nofib.git
14 years ago[project @ 2005-05-09 09:09:20 by simonmar] Initial_conversion_from_CVS_complete
simonmar [Mon, 9 May 2005 09:09:20 +0000 (09:09 +0000)] 
[project @ 2005-05-09 09:09:20 by simonmar]
remove slow way; it generates too much output

14 years ago[project @ 2005-05-04 09:58:50 by simonmar]
simonmar [Wed, 4 May 2005 10:00:37 +0000 (10:00 +0000)] 
[project @ 2005-05-04 09:58:50 by simonmar]
add x86_64 output

14 years ago[project @ 2005-05-04 09:44:09 by simonmar]
simonmar [Wed, 4 May 2005 09:44:09 +0000 (09:44 +0000)] 
[project @ 2005-05-04 09:44:09 by simonmar]
update output

14 years ago[project @ 2005-05-04 09:37:53 by simonmar]
simonmar [Wed, 4 May 2005 09:40:13 +0000 (09:40 +0000)] 
[project @ 2005-05-04 09:37:53 by simonmar]
slow output

14 years ago[project @ 2005-05-04 09:34:36 by simonmar]
simonmar [Wed, 4 May 2005 09:34:36 +0000 (09:34 +0000)] 
[project @ 2005-05-04 09:34:36 by simonmar]
add slow output

14 years ago[project @ 2005-04-28 13:34:25 by simonmar]
simonmar [Thu, 28 Apr 2005 13:34:27 +0000 (13:34 +0000)] 
[project @ 2005-04-28 13:34:25 by simonmar]
mode=slow

That completes the imaginary and spectral suites.  Most of the
programs in here will now run in 1-2s with mode=slow (on a 3GHz P4).
I had to modify the programs in a few places, so there will be some
slight discrepancies in the size of Main.o due to these changes.

14 years ago[project @ 2005-04-28 12:08:12 by simonmar]
simonmar [Thu, 28 Apr 2005 12:08:16 +0000 (12:08 +0000)] 
[project @ 2005-04-28 12:08:12 by simonmar]
mode=slow

14 years ago[project @ 2005-04-28 11:34:22 by simonmar]
simonmar [Thu, 28 Apr 2005 11:34:22 +0000 (11:34 +0000)] 
[project @ 2005-04-28 11:34:22 by simonmar]
back out: attempted repetition wasn't really working

14 years ago[project @ 2005-04-28 10:57:21 by simonmar]
simonmar [Thu, 28 Apr 2005 10:57:21 +0000 (10:57 +0000)] 
[project @ 2005-04-28 10:57:21 by simonmar]
mode=slow

14 years ago[project @ 2005-04-28 10:52:59 by simonmar]
simonmar [Thu, 28 Apr 2005 10:52:59 +0000 (10:52 +0000)] 
[project @ 2005-04-28 10:52:59 by simonmar]
mode=slow

14 years ago[project @ 2005-04-28 10:48:41 by simonmar]
simonmar [Thu, 28 Apr 2005 10:48:41 +0000 (10:48 +0000)] 
[project @ 2005-04-28 10:48:41 by simonmar]
fish doesn't take any args; this looks like a cut-n-pasto

14 years ago[project @ 2005-04-28 10:45:09 by simonmar]
simonmar [Thu, 28 Apr 2005 10:45:09 +0000 (10:45 +0000)] 
[project @ 2005-04-28 10:45:09 by simonmar]
mode=slow

14 years ago[project @ 2005-04-28 10:36:38 by simonmar]
simonmar [Thu, 28 Apr 2005 10:36:38 +0000 (10:36 +0000)] 
[project @ 2005-04-28 10:36:38 by simonmar]
mode=slow

14 years ago[project @ 2005-04-28 10:26:30 by simonmar]
simonmar [Thu, 28 Apr 2005 10:26:30 +0000 (10:26 +0000)] 
[project @ 2005-04-28 10:26:30 by simonmar]
mode=slow

14 years ago[project @ 2005-04-28 10:17:56 by simonmar]
simonmar [Thu, 28 Apr 2005 10:17:56 +0000 (10:17 +0000)] 
[project @ 2005-04-28 10:17:56 by simonmar]
mode=slow

14 years ago[project @ 2005-04-28 10:13:44 by simonmar]
simonmar [Thu, 28 Apr 2005 10:13:44 +0000 (10:13 +0000)] 
[project @ 2005-04-28 10:13:44 by simonmar]
mode=slow

