Re: Unaligning ghc

2011-08-18 Thread Alexander Kjeldaas
On 18 August 2011 23:38, Simon Marlow wrote: > On 18/08/11 11:47, Johan Tibell wrote: > >> On Thu, Aug 18, 2011 at 12:43 PM, Alexander Kjeldaas >> wrote: >> >>> Unaligned word-sized loads work fine on x86, and this would be x86-64 >>> only, >>> or even Nehalem (and later) only. Or, from a co

pgj-freebsd-amd64-stable (amd64 FreeBSD STABLE), build 333, Success

2011-08-18 Thread Builder
pgj-freebsd-amd64-stable (amd64 FreeBSD STABLE), build 333 Build succeeded Details: http://darcs.haskell.org/ghcBuilder/builders/pgj-freebsd-amd64-stable/333.html git clone| Success create mk/build.mk | Success get subrepos | Success repo versions| Success setting v

pgj-freebsd-i386-stable (x86 FreeBSD STABLE), build 316, Success

2011-08-18 Thread Builder
pgj-freebsd-i386-stable (x86 FreeBSD STABLE), build 316 Build succeeded Details: http://darcs.haskell.org/ghcBuilder/builders/pgj-freebsd-i386-stable/316.html git clone| Success create mk/build.mk | Success get subrepos | Success repo versions| Success setting versi

pgj2 (amd64 FreeBSD HEAD), build 451, Success

2011-08-18 Thread Builder
pgj2 (amd64 FreeBSD HEAD), build 451 Build succeeded Details: http://darcs.haskell.org/ghcBuilder/builders/pgj2/451.html git clone| Success create mk/build.mk | Success get subrepos | Success repo versions| Success setting version date | Success booting

[nightly] 18-Aug-2011 build of STABLE on i386-unknown-linux (cam-02-unx)

2011-08-18 Thread GHC Build Reports
Build description = STABLE on i386-unknown-linux (cam-02-unx) Build location= /playpen/simonmar/nightly/STABLE Build config file = /home/simonmar/nightly/site/msrc/conf-STABLE-cam-02-unx Nightly build started on cam-02-unx at Thu Aug 18 18:10:01 BST 2011. checking out new source tree

[commit: testsuite] master: .gitignore (2afdb7c)

2011-08-18 Thread Manuel Chakravarty
Repository : ssh://darcs.haskell.org//srv/darcs/testsuite On branch : master http://hackage.haskell.org/trac/ghc/changeset/2afdb7c5c3413d8d01230551cc7a4ed9c15893f3 >--- commit 2afdb7c5c3413d8d01230551cc7a4ed9c15893f3 Author: Manuel M

Re: validates breaks on x86_64 (OS X)

2011-08-18 Thread Manuel M T Chakravarty
Johan Tibell: > Just tried on my OS X/x86_64 and it builds fine for me. Make sure you have > > https://github.com/ghc/packages-ghc-prim/commit/cefc19afafe5107ff98d5205c204b190da1d497b I double checked and it works now on 64bit. I may not have properly cleaned the tree. (I think it is still bro

pgj (x86 FreeBSD HEAD), build 453, Failure

2011-08-18 Thread Builder
pgj (x86 FreeBSD HEAD), build 453 Build failed Details: http://darcs.haskell.org/ghcBuilder/builders/pgj/453.html git clone| Success create mk/build.mk | Success get subrepos | Success repo versions| Success setting version date | Success booting | Succe

[nightly] 18-Aug-2011 build of HEAD on x86_64-unknown-linux (cam-04-unx)

2011-08-18 Thread GHC Build Reports
Build description = HEAD on x86_64-unknown-linux (cam-04-unx) Build location= /64playpen/simonmar/nightly/HEAD-cam-04-unx Build config file = /home/simonmar/nightly/site/msrc/conf-HEAD-cam-04-unx Nightly build started on cam-04-unx at Thu Aug 18 18:00:01 BST 2011. checking out new source

[nightly] 18-Aug-2011 build of STABLE on x86_64-unknown-linux (cam-04-unx)

2011-08-18 Thread GHC Build Reports
Build description = STABLE on x86_64-unknown-linux (cam-04-unx) Build location= /64playpen/simonmar/nightly/STABLE-cam-04-unx Build config file = /home/simonmar/nightly/site/msrc/conf-STABLE-cam-04-unx Nightly build started on cam-04-unx at Thu Aug 18 18:10:01 BST 2011. checking out new s

