions based on these naming guidelines (which would have t=
o be formalized) could allow us to do some quick compile time conservative =
escape analysis to avoid allocating objects on the garbage collected heap, =
but instead work on a one-bit reference count. If at the end of the method =
they're still referenced, copy them from a cheap allocation source to the g=
arbage collected heap, effectively creating a nursery for objects which sho=
uld be able to allocate objects given a few instructions (basically be a po=
inter bump). Most objects in Io are very short lived (almost all messages f=
or instance, which are at LEAST guaranteed to be half of the objects in any=
 live Io system over the course of its life, based how on the language work=
s now).

All these items take significant effort to implement, and would produce an =
entirely different implementation (and change language semantics in some ar=
eas) when done. I hope that at some point, this will become a priority, and=
 that those of us who know about these things, will sink some effort into p=
utting these items into practice; either in the mainline with Steve's bless=
ing, or otherwise. There's no reason Io has to be this slow. You'll note, t=
hat I didn't even talk about taking advantage of multiple cores, which is a=
nother area that Io could greatly benefit from for many classes of work, no=
t the least of which being how the interpreter works.


Regards,

Jeremy Tregunna


On Friday, 21 December, 2012 at 11:26 AM, [email protected] wrote:

>=20=20=20=20
> I'm playing around with Io and liking it. I'd like to understand its=20=20
> performance characteristics. With a dumb integer arithmetic benchmark I =
=20
> get Io as ten times slower than perl. Is this a more generally accurate =
=20
> picture or not?
>=20=20
> =3D=3D=3D fib.io (http://fib.io) =3D=3D=3D
> fib :=3D method(n,
> if (n =3D=3D 0, return 1)
> if (n < 0, return -1)
> a :=3D 1; b :=3D 1
> n repeat (
> c :=3D a + b
> a :=3D b; b :=3D c
> )
> return a
> )
>=20=20
> 100000 repeat( fib(40) )
> =3D=3D=3D=3D=3D=3D=3D=3D
>=20=20
> =3D=3D=3D fib.pl (http://fib.pl) =3D=3D=3D
> sub fib {
> my $n =3D shift;
> return 1 if $n =3D=3D 0;
> return -1 if $n < 0;
> my ($a, $b) =3D (1, 1);
> for (1..$n) {
> my $c =3D $a + $b;
> $a =3D $b; $b =3D $c;
> }
> return $a;
> }
>=20=20
> fib(40) for 1..100000;
> =3D=3D=3D=3D=3D=3D=3D=3D
>=20=20
> $ time perl fib.pl (http://fib.pl) && time io fib.io (http://fib.io)
>=20=20
> real 0m0.904s
> user 0m0.890s
> sys 0m0.007s
>=20=20
> real 0m9.454s
> user 0m9.403s
> sys 0m0.003s
>=20=20=20


--Boundary_(ID_/+qQlVWiepkTsmtXvZtMRg)
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable




<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/htm=
l4/strict.dtd">
<html>
<head>
</head>







<body style=3D"background-color: #fff;">
<span style=3D"display:none">&nbsp;</span>

<!--~-|**|PrettyHtmlStartT|**|-~-->
<div id=3D"ygrp-mlmsg" style=3D"position:relative;">
  <div id=3D"ygrp-msg" style=3D"z-index: 1;">
<!--~-|**|PrettyHtmlEndT|**|-~-->