14 years ago[project @ 2005-04-28 10:10:05 by simonmar]
simonmar [Thu, 28 Apr 2005 10:10:05 +0000 (10:10 +0000)] 
[project @ 2005-04-28 10:10:05 by simonmar]
mode=slow

14 years ago[project @ 2005-04-28 10:02:59 by simonmar]
simonmar [Thu, 28 Apr 2005 10:05:01 +0000 (10:05 +0000)] 
[project @ 2005-04-28 10:02:59 by simonmar]
mode=slow

14 years ago[project @ 2005-04-28 09:52:55 by simonmar]
simonmar [Thu, 28 Apr 2005 09:57:01 +0000 (09:57 +0000)] 
[project @ 2005-04-28 09:52:55 by simonmar]
speed up mode=slow

14 years ago[project @ 2005-04-28 09:46:25 by simonmar]
simonmar [Thu, 28 Apr 2005 09:46:25 +0000 (09:46 +0000)] 
[project @ 2005-04-28 09:46:25 by simonmar]
Run each nofib program multiple times, tunable by the $(NoFibRuns)
variable.  nofib-analyse already knows how to average the times from
multiple runs.

14 years ago[project @ 2005-04-27 17:03:17 by sof]
sof [Wed, 27 Apr 2005 17:03:18 +0000 (17:03 +0000)] 
[project @ 2005-04-27 17:03:17 by sof]
de^Monised

14 years ago[project @ 2005-03-23 09:18:20 by simonmar]
simonmar [Wed, 23 Mar 2005 09:18:20 +0000 (09:18 +0000)] 
[project @ 2005-03-23 09:18:20 by simonmar]
Comment out tuple instances that are now in GHC 6.5.

14 years ago[project @ 2005-02-09 11:28:07 by simonmar]
simonmar [Wed, 9 Feb 2005 11:28:07 +0000 (11:28 +0000)] 
[project @ 2005-02-09 11:28:07 by simonmar]
Add a comment

14 years ago[project @ 2005-02-04 13:08:28 by simonmar]
simonmar [Fri, 4 Feb 2005 13:08:28 +0000 (13:08 +0000)] 
[project @ 2005-02-04 13:08:28 by simonmar]
unbreak nofib

14 years ago[project @ 2004-11-30 15:28:12 by simonmar]
simonmar [Tue, 30 Nov 2004 15:28:13 +0000 (15:28 +0000)] 
[project @ 2004-11-30 15:28:12 by simonmar]
Rename Maybe to avoid class with Haskell 98 module of the same name.

14 years ago[project @ 2004-11-30 15:25:32 by simonmar]
simonmar [Tue, 30 Nov 2004 15:25:33 +0000 (15:25 +0000)] 
[project @ 2004-11-30 15:25:32 by simonmar]
The compiler now detects local modules that overlap with package
modules, so we must rename Random to avoid the clash.

14 years ago[project @ 2004-11-09 11:48:07 by simonpj]
simonpj [Tue, 9 Nov 2004 11:48:07 +0000 (11:48 +0000)] 
[project @ 2004-11-09 11:48:07 by simonpj]
Add another puzzle-solving program

14 years ago[project @ 2004-09-01 08:30:04 by simonmar]
simonmar [Wed, 1 Sep 2004 08:30:43 +0000 (08:30 +0000)] 
[project @ 2004-09-01 08:30:04 by simonmar]
accept output

14 years ago[project @ 2004-08-10 11:39:00 by simonpj]
simonpj [Tue, 10 Aug 2004 11:39:00 +0000 (11:39 +0000)] 
[project @ 2004-08-10 11:39:00 by simonpj]
Add README

15 years ago[project @ 2003-10-09 11:52:01 by simonpj]
simonpj [Thu, 9 Oct 2003 11:52:02 +0000 (11:52 +0000)] 
[project @ 2003-10-09 11:52:01 by simonpj]
Update mate problem

15 years ago[project @ 2003-10-06 12:51:40 by simonpj]
simonpj [Mon, 6 Oct 2003 12:51:40 +0000 (12:51 +0000)] 
[project @ 2003-10-06 12:51:40 by simonpj]
Add bernouilli

16 years ago[project @ 2003-06-07 11:20:49 by ross]
ross [Sat, 7 Jun 2003 11:20:49 +0000 (11:20 +0000)] 
[project @ 2003-06-07 11:20:49 by ross]
Add $($(HcFlavour)_HAPPY_OPTS) to HAPPY_OPTS, where HcFlavour defaults to GHC.
Packages may GHC_HAPPY_OPTS if they don't want -agc with ghc.

