| Commit message (Collapse) | Author | Age |
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4331 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
option. This is now the preferred format, as per the recent
discussion on the developers' mailing list.
Comment out a leftover non-commented part of the non-exising "x"
option.
Reformat the command line descriptions slightly.
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4330 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The option is "-b <mode>", or "-buffering=<mode>", where mode can be
"no", "line", or "full". This option uses the ANSI C function call
setvbuf() to select the corresponding buffering mode.
The buffering mode can be selected only once in a portable programme,
and it has to be selected before the first output occurs. This means
that it has to be implemented in the flesh.
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4329 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
| |
included file is actually found. The previous code required that each
included file be in the same directory as the including file. The new
code treats using modules and including files in the same manner.
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4328 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
build. This flag is equivalent to OPTIMISE=yes and DEBUG=yes.
Additional compiler options {C,CXX,F77,F90}_PROFILE_FLAGS are also
introduced.
The configuration stage sets the F77 flags to the F90 flags if an F90
compiler is found. This flag setting was done too early, namely
before the default values for the F90 flags were set. This flag
setting has been moved to a later time.
The link command used undefined make variables $(OPTIMISE_C) etc.
These variables have been removed. Instead $(CXX_OPTIMISE_FLAGS)
etc. are added to the link statement. This makes the linker pick up
the correct flags e.g. for profiling.
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4327 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4326 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4325 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
| |
* oops, another place where INOUT needs to be mentioned
* clarify that OUT or INOUT array arguments may indeed be modified
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4324 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
| |
http://www.cactuscode.org/old/pipermail/developers/2006-June/002882.html
and following discussion
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4323 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
reduces FARGUMENTS line length in cctk.h
http://www.cactuscode.org/old/pipermail/patches/2006-February/000138.html
"Spaces in _CCTK_FARGUMENTS removed to use as less space as possible.
This leaves thorn writers more space for their own variables. Space can
be a problem because of Fortran standard and/or compiler limitations."
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4322 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4321 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
| |
Tell people to use capabilities instead of make.configuration.defn.
Update FAQ entry about detecting MPI in make.configuration.defn.
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4320 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4319 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
1) Got rid of
--instantiate=used
which only irritates recent compilers.
2) Broke off warning settings which were previously part of
e.g. F90_OPTIMISE_FLAGS into F90_WARN_FLAGS
so now you see most of the "loop unroll" etc warnings only if
WARN is on.
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4318 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4317 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
WarnLevel bold off unless output is to tty
http://www.cactuscode.org/old/pipermail/patches/2006-April/000161.html
I got tired of seeing all the terminal control characters in Cactus
output that was piped to files.
Tested by watching Cactus output on a terminal, then piping ouput from the
same program to a file. E.g. in bash
make config-testsuite 2>&1 |tee > ts
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4316 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
GridFuncStuff bug fix and cleanup
http://www.cactuscode.org/old/pipermail/patches/2006-May/000165.html
It fixes two minor bugs (that is, bugs that weren't bugging anybody right
now). It also includes an overall cleanup of the code, which begain with
the first bug fix, and resulted in the discovery of the second bug.
The first bug fix was to put in
use strict;
(According to L. Wall, a Perl file without use strict constitutes a bug.)
This resulted in lots of errors about variables needing their scope to be
specified. So I did this.
At the end, I found that I couldn't fix a certain occurrance of $group.
There was no sensible way that it was being set. The code was to produce
a warning concerning creating groups with mixed dimensions. See
'warn_mixeddim_gfs'.
The Perl code was taking a global value of $group set at build time to be
the last group in a loop. However, the offending group is determined at
run time in CreateThornGroupInitializers. So the warning message would
typically be wrong.
Affect code written in
bindings/Variables/<thorn>.c
Also: Perl 5-ied function calls
deleted great wads of commented-out code
Testing
-------
Ran testsuites on my laptop with Whisky benchmark thornlist.
No change was detected in the results.
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4315 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fortran REAL*16 detection
http://www.cactuscode.org/old/pipermail/patches/2006-April/000164.html
... a patch that makes Cactus directly detect Fortran support
for a REAL*16 type.
The problem has been that Cactus based CCTK_REAL16 on support for the C type
'long double'. This would cause it to create Fortran code for REAL*16, a
type which was sometimes not supported.
In particular, on a 64-bit Athlon machine with Gnu 4.1.0, gfortran
defines REAL*8 but not REAL*16, while gcc defines 'long double'.
This caused thorn AEIThorns/Fortran to fail to compile.
This patch causes the configuration to directly compile Fortran code that
features REAL*16, and define CCTK_REAL16 only if that compiles *and* C's
'long double' is supported.
When you make config, you will see a new line about Fortran REAL*16.
I have tested the enclosed patch on
my laptop with Gnu 4.10 and Intel 9.0 compilers
64-bit AMD Athlon with same compilers
POWER4 with IBM XL compilers
BUT NOTE: I have severe doubts about using REAL*16 on any of these
architectures. That 16-byte (128-bit) type is not natively supported
on the Xeon, Athlon, or POWER machines. While some compilers support
such types, calculation is done in software, not on the FPU.
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4314 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
http://www.cactuscode.org/old/pipermail/patches/2006-April/000160.html
... a patch for the known-architectures file for Darwin for
the XLF compiler.
Primarily it is to improve the speed, by making the optimization
options to be
-O3 -qhot -qarch=auto -qtune=auto -qcache=auto
This gets quite good performance, but moves instructions. The various
"auto's" cause the compiler to generate optimal code for the platform
on which the compiler is running.