    <div id=3D"ygrp-text" >
=20=20=20=20=20=20
=20=20=20=20=20=20
      <p>
                <div>
                    <div>Io is a slow language. The implementation isn't tu=
ned for performance, it's tuned for flexibility. There are a few facts whic=
h make Io slow:</div><div><br></div><div>* Almost no optimizations above a =
micro level</div><div>* Primitive numbers in the range of -10 ~ &#43;250 (i=
irc) are cached at interpreter startup, but any numbers out of that range h=
ave to be allocated each time they're used. This is a potential source for =
more caching if you can justify more memory usage in the naive case. More a=
dvanced optimization techniques here can encompass things like making numbe=
rs not real objects as far as the interpreter is concerned internally (read=
: no allocations), but keep actual type information separate, and allow us =
to just reference this one allocated group of behaviours for all objects of=
 that type. It may surprise people to know that Io's type system is bonkers=
 (almost doesn't exist unless you interface with C bindings).</div><div>* C=
hanging the slot table layout to be backed by a dense array, and not a hash=
 table. There's no reason that has to make the core vm more complex either,=
 in this respect. Slow lookup could then be a simple array offset lookup in=
stead of calculating the hash of some message name and then accessing the e=
lement at that index.</div><div>* JIT techniques can be applied to Io quite=
 easily, but that's a big discussion that has major implementation design c=
onsequences.</div><div>* Copy on write object creation would be a net win i=
n terms of alleviating some of the load on the allocator and garbage collec=
tor when a user clones an object. Coupled with the type system suggestion a=
bove, this could yield the biggest performance increase Io has ever seen, o=
r ever will w/o some major overhaul work.</div><div>* Garbage collection =
=E2=80=94 For many cases, we can tell when we build up parse trees at a met=
hod level, what items will escape based on known keywords (yes Io has no re=
al keywords, but we have things that are effectively keywords based on nami=
ng conventions, think "call" and friends inside methods). Making assumption=
s based on these naming guidelines (which would have to be formalized) coul=
d allow us to do some quick compile time conservative escape analysis to av=
oid allocating objects on the garbage collected heap, but instead work on a=
 one-bit reference count. If at the end of the method they're still referen=
ced, copy them from a cheap allocation source to the garbage collected heap=
, effectively creating a nursery for objects which should be able to alloca=
te objects given a few instructions (basically be a pointer bump). Most obj=
ects in Io are very short lived (almost all messages for instance, which ar=
e at LEAST guaranteed to be half of the objects in any live Io system over =
the course of its life, based how on the language works now).</div><div><br=
></div><div>All these items take significant effort to implement, and would=
 produce an entirely different implementation (and change language semantic=
s in some areas) when done. I hope that at some point, this will become a p=
riority, and that those of us who know about these things, will sink some e=
ffort into putting these items into practice; either in the mainline with S=
teve's blessing, or otherwise. There's no reason Io has to be this slow. Yo=
u'll note, that I didn't even talk about taking advantage of multiple cores=
, which is another area that Io could greatly benefit from for many classes=
 of work, not the least of which being how the interpreter works.</div><div=
><br></div>
                </div>
                <div><div><br></div><div>Regards,</div><div><br></div><div>=
Jeremy Tregunna</div><div><br></div></div>
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20
                <p style=3D"color: #A0A0A8;">On Friday, 21 December, 2012 a=
t 11:26 AM, [email protected] wrote:</p>
                <blockquote type=3D"cite" style=3D"border-left-style:solid;=
border-width:1px;">
                    <span><div><div>