16 years ago[project @ 2003-05-08 09:14:38 by simonmar]
simonmar [Thu, 8 May 2003 09:14:38 +0000 (09:14 +0000)] 
[project @ 2003-05-08 09:14:38 by simonmar]
Add lcss to the mix.

16 years ago[project @ 2003-03-10 08:58:31 by simonpj]
simonpj [Mon, 10 Mar 2003 08:58:31 +0000 (08:58 +0000)] 
[project @ 2003-03-10 08:58:31 by simonpj]
Add bernouilli test

16 years ago[project @ 2003-03-08 19:02:39 by panne]
panne [Sat, 8 Mar 2003 19:02:39 +0000 (19:02 +0000)] 
[project @ 2003-03-08 19:02:39 by panne]
Fixed some broken/redirected/canonicalized links found by a very picky
link checker.

16 years ago[project @ 2003-02-17 15:11:08 by simonpj]
simonpj [Mon, 17 Feb 2003 15:11:08 +0000 (15:11 +0000)] 
[project @ 2003-02-17 15:11:08 by simonpj]
More notes

16 years ago[project @ 2003-01-09 15:51:10 by simonmar]
simonmar [Thu, 9 Jan 2003 15:51:10 +0000 (15:51 +0000)] 
[project @ 2003-01-09 15:51:10 by simonmar]
Suppress printing of the runstdtest command line in GHCi mode.

16 years ago[project @ 2003-01-09 15:31:40 by simonmar]
simonmar [Thu, 9 Jan 2003 15:31:40 +0000 (15:31 +0000)] 
[project @ 2003-01-09 15:31:40 by simonmar]
More fixes for the GHCi way

16 years ago[project @ 2003-01-09 14:59:22 by simonmar]
simonmar [Thu, 9 Jan 2003 14:59:22 +0000 (14:59 +0000)] 
[project @ 2003-01-09 14:59:22 by simonmar]
Fixes to get NoFibWithGHCi going again.

16 years ago[project @ 2002-11-22 11:13:28 by simonmar]
simonmar [Fri, 22 Nov 2002 11:13:28 +0000 (11:13 +0000)] 
[project @ 2002-11-22 11:13:28 by simonmar]
update sample output to match reality

16 years ago[project @ 2002-11-22 10:32:24 by simonmar]
simonmar [Fri, 22 Nov 2002 10:32:24 +0000 (10:32 +0000)] 
[project @ 2002-11-22 10:32:24 by simonmar]
Add Andrew Bromage's lcss code as a spectral benchmark.

16 years ago[project @ 2002-11-19 14:31:31 by simonmar]
simonmar [Tue, 19 Nov 2002 14:31:31 +0000 (14:31 +0000)] 
[project @ 2002-11-19 14:31:31 by simonmar]
include boilerplate.mk in the right place

16 years ago[project @ 2002-11-15 17:25:54 by simonmar]
simonmar [Fri, 15 Nov 2002 17:25:54 +0000 (17:25 +0000)] 
[project @ 2002-11-15 17:25:54 by simonmar]
oops, I broke overriding of suffix rules when I moved the include of
suffix.mk into target.mk.  Hence, hack around it here by moving
nofib's suffix.mk after the include of target.mk.

16 years ago[project @ 2002-11-07 13:39:55 by simonmar]
simonmar [Thu, 7 Nov 2002 13:39:56 +0000 (13:39 +0000)] 
[project @ 2002-11-07 13:39:55 by simonmar]
Changes for 'make mode=fast': I've picked on some of the
longest-running programs, but left most of them alone.  Most runtimes
are <1s, but a few are 2-3s.

16 years ago[project @ 2002-11-07 13:17:43 by simonmar]
simonmar [Thu, 7 Nov 2002 13:17:44 +0000 (13:17 +0000)] 
[project @ 2002-11-07 13:17:43 by simonmar]
Make 'make mode=fast' work, and tweak the settings slightly so that
all of the tests run in <1s (except exp3_8, which runs in just over
1s), with most of the tests running in about 0.3s.  This is on a 1Gz
PIII.

The normal settings are unchanged.  In fact in some cases, the fast
settings are slower than the normal settings... at some point we'll
need to bump all the normal settings too, but that needs to be done in
one large sweep so we invalidate all the old results in one go.

