Summarize #7602

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

Summarize #7602

Andrew Farmer
I know that #7602 has been mentioned as a final loose end in the RC. I was
wondering if someone in the know could summarize its status in regards to
how it effects GHC users.

Feel free to be blisteringly brief... ;-) I'm just looking for things like:
"this is a non-issue on modern OS X/Clang" or "it'll be fixed soon just be
patient" or "we've done our part, this is Apple's problem now".

I gave the trac thread a read-through, but there seems to be several
degrees of flexibility/workarounds, so I'm still not sure if this is
something I should worry about on my OS X machine.

Thanks for your cycles!
Andrew
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.haskell.org/pipermail/ghc-devs/attachments/20140204/ecd0f3df/attachment.html>

Reply | Threaded
Open this post in threaded view
|

Summarize #7602

Austin Seipp-5
TL;DR OS X suffers a performance hit when build with llvm-gcc/clang.
The binary builds are built this way. The fix I outlined in the ticket
(a fast `pthread_getspecific`/`pthread_setspecific`) could still work
to recover most of the performance, but Apple accidentally messed up
the Mavericks source code release, so I'm not sure if anything
regarding the pthread implementation has changed. If it hasn't, then I
think it should be reasonably easy to fix with some testing.

(The ticket is a bit of a long and dry read I'm afraid.)

On Tue, Feb 4, 2014 at 3:22 PM, Andrew Farmer <afarmer at ittc.ku.edu> wrote:

> I know that #7602 has been mentioned as a final loose end in the RC. I was
> wondering if someone in the know could summarize its status in regards to
> how it effects GHC users.
>
> Feel free to be blisteringly brief... ;-) I'm just looking for things like:
> "this is a non-issue on modern OS X/Clang" or "it'll be fixed soon just be
> patient" or "we've done our part, this is Apple's problem now".
>
> I gave the trac thread a read-through, but there seems to be several degrees
> of flexibility/workarounds, so I'm still not sure if this is something I
> should worry about on my OS X machine.
>
> Thanks for your cycles!
> Andrew
>
> _______________________________________________
> ghc-devs mailing list
> ghc-devs at haskell.org
> http://www.haskell.org/mailman/listinfo/ghc-devs
>



--
Regards,

Austin Seipp, Haskell Consultant
Well-Typed LLP, http://www.well-typed.com/

Reply | Threaded
Open this post in threaded view
|

Summarize #7602

Carter Schonwald
@Andrew, if you have GHC built with a real GCC, theres no perf issue.
(though if you built GHC with clang, thereis)

I've a bindist thats from the 7.8 branch from 5 days ago that I built wiht
GCC-4.8 here
https://github.com/cartazio/ghc/releases/tag/carters-build-7.8-preRC1
you may need to patch the settings file to point to your own GCC version

point being, no issue if you built with GCC. :)


On Tue, Feb 4, 2014 at 4:31 PM, Austin Seipp <austin at well-typed.com> wrote:

> TL;DR OS X suffers a performance hit when build with llvm-gcc/clang.
> The binary builds are built this way. The fix I outlined in the ticket
> (a fast `pthread_getspecific`/`pthread_setspecific`) could still work
> to recover most of the performance, but Apple accidentally messed up
> the Mavericks source code release, so I'm not sure if anything
> regarding the pthread implementation has changed. If it hasn't, then I
> think it should be reasonably easy to fix with some testing.
>
> (The ticket is a bit of a long and dry read I'm afraid.)
>
> On Tue, Feb 4, 2014 at 3:22 PM, Andrew Farmer <afarmer at ittc.ku.edu> wrote:
> > I know that #7602 has been mentioned as a final loose end in the RC. I
> was
> > wondering if someone in the know could summarize its status in regards to
> > how it effects GHC users.
> >
> > Feel free to be blisteringly brief... ;-) I'm just looking for things
> like:
> > "this is a non-issue on modern OS X/Clang" or "it'll be fixed soon just
> be
> > patient" or "we've done our part, this is Apple's problem now".
> >
> > I gave the trac thread a read-through, but there seems to be several
> degrees
> > of flexibility/workarounds, so I'm still not sure if this is something I
> > should worry about on my OS X machine.
> >
> > Thanks for your cycles!
> > Andrew
> >
> > _______________________________________________
> > ghc-devs mailing list
> > ghc-devs at haskell.org
> > http://www.haskell.org/mailman/listinfo/ghc-devs
> >
>
>
>
> --
> Regards,
>
> Austin Seipp, Haskell Consultant
> Well-Typed LLP, http://www.well-typed.com/
> _______________________________________________
> ghc-devs mailing list
> ghc-devs at haskell.org
> http://www.haskell.org/mailman/listinfo/ghc-devs
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.haskell.org/pipermail/ghc-devs/attachments/20140204/24b43bef/attachment.html>