Re: Last two vectoriser patches

2011-08-18 Thread Manuel M T Chakravarty
It was a bug in the test that only showed with some recent changes to the vectoriser. I just pushed a patch. Manuel Daniel Fischer: > On Thursday 18 August 2011, 17:19:57, Manuel M T Chakravarty wrote: >> I did validate them, but with Johan's x86_64 breaking patch removed >> (which my patches s

[commit: testsuite] master: dph/quickhull: Import vectorised Double (be8b8c5)

2011-08-18 Thread Manuel Chakravarty
Repository : ssh://darcs.haskell.org//srv/darcs/testsuite On branch : master http://hackage.haskell.org/trac/ghc/changeset/be8b8c546f510f2ca1b7411b27aa5a71622fbdd5 >--- commit be8b8c546f510f2ca1b7411b27aa5a71622fbdd5 Author: Manuel M

Re: validates breaks on x86_64 (OS X)

2011-08-18 Thread Johan Tibell
On Thu, Aug 18, 2011 at 11:45 PM, Simon Marlow wrote: > It's broken on 32-bit platforms (like Johan, I validated on 64-bit and > didn't notice the 32-bit breakage).  I have a fix on the way... I got alerted to the problem this morning (which is 32-bit only) and had a fix that Simon was validating

Re: validates breaks on x86_64 (OS X)

2011-08-18 Thread Simon Marlow
On 18/08/11 19:46, Daniel Fischer wrote: On Thursday 18 August 2011, 20:14:16, David Terei wrote: So it fails for me on OSX as well. "inplace/bin/genprimopcode" --data-decl< compiler/prelude/primops.txt> compiler/primop-data-decl.hs-incl genprimopcode: type of primop PopCnt64Op doesn't make se

Re: Unaligning ghc

2011-08-18 Thread Simon Marlow
On 18/08/11 11:47, Johan Tibell wrote: On Thu, Aug 18, 2011 at 12:43 PM, Alexander Kjeldaas wrote: Unaligned word-sized loads work fine on x86, and this would be x86-64 only, or even Nehalem (and later) only. Or, from a cost perspective, it could be interesting for non-Nehalem as well, as R

Re: Last two vectoriser patches

2011-08-18 Thread Daniel Fischer
On Thursday 18 August 2011, 17:19:57, Manuel M T Chakravarty wrote: > I did validate them, but with Johan's x86_64 breaking patch removed > (which my patches shouldn't interact with). > > Manuel > I don't know if those patches broke it because I haven't run validate for a couple of days, but dp

Re: validates breaks on x86_64 (OS X)

2011-08-18 Thread Daniel Fischer
On Thursday 18 August 2011, 20:14:16, David Terei wrote: > So it fails for me on OSX as well. > > "inplace/bin/genprimopcode" --data-decl < > compiler/prelude/primops.txt > compiler/primop-data-decl.hs-incl > genprimopcode: type of primop PopCnt64Op doesn't make sense w.r.t > category Mon

Re: validates breaks on x86_64 (OS X)

2011-08-18 Thread David Terei
So it fails for me on OSX as well. "inplace/bin/genprimopcode" --data-decl < compiler/prelude/primops.txt > compiler/primop-data-decl.hs-incl genprimopcode: type of primop PopCnt64Op doesn't make sense w.r.t category Monadic I've got all the latest patches. Running OSX 10.7, Core i7, 32b

simonmar-win32-head (x86 Windows HEAD), build 373, Failure

2011-08-18 Thread Builder
simonmar-win32-head (x86 Windows HEAD), build 373 Build failed Details: http://darcs.haskell.org/ghcBuilder/builders/simonmar-win32-head/373.html git clone| Success create mk/build.mk | Success get subrepos | Success repo versions| Success setting version date | Suc

[nightly] 18-Aug-2011 build of HEAD on i386-unknown-linux (cam-02-unx)

2011-08-18 Thread GHC Build Reports
Build description = HEAD on i386-unknown-linux (cam-02-unx) Build location= /playpen/simonmar/nightly/HEAD Build config file = /home/simonmar/nightly/site/msrc/conf-HEAD-cam-02-unx Nightly build started on cam-02-unx at Thu Aug 18 18:00:01 BST 2011. checking out new source tree

Re: validates breaks on x86_64 (OS X)

2011-08-18 Thread Johan Tibell
Just tried on my OS X/x86_64 and it builds fine for me. Make sure you have https://github.com/ghc/packages-ghc-prim/commit/cefc19afafe5107ff98d5205c204b190da1d497b On Thu, Aug 18, 2011 at 4:36 PM, Johan Tibell wrote: > It works on my machine, could it be due to different OS X versions? > I'm sti