16 years ago[project @ 2002-11-07 13:14:16 by simonmar]
simonmar [Thu, 7 Nov 2002 13:14:16 +0000 (13:14 +0000)] 
[project @ 2002-11-07 13:14:16 by simonmar]
wibbles to do with the FAST/NORM/SLOW modes.  We default to the normal
.stdout file if the .$(mode)stdout file doesn't exist, and similarly
for stderr.

16 years ago[project @ 2002-09-07 11:25:34 by panne]
panne [Sat, 7 Sep 2002 11:25:34 +0000 (11:25 +0000)] 
[project @ 2002-09-07 11:25:34 by panne]
Remove the clash of runstdtest's flag -t and the one used for this
example by renaming the latter to -u.

16 years ago[project @ 2002-08-29 14:46:50 by simonpj]
simonpj [Thu, 29 Aug 2002 14:46:50 +0000 (14:46 +0000)] 
[project @ 2002-08-29 14:46:50 by simonpj]
Add chess mate test

17 years ago[project @ 2002-07-10 13:11:25 by simonpj]
simonpj [Wed, 10 Jul 2002 13:11:25 +0000 (13:11 +0000)] 
[project @ 2002-07-10 13:11:25 by simonpj]
More notes

17 years ago[project @ 2002-07-04 09:11:54 by simonmar]
simonmar [Thu, 4 Jul 2002 09:11:54 +0000 (09:11 +0000)] 
[project @ 2002-07-04 09:11:54 by simonmar]
Fix up this makefile after I renamed Fulsom.hs

17 years ago[project @ 2002-07-03 12:49:01 by simonmar]
simonmar [Wed, 3 Jul 2002 12:49:01 +0000 (12:49 +0000)] 
[project @ 2002-07-03 12:49:01 by simonmar]
Update expected output (the checked in version appeared to be completely bogus).

17 years ago[project @ 2002-07-03 12:41:57 by simonmar]
simonmar [Wed, 3 Jul 2002 12:43:58 +0000 (12:43 +0000)] 
[project @ 2002-07-03 12:41:57 by simonmar]
Rename the main module to Main.hs, to avoid confusing GHCi

17 years ago[project @ 2002-07-03 12:35:43 by simonmar]
simonmar [Wed, 3 Jul 2002 12:35:43 +0000 (12:35 +0000)] 
[project @ 2002-07-03 12:35:43 by simonmar]
fix up options

17 years ago[project @ 2002-07-03 12:30:33 by simonmar]
simonmar [Wed, 3 Jul 2002 12:30:33 +0000 (12:30 +0000)] 
[project @ 2002-07-03 12:30:33 by simonmar]
use STDIN_FILE rather than SRC_RUNTEST_OPTS

17 years ago[project @ 2002-07-03 12:25:46 by simonmar]
simonmar [Wed, 3 Jul 2002 12:25:46 +0000 (12:25 +0000)] 
[project @ 2002-07-03 12:25:46 by simonmar]
use FAST_OPTS etc rather than SRC_RUNTEST_OPTS

17 years ago[project @ 2002-07-03 12:19:09 by simonmar]
simonmar [Wed, 3 Jul 2002 12:19:09 +0000 (12:19 +0000)] 
[project @ 2002-07-03 12:19:09 by simonmar]
This test uses args, not stdin

17 years ago[project @ 2002-06-06 16:02:25 by simonpj]
simonpj [Thu, 6 Jun 2002 16:02:25 +0000 (16:02 +0000)] 
[project @ 2002-06-06 16:02:25 by simonpj]
Add a few notes

17 years ago[project @ 2002-05-20 12:26:25 by simonmar]
simonmar [Mon, 20 May 2002 12:26:25 +0000 (12:26 +0000)] 
[project @ 2002-05-20 12:26:25 by simonmar]
Fix to make NoFib work with GHCi again.

17 years ago[project @ 2002-05-17 11:47:06 by simonpj]
simonpj [Fri, 17 May 2002 11:47:06 +0000 (11:47 +0000)] 
[project @ 2002-05-17 11:47:06 by simonpj]
last piece

17 years ago[project @ 2002-05-17 11:46:45 by simonpj]
simonpj [Fri, 17 May 2002 11:46:45 +0000 (11:46 +0000)] 
[project @ 2002-05-17 11:46:45 by simonpj]
Add last piece puzzle

17 years ago[project @ 2002-04-01 13:58:09 by simonpj]
simonpj [Mon, 1 Apr 2002 13:58:09 +0000 (13:58 +0000)] 
[project @ 2002-04-01 13:58:09 by simonpj]
Notes on mandel

17 years ago[project @ 2002-03-15 16:14:04 by rje]
rje [Fri, 15 Mar 2002 16:14:07 +0000 (16:14 +0000)] 
[project @ 2002-03-15 16:14:04 by rje]
Extended the imaginary tests so that they can be run in slow mode by setting
mode=slow.

