<html><body><div style="color:#000; background-color:#fff; font-family:arial, helvetica, sans-serif;font-size:12pt">Would you like a paste from jitviewer or the source code to run and examine with jitviewer?<br><br>-Roger<br><div><span><br></span></div><div><br></div> <div style="font-family: arial, helvetica, sans-serif; font-size: 12pt;"> <div style="font-family: times new roman, new york, times, serif; font-size: 12pt;"> <div dir="ltr"> <font face="Arial" size="2"> <hr size="1"> <b><span style="font-weight:bold;">From:</span></b> Alex Gaynor <alex.gaynor@gmail.com><br> <b><span style="font-weight: bold;">To:</span></b> Roger Flores <aidembb@yahoo.com> <br><b><span style="font-weight: bold;">Cc:</span></b> "pypy-dev@python.org" <pypy-dev@python.org> <br> <b><span style="font-weight: bold;">Sent:</span></b> Wednesday, February 27, 2013 12:23 PM<br> <b><span style="font-weight: bold;">Subject:</span></b> Re: [pypy-dev] Slow int
code<br> </font> </div> <br>
<div id="yiv1515662262"><div dir="ltr">In that context large longs means HUNDREDS or THOUSANDS of bits, not 64 :) Can you show us a full runnable example that illustrates this?<div><br></div><div>Alex</div></div><div class="yiv1515662262gmail_extra"><br><br><div class="yiv1515662262gmail_quote">
On Wed, Feb 27, 2013 at 12:10 PM, Roger Flores <span dir="ltr"><<a rel="nofollow" ymailto="mailto:aidembb@yahoo.com" target="_blank" href="mailto:aidembb@yahoo.com">aidembb@yahoo.com</a>></span> wrote:<br><blockquote class="yiv1515662262gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
Hi guys. I've been looking at two simple routines using jitviewer to figure out why they're so much slower than expected.<br>
<br>
<br>
I've also noticed that http://pypy.org/performance.html has the line "Bad examples include doing computations with<br>
large longs – which is performed by unoptimizable support code.". I'm worried that my 32 bit int code is falling into this, and I'm wondering what I can do to avoid it?<br>
<br>
Trivial code like<br>
<br>
if (self.low ^ self.high) & 0x80000000 == 0:<br>
<br>
<br>
is expanding into several dozen asm instructions. I'm suspecting that lines like <br>
<br>
self.low = (self.low << 1) & 0xffffffff<br>
<br>
<br>
with it's shift left are convincing the jit to consider the int to need 64 bits (large long?) instead of 32.<br>
<br>
<br>
Ideas? The asm is clearly operating on QWORDs and calling routines to do the bit arithmetic instead of single instructions. Is this what that line in performance.html is warning about?<br>
<br>
<br>
<br>
-Roger<br>
<br>
BTW Fijal's jitviewer is a *must see* for anyone interested in how pypy makes their code fast!<br>
<br>
_______________________________________________<br>
pypy-dev mailing list<br>
<a rel="nofollow" ymailto="mailto:pypy-dev@python.org" target="_blank" href="mailto:pypy-dev@python.org">pypy-dev@python.org</a><br>
http://mail.python.org/mailman/listinfo/pypy-dev<br>
</blockquote></div><br><br clear="all"><div><br></div>-- <br>"I disapprove of what you say, but I will defend to the death your right to say it." -- Evelyn Beatrice Hall (summarizing Voltaire)<br>"The people's good is the highest law." -- Cicero<br>
</div>
</div><br><br> </div> </div> </div></body></html>