<div dir="ltr"><div class="gmail_default" style="font-family:times new roman,serif;font-size:large">7.10.1.20150612 <br><br>It's very sensitive, I see today. Sometimes it fails with <br><br><div style="margin-left:40px">Simplifier ticks exhausted<br><br></div>other times it works. Bizarre<br><br>I think the latter is the more common case and I think the former case has something to do with existing .o file or order of args and -fforce-recomp may mean it always works but that is very speculative.<br><br></div><div class="gmail_default" style="font-family:times new roman,serif;font-size:large">The two casesor <br></div><div class="gmail_default" style="font-family:times new roman,serif;font-size:large"><br><div style="margin-left:40px">$ time ghc -fno-specialise -O2 Slice.hs<br>[1 of 1] Compiling Data.Array.Accelerate.CUDA.Array.Slice ( Slice.hs, Slice.o )<br>ghc: panic! (the 'impossible' happened)<br> (GHC version 7.10.1.20150612 for x86_64-apple-darwin):<br> Simplifier ticks exhausted<br> When trying UnfoldingDone $j_s1FS5<br> To increase the limit, use -fsimpl-tick-factor=N (default 100)<br> If you need to do this, let GHC HQ know, and what factor you needed<br> To see detailed counts use -ddump-simpl-stats<br> Total ticks: 1668604<br><br>Please report this as a GHC bug: <a href="http://www.haskell.org/ghc/reportabug">http://www.haskell.org/ghc/reportabug</a><br><br><br>real 0m19.714s<br>user 0m18.264s<br>sys 0m0.875s<br><br>$ time ghc -O2 -fno-specialise Slice.hs <br>[1 of 1] Compiling Data.Array.Accelerate.CUDA.Array.Slice ( Slice.hs, Slice.o )<br><br>real 0m3.228s<br>user 0m3.040s<br>sys 0m0.155s<br></div></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Sat, Jun 20, 2015 at 2:03 PM, GHC <span dir="ltr"><<a href="mailto:ghc-devs@haskell.org" target="_blank">ghc-devs@haskell.org</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">#10491: Regression, simplifier explosion with Accelerate, cannot compile,<br>
increasing tick factor is not a workaround<br>
-------------------------------------+-------------------------------------<br>
Reporter: robertce | Owner:<br>
Type: bug | Status: new<br>
Priority: highest | Milestone: 7.10.2<br>
Component: Compiler | Version: 7.10.1<br>
Resolution: | Keywords:<br>
Operating System: Unknown/Multiple | Architecture:<br>
Type of failure: Compile-time | Unknown/Multiple<br>
performance bug | Test Case:<br>
Blocked By: | Blocking:<br>
Related Tickets: | Differential Revisions:<br>
-------------------------------------+-------------------------------------<br>
<br>
</span>Comment (by bgamari):<br>
<br>
George, what GHC version did you test `-fno-specialise` on? While<br>
yesterday I was able to confirm that `-fno-specialise` seemed to make no<br>
difference on a test machine running what should have been 7.10.1 I am now<br>
having trouble replicating this on my laptop. Unfortunately I no longer<br>
have access to the test environment on which I tested this yesterday but<br>
clearly something was inconsistent.<br>
<br>
I am now seeing on multiple machines that `-fno-specialise` indeed allows<br>
things to compile,<br>
{{{<br>
$ ghc -V<br>
The Glorious Glasgow Haskell Compilation System, version 7.10.1<br>
$$ time ghc Slice.hs -fforce-recomp -O2 -fno-specialise<br>
[1 of 1] Compiling Slice ( Slice.hs, Slice.o )<br>
<br>
real 0m3.759s<br>
user 0m1.688s<br>
sys 0m0.044s<br>
$ time ghc Slice.hs -fforce-recomp -O2<br>
[1 of 1] Compiling Slice ( Slice.hs, Slice.o )<br>
^C<br>
<br>
real 0m51.103s<br>
user 0m44.336s<br>
sys 0m0.948s<br>
}}}<br>
<br>
I am now looking at whether disabling only cross-module specialisation is<br>
enough to eliminate the blow-up.<br>
<br>
--<br>
Ticket URL: <<a href="http://ghc.haskell.org/trac/ghc/ticket/10491#comment:27" rel="noreferrer" target="_blank">http://ghc.haskell.org/trac/ghc/ticket/10491#comment:27</a>><br>
<div class="HOEnZb"><div class="h5">GHC <<a href="http://www.haskell.org/ghc/" rel="noreferrer" target="_blank">http://www.haskell.org/ghc/</a>><br>
The Glasgow Haskell Compiler<br>
</div></div></blockquote></div><br></div>