<span>&nbsp;</span>



    <div>
=20=20=20=20=20=20=20
=20=20=20=20=20=20=20
      <p>I'm playing around with Io and liking it. I'd like to understand i=
ts <br>
performance characteristics. With a dumb integer arithmetic benchmark I <br=
>
get Io as ten times slower than perl. Is this a more generally accurate <br=
>
picture or not?<br>
<br>
=3D=3D=3D <a href=3D"http://fib.io">fib.io</a> =3D=3D=3D<br>
fib :=3D method(n,<br>
     if (n =3D=3D 0, return 1)<br>
     if (n &lt; 0, return -1)<br>
     a :=3D 1; b :=3D 1<br>
     n repeat (<br>
         c :=3D a + b<br>
         a :=3D b; b :=3D c<br>
     )<br>
     return a<br>
)<br>
<br>
100000 repeat( fib(40) )<br>
=3D=3D=3D=3D=3D=3D=3D=3D<br>
<br>
=3D=3D=3D <a href=3D"http://fib.pl">fib.pl</a> =3D=3D=3D<br>
sub fib {<br>
     my $n =3D shift;<br>
     return 1 if $n =3D=3D 0;<br>
     return -1 if $n &lt; 0;<br>
     my ($a, $b) =3D (1, 1);<br>
     for (1..$n) {<br>
         my $c =3D $a + $b;<br>
         $a =3D $b; $b =3D $c;<br>
     }<br>
     return $a;<br>
}<br>
<br>
fib(40) for 1..100000;<br>
=3D=3D=3D=3D=3D=3D=3D=3D<br>
<br>
$ time perl <a href=3D"http://fib.pl">fib.pl</a> &amp;&amp; time io <a href=
=3D"http://fib.io">fib.io</a><br>
<br>
real    0m0.904s<br>
user    0m0.890s<br>
sys     0m0.007s<br>
<br>
real    0m9.454s<br>
user    0m9.403s<br>
sys     0m0.003s<br>
</p>

    </div>
=20=20=20=20=20=20

=20=20=20=20





<!-- end group email -->

</div></div></span>
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20
                </blockquote>
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20
                <div>
                    <br>
                </div>
</p>

    </div>
=20=20=20=20=20

    <!--~-|**|PrettyHtmlStart|**|-~-->
    <div style=3D"color: #fff; height: 0;">__._,_.___</div>

=20=20=20=20=20=20=20=20
=20=20
=20=20
    <table cellspacing=3D4px style=3D"margin-top: 20px; margin-bottom: 10px=
;">
      <tbody>
        <tr>
          <td style=3D"font-size: 12px; font-family: arial; font-weight: bo=
ld; padding: 7px 5px 5px; color: #FFF; background-color: #F2F2F2; border: 1=
px solid #EAEAEA "  >
                          <a style=3D"text-decoration: none; color: #2D50FD=
" href=3D"http://groups.yahoo.com/group/iolanguage/post;_ylc=3DX3oDMTJxbHNn=
MWVvBF9TAzk3MzU5NzE0BGdycElkAzY2Nzc0NzYEZ3Jwc3BJZAMxNzA1MDA2NzY0BG1zZ0lkAzE=
zMzAwBHNlYwNmdHIEc2xrA3JwbHkEc3RpbWUDMTM1NjI3NTA4Nw--?act=3Dreply&messageNu=
m=3D13300">Reply via web post</a>
                      </td>
          <td style=3D"font-size: 12px; font-family: arial; padding: 7px 5p=
x 5px; color: #FFF; background-color: #F2F2F2; border: 1px solid #EAEAEA; "=
 >
            <a href=3D"mailto:[email protected]?subject=3DRe%3A%20%5BI=
o%5D%20newbie%20curious%20of%20general%20relative%20performance" style=3D"t=
ext-decoration: none; color: #2D50FD;">
              Reply to sender            </a>=20
          </td>
          <td style=3D"font-size: 12px; font-family: arial; padding: 7px 5p=
x 5px; color: #FFF; background-color: #F2F2F2; border: 1px solid #EAEAEA; "=
>
            <a href=3D"mailto:[email protected]?subject=3DRe%3A%20=
%5BIo%5D%20newbie%20curious%20of%20general%20relative%20performance" style=
=3D"text-decoration: none; color: #2D50FD">
              Reply to group            </a>=20
          </td>
          <td style=3D"font-size: 12px; font-family: arial; padding: 7px 5p=
x 5px; color: #FFF; background-color: #F2F2F2; border: 1px solid #EAEAEA; "=
 >
            <a href=3D"http://groups.yahoo.com/group/iolanguage/post;_ylc=
=3DX3oDMTJlbWI0MjBiBF9TAzk3MzU5NzE0BGdycElkAzY2Nzc0NzYEZ3Jwc3BJZAMxNzA1MDA2=
NzY0BHNlYwNmdHIEc2xrA250cGMEc3RpbWUDMTM1NjI3NTA4Nw--" style=3D"text-decorat=
ion: none; color: #2D50FD">Start a New Topic</a>
          </td>
          <td style=3D"font-size: 12px; font-family: arial; padding: 7px 5p=
x 5px; color: #2D50FD; background-color: #F2F2F2; border: 1px solid #EAEAEA=
; " >
                            <a href=3D"http://groups.yahoo.com/group/iolang=
uage/message/13299;_ylc=3DX3oDMTM2b2owZ2RpBF9TAzk3MzU5NzE0BGdycElkAzY2Nzc0N=
zYEZ3Jwc3BJZAMxNzA1MDA2NzY0BG1zZ0lkAzEzMzAwBHNlYwNmdHIEc2xrA3Z0cGMEc3RpbWUD=
MTM1NjI3NTA4NwR0cGNJZAMxMzI5OQ--" style=3D"text-decoration: none; color: #2=
D50FD;">Messages in this topic</a>
                (2)
                      </td>
        </tr>
      </tbody>
    </table>

