<div dir="ltr">I've since run NoFib. You can find the results here: <a href="https://phabricator.haskell.org/D4244#119697">https://phabricator.haskell.org/D4244#119697</a><div><br></div><div>I wonder if you feel that any more notes are needed? The general idea of CPR remained the same, it's just the extension of the DmdResult lattice that needs some rationale as to why and when these new values are needed.</div><div><br></div><div>I also wonder what the impact of "Slightly strengthen the strictness analysis" (<a href="https://ghc.haskell.org/trac/ghc/wiki/NestedCPR/Akio2017#Changestothedemandanalyzer">https://ghc.haskell.org/trac/ghc/wiki/NestedCPR/Akio2017#Changestothedemandanalyzer</a>) would be if regarded in isolation.</div></div><div class="gmail_extra"><br><div class="gmail_quote">On Tue, Jan 2, 2018 at 11:52 AM, Matthew Pickering <span dir="ltr"><<a href="mailto:matthewtpickering@gmail.com" target="_blank">matthewtpickering@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">I don't think anyone has run nofib on the rebased branch yet.<br>
<br>
The Akio2017 subpage is a more accurate summary. Sebastian has also<br>
been adding notes to explain the more intricate parts.<br>
<br>
Matt<br>
<div class="HOEnZb"><div class="h5"><br>
On Fri, Dec 22, 2017 at 5:27 PM, Simon Peyton Jones<br>
<<a href="mailto:simonpj@microsoft.com">simonpj@microsoft.com</a>> wrote:<br>
> Terrific!<br>
><br>
> What are the nofib results?<br>
><br>
> Can we have a couple of artificial benchmarks in cpranal/should_run that show substantial perf improvements because the nested CPR wins in some inner loop?<br>
><br>
> Is <a href="https://ghc.haskell.org/trac/ghc/wiki/NestedCPR" rel="noreferrer" target="_blank">https://ghc.haskell.org/trac/<wbr>ghc/wiki/NestedCPR</a> still an accurate summary of the idea? And the Akio2017 sub-page? It would be easier to review the code if the design documentation accurately described it.<br>
><br>
> I'll look in the new year. Thanks!<br>
><br>
> Simon<br>
><br>
> | -----Original Message-----<br>
> | From: Matthew Pickering [mailto:<a href="mailto:matthewtpickering@gmail.com">matthewtpickering@<wbr>gmail.com</a>]<br>
> | Sent: 22 December 2017 17:09<br>
> | To: GHC developers <<a href="mailto:ghc-devs@haskell.org">ghc-devs@haskell.org</a>>; Simon Peyton Jones<br>
> | <<a href="mailto:simonpj@microsoft.com">simonpj@microsoft.com</a>>; Joachim Breitner <<a href="mailto:mail@joachim-breitner.de">mail@joachim-breitner.de</a>>;<br>
> | <a href="mailto:tkn.akio@gmail.com">tkn.akio@gmail.com</a>; Sebastian Graf <<a href="mailto:sgraf1337@gmail.com">sgraf1337@gmail.com</a>><br>
> | Subject: Nested CPR patch review<br>
> |<br>
> | Hi all,<br>
> |<br>
> | I recently resurrected akio's nested cpr branch and put it on phabricator<br>
> | for review.<br>
> |<br>
> | <a href="https://phabricator.haskell.org/D4244" rel="noreferrer" target="_blank">https://phabricator.haskell.<wbr>org/D4244</a><br>
> |<br>
> | Sebastian has kindly been going over it and ironed out a few kinks in the<br>
> | last few days. He says now that he believes the patch is correct.<br>
> |<br>
> | Is there anything else which needs to be done before merging this patch?<br>
> |<br>
> | Simon, would you perhaps be able to give the patch a look over?<br>
> |<br>
> | Cheers,<br>
> |<br>
> | Matt<br>
</div></div></blockquote></div><br></div>