Last two vectoriser patches

2011-08-18 Thread Manuel M T Chakravarty
I did validate them, but with Johan's x86_64 breaking patch removed (which my patches shouldn't interact with). Manuel ___ Cvs-ghc mailing list Cvs-ghc@haskell.org http://www.haskell.org/mailman/listinfo/cvs-ghc

[commit: ghc] master: Add VECTORISE [SCALAR] type pragma (46fa261)

2011-08-18 Thread Manuel Chakravarty
Repository : ssh://darcs.haskell.org//srv/darcs/ghc On branch : master http://hackage.haskell.org/trac/ghc/changeset/46fa261eee74c1c1a1be52f9394ff131183024da >--- commit 46fa261eee74c1c1a1be52f9394ff131183024da Author: Manuel M T Chak

[commit: ghc] master: Improve import and export of vectorisation information (9282550)

2011-08-18 Thread Manuel Chakravarty
Repository : ssh://darcs.haskell.org//srv/darcs/ghc On branch : master http://hackage.haskell.org/trac/ghc/changeset/9282550cc1f1b7915642871cb6010ba45988683a >--- commit 9282550cc1f1b7915642871cb6010ba45988683a Author: Manuel M T Chak

Re: validates breaks on x86_64 (OS X)

2011-08-18 Thread Johan Tibell
It works on my machine, could it be due to different OS X versions? I'm still on Snow Leopard. The symbols are defined in ghc-prim, perhaps you need to pull the latest version? On Thu, Aug 18, 2011 at 4:29 PM, Manuel M T Chakravarty wrote: > This popCnt stuff doesn't seem to work on OS X/x86_64:

validates breaks on x86_64 (OS X)

2011-08-18 Thread Manuel M T Chakravarty
This popCnt stuff doesn't seem to work on OS X/x86_64: > < (6 samples), 119M in use, 0.00 INIT (0.00 elapsed), 0.83 MUT (0.93 elapsed), > 0.54 GC (0.56 elapsed) :ghc>> > "/usr/bin/ghc" -o ghc/stage1/build/tmp/ghc-stage1 -H64m -O -fasm -W > -fno-warn-unused-matches -fwarn-unused-imports -packa

Re: Unaligning ghc

2011-08-18 Thread Ben Lippmeier
On 18/08/2011, at 20:22 , Alexander Kjeldaas wrote: > The Nehalem micro-architecture has made unaligned loads very cheap, as long > as they do not cross a cache line boundary. > > I am thinking that this makes it possible for ghc to use 40-bit pointers, and > generally use "packed" structure l

Re: Unaligning ghc

2011-08-18 Thread Johan Tibell
On Thu, Aug 18, 2011 at 12:43 PM, Alexander Kjeldaas wrote: > Unaligned word-sized loads work fine on x86, and this would be x86-64 only, > or even Nehalem (and later) only.   Or,  from a cost perspective, it could > be interesting for non-Nehalem as well, as RAM is (usually) the most > expensive

Re: Unaligning ghc

2011-08-18 Thread Alexander Kjeldaas
On 18 August 2011 12:29, Johan Tibell wrote: > On Thu, Aug 18, 2011 at 12:22 PM, Alexander Kjeldaas > wrote: > > The Nehalem micro-architecture has made unaligned loads very cheap, as > long > > as they do not cross a cache line boundary. > > I am thinking that this makes it possible for ghc to

Re: Unaligning ghc

2011-08-18 Thread Johan Tibell
On Thu, Aug 18, 2011 at 12:22 PM, Alexander Kjeldaas wrote: > The Nehalem micro-architecture has made unaligned loads very cheap, as long > as they do not cross a cache line boundary. > I am thinking that this makes it possible for ghc to use 40-bit pointers, > and generally use "packed" structure

New patches in /srv/darcs/git-mirrors/random

2011-08-18 Thread Ian Lynagh
New patches in /srv/darcs/git-mirrors/random commit f9a348ff46e92d64b70f026109765bbfd63000fd Merge: eda121f... 3fb0e9e... Author: Ryan Newton Date: Mon Aug 8 20:51:50 2011 -0700 Merge pull request #1 from dterei/master Patch for random to use Safe Haskell commit 3fb0e9e42b54d7b0