=20=20=20=20=20=20=20=20
<!------- Start Nav Bar ------>
<!-- |**|begin egp html banner|**| -->
<!-- |**|end egp html banner|**| -->

<!-- |**|begin egp html banner|**| -->
<div id=3D"ygrp-vital" style=3D"background-color: #f2f2f2; font-family: Ver=
dana; font-size: 10px; margin-bottom: 10px; padding: 10px;">
      <span id=3D"vithd" style=3D"font-weight: bold; color: #333; text-tran=
sform: uppercase; ">Recent Activity:</span>

    <ul style=3D"list-style-type: none; margin: 0; padding: 0; display: inl=
ine;">
            <li style=3D"border-right: 1px solid #000; font-weight: 700; di=
splay: inline; padding: 0 5px; margin-left: 0;">
      <span class=3D"cat"><a href=3D"http://groups.yahoo.com/group/iolangua=
ge/members;_ylc=3DX3oDMTJmbjFtdjIwBF9TAzk3MzU5NzE0BGdycElkAzY2Nzc0NzYEZ3Jwc=
3BJZAMxNzA1MDA2NzY0BHNlYwN2dGwEc2xrA3ZtYnJzBHN0aW1lAzEzNTYyNzUwODY-?o=3D6" =
style=3D"text-decoration: none;">New Members</a></span>
      <span class=3D"ct" style=3D"color: #ff7900;">3</span>
    </li>
                                              </ul>
=20=20=20=20
  <div style=3D"clear: both; padding-top: 2px; color: #1e66ae;">
    <a href=3D"http://groups.yahoo.com/group/iolanguage;_ylc=3DX3oDMTJlOXNu=
dmZsBF9TAzk3MzU5NzE0BGdycElkAzY2Nzc0NzYEZ3Jwc3BJZAMxNzA1MDA2NzY0BHNlYwN2dGw=
Ec2xrA3ZnaHAEc3RpbWUDMTM1NjI3NTA4Ng--" style=3D"text-decoration: none;">Vis=
it Your Group</a>
  </div>
</div>