Some of these tests required minor changes in order to be able to run at
more than one speed. In particular, gen_regexps now only outputs the number
of chars in its output, rather than all the regular expressions. This is
to avoid the slow stdout file having to be insanely huge.

I've done a half-hearted attempt at making the speculative tests go slowly
too, but that is too incomplete and too messy to be committed just yet.

17 years ago[project @ 2002-03-15 16:07:58 by rje]
rje [Fri, 15 Mar 2002 16:07:58 +0000 (16:07 +0000)] 
[project @ 2002-03-15 16:07:58 by rje]
Add ability for NoFib to be run with various different modes.

The mode is set by changing a variable called "mode".

This allows tests to be run for different lengths of time, depending on what kin
d of test is wanted.

Current modes are:

<unset> = default mode. Like previous NoFib.
"slow" = tries to run for around 10-20 seconds on my workstation.

A test is given several modes by giving it the following:
        <test>.stdout -> <test>.<mode>stdout
        <test>.stdin -> <test>.<mode>stdin

Args are set in the makefile as

<mode>_OPTS

17 years ago[project @ 2002-03-06 13:32:59 by keithw]
keithw [Wed, 6 Mar 2002 13:32:59 +0000 (13:32 +0000)] 
[project @ 2002-03-06 13:32:59 by keithw]
Add way to ==nofib== banner (it's now `==nofib==' for normal and
`==nofib_way==' for way `way'.  At the moment nofib-analyse only looks
for `==nofib==', and so now it won't be misled by, e.g., excessive
runtimes for the ticky way.

To collect stats from non-standard ways, nofib-analyse will have to be
modified.

17 years ago[project @ 2002-03-02 18:13:18 by sof]
sof [Sat, 2 Mar 2002 18:13:19 +0000 (18:13 +0000)] 
[project @ 2002-03-02 18:13:18 by sof]
make it work with new libs

17 years ago[project @ 2002-03-02 18:12:35 by sof]
sof [Sat, 2 Mar 2002 18:12:35 +0000 (18:12 +0000)] 
[project @ 2002-03-02 18:12:35 by sof]
HS_SRCS:= --> EXCLUDED_SRCS

17 years ago[project @ 2002-02-12 15:17:35 by simonmar]
simonmar [Tue, 12 Feb 2002 15:17:36 +0000 (15:17 +0000)] 
[project @ 2002-02-12 15:17:35 by simonmar]
Switch over to the new hierarchical libraries
---------------------------------------------

This commit reorganises our libraries to use the new hierarchical
module namespace extension.

The basic story is this:

   - fptools/libraries contains the new hierarchical libraries.
     Everything in here is "clean", i.e. most deprecated stuff has
     been removed.

- fptools/libraries/base is the new base package
  (replacing "std") and contains roughly what was previously
  in std, lang, and concurrent, minus deprecated stuff.
  Things that are *not allowed* in libraries/base include:
Addr, ForeignObj, ByteArray, MutableByteArray,
_casm_, _ccall_, ``'', PrimIO

  For ByteArrays and MutableByteArrays we use UArray and
  STUArray/IOUArray respectively now.

  Modules previously called PrelFoo are now under
  fptools/libraries/GHC.  eg. PrelBase is now GHC.Base.

- fptools/libraries/haskell98 provides the Haskell 98 std.
  libraries (Char, IO, Numeric etc.) as a package.  This
  package is enabled by default.

- fptools/libraries/network is a rearranged version of
  the existing net package (the old package net is still
  available; see below).

- Other packages will migrate to fptools/libraries in
  due course.

     NB. you need to checkout fptools/libraries as well as
     fptools/hslibs now.  The nightly build scripts will need to be
     tweaked.

   - fptools/hslibs still contains (almost) the same stuff as before.
     Where libraries have moved into the new hierarchy, the hslibs
     version contains a "stub" that just re-exports the new version.
     The idea is that code will gradually migrate from fptools/hslibs
     into fptools/libraries as it gets cleaned up, and in a version or
     two we can remove the old packages altogether.

   - I've taken the opportunity to make some changes to the build
     system, ripping out the old hslibs Makefile stuff from
     mk/target.mk; the new package building Makefile code is in
     mk/package.mk (auto-included from mk/target.mk).