New patches in /srv/darcs/git-mirrors/terminfo

2011-08-18 Thread Ian Lynagh
New patches in /srv/darcs/git-mirrors/terminfo commit baec6aff59d13ba294b370f9563e8068706392ce Author: Judah Jacobson Date: Sun Aug 14 00:38:45 2011 + Require ghc-6.10, and don't use the extensible-exceptions package. commit 1fdbf694a26192ca8c0c4ecf036729a2eca0da30 Author: Judah Jaco

New patches in /srv/darcs/git-mirrors/pretty

2011-08-18 Thread Ian Lynagh
New patches in /srv/darcs/git-mirrors/pretty commit 036fb8dfbb9d4a787fcd150c2756b4899be4e942 Author: David Terei Date: Tue Aug 9 10:42:17 2011 -0700 Fix conditional pragma to work with 6.12 ___ Cvs-ghc mailing list Cvs-ghc@haskell.org http://ww

New patches in /srv/darcs/git-mirrors/primitive

2011-08-18 Thread Ian Lynagh
New patches in /srv/darcs/git-mirrors/primitive commit 5088de4ffb651982ca09319d5048426b3e3151f9 Author: Roman Leshchinskiy Date: Wed Aug 17 22:37:10 2011 + Changelog commit 74479e07b92b8859eae473e5cc86b40decae1d6e Author: Roman Leshchinskiy Date: Wed Aug 17 22:19:33 2011 +

New patches in /srv/darcs/git-mirrors/containers

2011-08-18 Thread Ian Lynagh
New patches in /srv/darcs/git-mirrors/containers commit 2575ba6eec15ef99c046c7e09a8ab879f8b21376 Merge: 770fbc3... 667591b... Author: Milan Straka Date: Tue Aug 9 11:50:00 2011 -0700 Merge pull request #2 from dterei/master fix mistake in last patch commit 667591b168c804d3eeae50

New patches in /srv/darcs/git-mirrors/Cabal

2011-08-18 Thread Ian Lynagh
New patches in /srv/darcs/git-mirrors/Cabal commit a2ae5a8d5f15f346eee030ae9800c7a47c11b471 Author: Duncan Coutts Date: Fri Aug 12 11:08:46 2011 + Relax cabal-install's deps for ghc-7.2 commit a792a50a73a4b606e4886c51e977810b500f575f Author: Duncan Coutts Date: Fri Aug 12 11:08:20

New patches in /srv/darcs/git-mirrors/hoopl

2011-08-18 Thread Ian Lynagh
New patches in /srv/darcs/git-mirrors/hoopl commit 6ffca823d05780f8b1c0de598da814bf248d1c06 Author: David Terei Date: Tue Aug 16 15:27:04 2011 -0700 Fix safe haskell issue with ghc 7.2.1 We can't rely on the containers package using Safe Haskell yet, so must have XUtil be Tru

New patches in /srv/darcs/git-mirrors/vector

2011-08-18 Thread Ian Lynagh
New patches in /srv/darcs/git-mirrors/vector commit 0da81dc2ced5ae6c4e179917f9861a45ff02b7e0 Author: Roman Leshchinskiy Date: Wed Aug 17 22:47:52 2011 + Remove deprecated functions commit 0832727605a1758302f8f621aca40bc0e7cedfdb Author: Roman Leshchinskiy Date: Wed Aug 17 22:37:40

New patches in /srv/darcs/git-mirrors/haskeline

2011-08-18 Thread Ian Lynagh
New patches in /srv/darcs/git-mirrors/haskeline commit 8787a64417500efffc9c48032ee7c37315fb2547 Author: Judah Jacobson Date: Sun Aug 14 00:01:00 2011 + Fix bug on ghc-7.2.1 with tab-completion of Unicode filenames. This was due to the changed behavior of getDirectoryContents.

Unaligning ghc

2011-08-18 Thread Alexander Kjeldaas
The Nehalem micro-architecture has made unaligned loads very cheap, as long as they do not cross a cache line boundary. I am thinking that this makes it possible for ghc to use 40-bit pointers, and generally use "packed" structure layout. This again should improve performance by increasing the ef

simonmar-win32-stable (x86 Windows STABLE), build 305, Success

2011-08-18 Thread Builder
simonmar-win32-stable (x86 Windows STABLE), build 305 Build succeeded Details: http://darcs.haskell.org/ghcBuilder/builders/simonmar-win32-stable/305.html git clone| Success create mk/build.mk | Success get subrepos | Success repo versions| Success setting version d