=20=20
<div id=3D"ft" style=3D"font-family: Arial; font-size: 11px; margin-top: 5p=
x; padding: 0 2px 0 0; clear: both;">
  <a href=3D"http://groups.yahoo.com/;_ylc=3DX3oDMTJkdGYxNGowBF9TAzk3NDc2NT=
kwBGdycElkAzY2Nzc0NzYEZ3Jwc3BJZAMxNzA1MDA2NzY0BHNlYwNmdHIEc2xrA2dmcARzdGltZ=
QMxMzU2Mjc1MDg3" style=3D"float: left;"><img src=3D"http://l.yimg.com/a/i/u=
s/yg/logo/us.gif" height=3D"15" width=3D"137" alt=3D"Yahoo! Groups" style=
=3D"border: 0;"/></a>
  <div style=3D"color: #747575; float: right;">Switch to: <a href=3D"mailto=
:[email protected]?subject=3DChange Delivery Format: T=
raditional" style=3D"text-decoration: none;">Text-Only</a>, <a href=3D"mail=
to:[email protected]?subject=3DEmail Delivery: Digest" clas=
s=3D"margin-rt" style=3D"text-decoration: none;">Daily Digest</a> &bull; <a=
 href=3D"mailto:[email protected]?subject=3DUnsubscrib=
e" style=3D"text-decoration: none;">Unsubscribe</a> &bull; <a href=3D"http:=
//docs.yahoo.com/info/terms/" style=3D"text-decoration: none;">Terms of Use=
</a> &bull; <a href=3D"mailto:[email protected]?subject=
=3DFeedback on the redesigned individual mail v1" style=3D"text-decoration:=
 none;">Send us Feedback </a></div>
</div>

<!-- |**|end egp html banner|**| -->

  </div> <!-- ygrp-msg -->

  <!-- Sponsor -->
  <!-- |**|begin egp html banner|**| -->
  <div id=3D"ygrp-sponsor" style=3D"width:160px; float:right; clear:none; m=
argin:0 0 25px 0; background: #fff;">

<!-- Start Recommendations -->
<div id=3D"ygrp-reco">
     </div>
<!-- End Recommendations -->



  </div>   <!-- |**|end egp html banner|**| -->

  <div style=3D"clear:both; color: #FFF; font-size:1px;">.</div>
</div>

  <img src=3D"http://geo.yahoo.com/serv?s=3D97359714/grpId=3D6677476/grpspI=
d=3D1705006764/msgId=3D13300/stime=3D1356275087/nc1=3D3848642/nc2=3D5758221=
/nc3=3D4507179" width=3D"1" height=3D"1"> <br>

<div style=3D"color: #fff; height: 0;">__,_._,___</div>
<!--~-|**|PrettyHtmlEnd|**|-~-->

</body>

<!--~-|**|PrettyHtmlStart|**|-~-->
<head>
  <style type=3D"text/css">
  <!--
  #ygrp-mkp {
  border: 1px solid #d8d8d8;
  font-family: Arial;
  margin: 10px 0;
  padding: 0 10px;
}

#ygrp-mkp hr {
  border: 1px solid #d8d8d8;
}

#ygrp-mkp #hd {
  color: #628c2a;
  font-size: 85%;
  font-weight: 700;
  line-height: 122%;
  margin: 10px 0;
}

#ygrp-mkp #ads {
  margin-bottom: 10px;
}

#ygrp-mkp .ad {
  padding: 0 0;
}

#ygrp-mkp .ad p {
  margin: 0;
}

#ygrp-mkp .ad a {
  color: #0000ff;
  text-decoration: none;
}
  #ygrp-sponsor #ygrp-lc {
  font-family: Arial;
}

#ygrp-sponsor #ygrp-lc #hd {
  margin: 10px 0px;
  font-weight: 700;
  font-size: 78%;
  line-height: 122%;
}

#ygrp-sponsor #ygrp-lc .ad {
  margin-bottom: 10px;
  padding: 0 0;
}

  #actions {
    font-family: Verdana;
    font-size: 11px;
    padding: 10px 0;
  }

  #activity {
    background-color: #e0ecee;
    float: left;
    font-family: Verdana;
    font-size: 10px;
    padding: 10px;
  }

  #activity span {
    font-weight: 700;
  }

  #activity span:first-child {
    text-transform: uppercase;
  }

  #activity span a {
    color: #5085b6;
    text-decoration: none;
  }

  #activity span span {
    color: #ff7900;
  }

  #activity span .underline {
    text-decoration: underline;
  }

  .attach {
    clear: both;
    display: table;
    font-family: Arial;
    font-size: 12px;
    padding: 10px 0;
    width: 400px;
  }