     The main improvement is that packages now register themselves at
     make boot time using ghc-pkg, and the monolithic package.conf
     in ghc/driver is gone.

     I've updated the standard packages but haven't tested win32,
     graphics, xlib, object-io, or OpenGL yet.  The Makefiles in
     these packages may need some further tweaks, and they'll need
     pkg.conf.in files added.

   - Unfortunately all this rearrangement meant I had to bump the
     interface-file version and create a bunch of .hi-boot-6 files :-(

17 years ago[project @ 2002-02-01 10:22:48 by simonmar]
simonmar [Fri, 1 Feb 2002 10:22:48 +0000 (10:22 +0000)] 
[project @ 2002-02-01 10:22:48 by simonmar]
An ugly hack is now needed in here after my build system changes the
other day.  Oh dear, I hope I haven't created a monster.  Well, it can
always be backed out I suppose.

17 years ago[project @ 2002-02-01 10:12:46 by simonmar]
simonmar [Fri, 1 Feb 2002 10:12:46 +0000 (10:12 +0000)] 
[project @ 2002-02-01 10:12:46 by simonmar]
I found a fossil

17 years ago[project @ 2002-01-29 11:04:05 by simonmar]
simonmar [Tue, 29 Jan 2002 11:04:05 +0000 (11:04 +0000)] 
[project @ 2002-01-29 11:04:05 by simonmar]
Allow the stdin file to be specified by setting STDIN_FILE.

17 years ago[project @ 2002-01-29 11:03:21 by simonmar]
simonmar [Tue, 29 Jan 2002 11:03:22 +0000 (11:03 +0000)] 
[project @ 2002-01-29 11:03:21 by simonmar]
Tweaks to make the real suite run with GHCi.

17 years ago[project @ 2002-01-29 10:36:40 by simonmar]
simonmar [Tue, 29 Jan 2002 10:36:40 +0000 (10:36 +0000)] 
[project @ 2002-01-29 10:36:40 by simonmar]
Update sample output, and fix the test to work with current GHC.

17 years ago[project @ 2002-01-29 10:36:06 by simonmar]
simonmar [Tue, 29 Jan 2002 10:36:06 +0000 (10:36 +0000)] 
[project @ 2002-01-29 10:36:06 by simonmar]
Add scs, which was a while back but never enabled.

17 years ago[project @ 2002-01-28 15:51:29 by simonmar]
simonmar [Mon, 28 Jan 2002 15:51:29 +0000 (15:51 +0000)] 
[project @ 2002-01-28 15:51:29 by simonmar]
Wibbles to the GHCi support.  Most of the spectral suite goes through now.

17 years ago[project @ 2002-01-28 14:15:02 by simonmar]
simonmar [Mon, 28 Jan 2002 14:15:03 +0000 (14:15 +0000)] 
[project @ 2002-01-28 14:15:02 by simonmar]
More hacks to run the Nofib with GHCi:

    - allow arguments to be given the program: these are now
      set using PROG_ARGS in the Makefile rather than just
      adding to SRC_RUNTEST_OPTS as we did before.

    - pass $(HC_OPTS) to GHCi (after filtering out some
      undesirable stuff).

17 years ago[project @ 2002-01-28 13:48:33 by simonmar]
simonmar [Mon, 28 Jan 2002 13:48:33 +0000 (13:48 +0000)] 
[project @ 2002-01-28 13:48:33 by simonmar]
Nofib support for running with GHCi.  Run like this:

    make NoFibWithGHCi=YES

or set NoFibWithGHCi=YES in your build.mk.

17 years ago[project @ 2002-01-02 12:37:25 by simonmar]
simonmar [Wed, 2 Jan 2002 12:37:25 +0000 (12:37 +0000)] 
[project @ 2002-01-02 12:37:25 by simonmar]
Fix compilation problems on HEAD (shiftL# & friends now come from PrelBase).

17 years ago[project @ 2001-12-10 16:37:25 by simonmar]
simonmar [Mon, 10 Dec 2001 16:37:25 +0000 (16:37 +0000)] 
[project @ 2001-12-10 16:37:25 by simonmar]
Add a note about maillist being a very badly written program :-)

17 years ago[project @ 2001-11-27 12:18:11 by simonmar]
simonmar [Tue, 27 Nov 2001 12:18:11 +0000 (12:18 +0000)] 
[project @ 2001-11-27 12:18:11 by simonmar]
Add things I discovered on my last performance sweep (this was several
weeks ago).