Also, I made it supress the congratulatory message ("Yay, I compiled").
Also, I test the compiler version major and minor numbers so that darned
"-no-cpp-precomp" option is included only when the version is less than 3.3.
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4313 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
aix known-architectures for XLC/XLF
http://www.cactuscode.org/old/pipermail/patches/2006-April/000159.html
> ... a patch for the AIX known-architectures file, meant to
> improve performance for the IBM XLC/XLF compilers.
>
> I made the default optimization to be
> -O3 -qarch=auto -qtune=auto -qcache=auto -qhot
> for all the IBM compilers.
>
> I also made XLF to suppress congratulatory messages ("Yay! I compiled!)
> on successful compilation.
>
> I tested -02 through -O5. On BSSNCarpet runs, "-O3 -qhot" performed
> best--but beware, it moves instructions. While turning on -qstrict
> fixes this, it also degrades performance. The -qarch=auto causes the
> code to be customized for the architecture on which it is compiled.
> The higher levels turn on -qipa which does global optimizations and
> work best if this option is also supplied at link time.
>
> See http://publib.boulder.ibm.com/infocenter/comphelp/v7v91/index.jsp
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4312 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4311 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
|
|
|
| |
http://www.cactuscode.org/old/pipermail/patches/2006-May/000180.html
This documents the 5 (!) separate places in the code which must be changed
to add a new command-line option. See also the discussion in
http://www.cactuscode.org/mailman/private/cactusmaint/2006-May/021429.html
http://www.cactuscode.org/mailman/private/cactusmaint/2006-May/021430.html
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4310 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
|
|
| |
CCTK_MaxTimeLevelsGI()
CCTK_MaxTimeLevelsGN()
CCTK_MaxTimeLevelsVI()
CCTK_MaxTimeLevelsVN()
to be in alphabetical order (and all come before CCTK_MyProc())
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4309 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4308 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4307 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
CCTK_ReduceLocScalar()
CCTK_ReduceLocArrayToArray1D()
CCTK_ReduceLocArrayToArray2D()
CCTK_ReduceLocArrayToArray3D()
don't modify the GH they're passed (I verified this by looking at
the prototypes in the flesh), i.e. have that argument be of type
const cGH *GH
rather than
cGH *GH
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4306 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4305 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4304 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
| |
(no change in non-comment latex code)
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4303 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
|
|
| |
* use modern FunctionDescription environment
* document that GH argument is pointer-to-const
(i.e. document that CCTK_SyncGroup doesn't modify the GH)
* document that this is a collective operation
* add an example
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4302 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
| |
variables.
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4301 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4300 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4299 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
|
|
| |
CCTK_GroupTagsTable()
CCTK_GroupTagsTableI()
(this change is *only* to comment wording -- there's no change to
executable code)
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4298 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
| |
CCTK_GroupTagsTable() and CCTK_GroupTagsTableI()
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4297 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
| |
and made it include libmpi_cxx
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4296 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
|
| |
http://www.cactuscode.org/old/pipermail/patches/2006-February/000147.html
based on discussions at today's access-grid "Cactus call":
"Inherits" et al get *groups* of variables from other *implementations*.
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4295 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
| |
Define HAVE_CCTK_BYTE etc.
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4294 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
| |
New command line option to stop after printing the schedule
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4293 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4292 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4291 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4290 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4289 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4288 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4287 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4286 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The symptoms were that if you
(a) set the same parameter twice, *and*
(b) the second setting is an invalid value,
then current-CVS Cactus core-dumps inside the flesh trying to free()
a garbage (uninitialized) pointer.
For example, the following par file suffices to trigger the core-dump,
using only flesh parameters (*no* thorns activated):
# this par file causes a core dump in current-CVS Cactus
Cactus::cctk_timer_output = "off"
Cactus::cctk_timer_output = "foo" # an illegal value for this parameter
Looking at the code,
599 /* check if a parameter is set more than once in a parfile */
600 if (cctk_parameter_set_mask == PARAMETER_RECOVERY_PRE &&
601 param->props->n_set > 0)
602 {
603 if (retval == 0)
604 {
605 new_value = CCTK_ParameterValString (param->props->name,
606 param->props->thorn);
607 retval = strcmp (old_value, new_value) ? -10 : -11;
608 }
609 free (new_value);
610 }
Notice that the free() at line 609 may be executed even if new_value
was *not* assigned a value at line 605. Alas, new_value is a local
variable in this function, and is not explicitly initialized, so it
has a garbage value initially... and line 605 is its only assignment
in this function.
The fix is easy -- just move the free() up to right after line 607,
so it's executed if and only if new_value has been assigned.
patch was
http://www.cactuscode.org/old/pipermail/patches/2006-April/000162.html
developers discussion was
http://www.cactuscode.org/old/pipermail/developers/2006-April/001758.html
http://www.cactuscode.org/old/pipermail/developers/2006-April/001759.html
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4285 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
| |
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4284 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
| |
There seem to be new Darwin releases every week or so.
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4283 17b73243-c579-4c4c-a9d2-2d5706c11dac
|
|
|
|
|
|
|
|
| |
(old versions of g++) don't support restrict in arrays which are
structure members.
git-svn-id: http://svn.cactuscode.org/flesh/trunk@4282 17b73243-c579-4c4c-a9d2-2d5706c11dac
|