  .attach div a {
    text-decoration: none;
  }

  .attach img {
    border: none;
    padding-right: 5px;
  }

  .attach label {
    display: block;
    margin-bottom: 5px;
  }

  .attach label a {
    text-decoration: none;
  }
=20=20
  blockquote {
    margin: 0 0 0 4px;
  }

  .bold {
    font-family: Arial;
    font-size: 13px;
    font-weight: 700;
  }

  .bold a {
    text-decoration: none;
  }

  dd.last p a {
    font-family: Verdana;
    font-weight: 700;
  }

  dd.last p span {
    margin-right: 10px;
    font-family: Verdana;
    font-weight: 700;
  }

  dd.last p span.yshortcuts {
    margin-right: 0;
  }

  div.attach-table div div a {
    text-decoration: none;
  }

  div.attach-table {
    width: 400px;
  }

  div.file-title a, div.file-title a:active, div.file-title a:hover, div.fi=
le-title a:visited {
    text-decoration: none;
  }

  div.photo-title a, div.photo-title a:active, div.photo-title a:hover, div=
.photo-title a:visited {
    text-decoration: none;
  }

  div#ygrp-mlmsg #ygrp-msg p a span.yshortcuts {
    font-family: Verdana;
    font-size: 10px;
    font-weight: normal;
  }

  .green {
    color: #628c2a;
  }

  .MsoNormal {
    margin: 0 0 0 0;
  }

  o {
    font-size: 0;
  }

  #photos div {
    float: left;
    width: 72px;
  }

  #photos div div {
    border: 1px solid #666666;
    height: 62px;
    overflow: hidden;
    width: 62px;
  }

  #photos div label {
    color: #666666;
    font-size: 10px;
    overflow: hidden;
    text-align: center;
    white-space: nowrap;
    width: 64px;
  }

  #reco-category {
    font-size: 77%;
  }

  #reco-desc {
    font-size: 77%;
  }

  .replbq {
    margin: 4px;
  }

  #ygrp-actbar div a:first-child {
   /* border-right: 0px solid #000;*/
    margin-right: 2px;
    padding-right: 5px;
  }

  #ygrp-mlmsg {
    font-size: 13px;
    font-family: Arial, helvetica,clean, sans-serif;
    *font-size: small;
    *font: x-small;
  }

  #ygrp-mlmsg table {
    font-size: inherit;
    font: 100%;
  }

  #ygrp-mlmsg select, input, textarea {
    font: 99% Arial, Helvetica, clean, sans-serif;
  }

  #ygrp-mlmsg pre, code {
    font:115% monospace;
    *font-size:100%;
  }

  #ygrp-mlmsg * {
    line-height: 1.22em;
  }

  #ygrp-mlmsg #logo {
    padding-bottom: 10px;
  }


  #ygrp-msg p a {
    font-family: Verdana;
  }

  #ygrp-msg p#attach-count span {
    color: #1E66AE;
    font-weight: 700;
  }

  #ygrp-reco #reco-head {
    color: #ff7900;
    font-weight: 700;
  }

  #ygrp-reco {
    margin-bottom: 20px;
    padding: 0px;
  }

  #ygrp-sponsor #ov li a {
    font-size: 130%;
    text-decoration: none;
  }

  #ygrp-sponsor #ov li {
    font-size: 77%;
    list-style-type: square;
    padding: 6px 0;
  }=20

  #ygrp-sponsor #ov ul {
    margin: 0;
    padding: 0 0 0 8px;
  }

  #ygrp-text {
    font-family: Georgia;
  }

  #ygrp-text p {
    margin: 0 0 1em 0;
  }

  #ygrp-text tt {
    font-size: 120%;
  }

  #ygrp-vital ul li:last-child {
    border-right: none !important;=20
  }=20
  -->
  </style>
</head>

<!--~-|**|PrettyHtmlEnd|**|-~-->
</html>
<!-- end group email -->


--Boundary_(ID_/+qQlVWiepkTsmtXvZtMRg)--