17 years ago[project @ 2001-11-02 14:27:36 by simonpj]
simonpj [Fri, 2 Nov 2001 14:27:36 +0000 (14:27 +0000)] 
[project @ 2001-11-02 14:27:36 by simonpj]
new program

17 years ago[project @ 2001-10-18 10:27:15 by rrt]
rrt [Thu, 18 Oct 2001 10:28:06 +0000 (10:28 +0000)] 
[project @ 2001-10-18 10:27:15 by rrt]
catchAllIO -> catch

17 years ago[project @ 2001-10-01 12:23:42 by rrt]
rrt [Mon, 1 Oct 2001 12:23:42 +0000 (12:23 +0000)] 
[project @ 2001-10-01 12:23:42 by rrt]
Stop using ANNA_DIR environment variable, which for various sad reasons
doesn't work on Windows.

17 years ago[project @ 2001-08-31 08:43:55 by simonmar]
simonmar [Fri, 31 Aug 2001 08:43:55 +0000 (08:43 +0000)] 
[project @ 2001-08-31 08:43:55 by simonmar]
Add an (Ord a) context on an Ix instance which is required now.

17 years ago[project @ 2001-08-20 11:41:13 by simonpj]
simonpj [Mon, 20 Aug 2001 11:41:13 +0000 (11:41 +0000)] 
[project @ 2001-08-20 11:41:13 by simonpj]
Add two digits-of-e programs

18 years ago[project @ 2001-07-16 12:01:43 by simonmar]
simonmar [Mon, 16 Jul 2001 12:01:43 +0000 (12:01 +0000)] 
[project @ 2001-07-16 12:01:43 by simonmar]
-syslib lang unnecessary

18 years ago[project @ 2001-07-03 17:08:17 by rrt]
rrt [Tue, 3 Jul 2001 17:08:17 +0000 (17:08 +0000)] 
[project @ 2001-07-03 17:08:17 by rrt]
Tart up comments

18 years ago[project @ 2001-07-03 15:54:34 by rrt]
rrt [Tue, 3 Jul 2001 15:54:34 +0000 (15:54 +0000)] 
[project @ 2001-07-03 15:54:34 by rrt]
More bogus results

18 years ago[project @ 2001-07-03 14:00:28 by rrt]
rrt [Tue, 3 Jul 2001 14:00:28 +0000 (14:00 +0000)] 
[project @ 2001-07-03 14:00:28 by rrt]
Update output because of changed defaulting rules

18 years ago[project @ 2001-05-27 17:37:59 by sof]
sof [Sun, 27 May 2001 17:37:59 +0000 (17:37 +0000)] 
[project @ 2001-05-27 17:37:59 by sof]
basic mingw headers doesn't define M_PI; make fft.c cope

18 years ago[project @ 2001-04-04 11:23:57 by sewardj]
sewardj [Wed, 4 Apr 2001 11:23:57 +0000 (11:23 +0000)] 
[project @ 2001-04-04 11:23:57 by sewardj]
Another out variant (FP wibbles).

18 years ago[project @ 2001-04-04 11:21:09 by sewardj]
sewardj [Wed, 4 Apr 2001 11:21:09 +0000 (11:21 +0000)] 
[project @ 2001-04-04 11:21:09 by sewardj]
Another output variant (FP wibbles).

18 years ago[project @ 2001-04-04 11:17:13 by sewardj]
sewardj [Wed, 4 Apr 2001 11:17:13 +0000 (11:17 +0000)] 
[project @ 2001-04-04 11:17:13 by sewardj]
Yet another output variant.

18 years ago[project @ 2001-04-04 11:12:09 by sewardj]
sewardj [Wed, 4 Apr 2001 11:12:09 +0000 (11:12 +0000)] 
[project @ 2001-04-04 11:12:09 by sewardj]
Add yet another output variant.  One could be forgiven for suspecting that
this program is not numerically stable, and gives not just differing coords
but also different numbers of points as a result.

18 years ago[project @ 2001-04-03 20:12:37 by hwloidl]
hwloidl [Tue, 3 Apr 2001 20:12:38 +0000 (20:12 +0000)] 
[project @ 2001-04-03 20:12:37 by hwloidl]
Cleanup in parallel section of nofib suite. The fibish fcts are ok, i.e.
std way should succeed, parallel way should build.
Parallel way has to be run by hand (problems redirecting stdout), so not
automatic check whether output is identical to the provided file.
Added a bunch of test prgs for GdH in distributed section. Still need
to be integrated into nofib build machinery (currently standalone Makefiles).
Modified Makefiles to work with a parallel build. Have to omit runtest for now.

18 years ago[project @ 2001-03-30 11:01:12 by simonmar]
simonmar [Fri, 30 Mar 2001 11:01:12 +0000 (11:01 +0000)] 
[project @ 2001-03-30 11:01:12 by simonmar]
put back hartel subdir that hwl removed.

18 years ago[project @ 2001-03-22 03:51:12 by hwloidl]
hwloidl [Thu, 22 Mar 2001 03:51:13 +0000 (03:51 +0000)] 
[project @ 2001-03-22 03:51:12 by hwloidl]
-*- outline -*-
Time-stamp: <Thu Mar 22 2001 03:50:16 Stardate: [-30]6365.79 hwloidl>

This commit covers changes in GHC to get GUM (way=mp) and GUM/GdH (way=md)
working. It is a merge of my working version of GUM, based on GHC 4.06,
with GHC 4.11. Almost all changes are in the RTS (see below).

GUM is reasonably stable, we used the 4.06 version in large-ish programs for
recent papers. Couple of things I want to change, but nothing urgent.
GUM/GdH has just been merged and needs more testing. Hope to do that in the
next weeks. It works in our working build but needs tweaking to run.
GranSim doesn't work yet (*sigh*). Most of the code should be in, but needs
more debugging.

ToDo: I still want to make the following minor modifications before the release
- Better wrapper skript for parallel execution [ghc/compiler/main]
- Update parallel docu: started on it but it's minimal [ghc/docs/users_guide]
- Clean up [nofib/parallel]: it's a real mess right now (*sigh*)
- Update visualisation tools (minor things only IIRC) [ghc/utils/parallel]
- Add a Klingon-English glossary

* RTS:

Almost all changes are restricted to ghc/rts/parallel and should not
interfere with the rest. I only comment on changes outside the parallel
dir:

- Several changes in Schedule.c (scheduling loop; createThreads etc);
  should only affect parallel code
- Added ghc/rts/hooks/ShutdownEachPEHook.c
- ghc/rts/Linker.[ch]: GUM doesn't know about Stable Names (ifdefs)!!
- StgMiscClosures.h: END_TSO_QUEUE etc now defined here (from StgMiscClosures.hc)
                     END_ECAF_LIST was missing a leading stg_
- SchedAPI.h: taskStart now defined in here; it's only a wrapper around
              scheduleThread now, but might use some init, shutdown later
- RtsAPI.h: I have nuked the def of rts_evalNothing

* Compiler:

- ghc/compiler/main/DriverState.hs
  added PVM-ish flags to the parallel way
  added new ways for parallel ticky profiling and distributed exec

- ghc/compiler/main/DriverPipeline.hs
  added a fct run_phase_MoveBinary which is called with way=mp after linking;
  it moves the bin file into a PVM dir and produces a wrapper script for
  parallel execution
  maybe cleaner to add a MoveBinary phase in DriverPhases.hs but this way
  it's less intrusive and MoveBinary makes probably only sense for mp anyway

* Nofib:

- nofib/spectral/Makefile, nofib/real/Makefile, ghc/tests/programs/Makefile:
  modified to skip some tests if HWL_NOFIB_HACK is set; only tmp to record
  which test prgs cause problems in my working build right now

18 years ago[project @ 2001-03-14 10:40:47 by simonmar]
simonmar [Wed, 14 Mar 2001 10:40:47 +0000 (10:40 +0000)] 
[project @ 2001-03-14 10:40:47 by simonmar]
remove import of fromInt

18 years ago[project @ 2001-03-14 09:47:17 by simonmar]
simonmar [Wed, 14 Mar 2001 09:47:17 +0000 (09:47 +0000)] 
[project @ 2001-03-14 09:47:17 by simonmar]
add -package lang

18 years ago[project @ 2001-03-12 16:00:46 by sewardj]
sewardj [Mon, 12 Mar 2001 16:00:46 +0000 (16:00 +0000)] 
[project @ 2001-03-12 16:00:46 by sewardj]
Fix imports.

18 years ago[project @ 2001-03-12 15:30:01 by sewardj]
sewardj [Mon, 12 Mar 2001 15:30:01 +0000 (15:30 +0000)] 
[project @ 2001-03-12 15:30:01 by sewardj]
import GlaExts is now out-of-date.