platform/upstream/isl.git
12 years agoscheduler: allow to bound the coefficients in the calculated schedule
Tobias Grosser [Sun, 19 Feb 2012 13:05:50 +0000 (14:05 +0100)]
scheduler: allow to bound the coefficients in the calculated schedule

isl already supports the bounding of the constant term of the calculated
schedule, but within polybench there are several cases where bounding
the coefficients of parameter and variable dimensions is also needed.

With a bound of 20 Polly can run the isl scheduler on all polybench 2.0 kernels
and takes for each less than 5 seconds, whereas without the bound five test
cases had scheduling times larger than 40 seconds. This is still slower, than
scheduling with gist simplified dependences but it yields better results.
The bounded, non-gist-simplified schedule yields for dynprog, lu, reg_detect
and trmm to faster code compared to scheduling with gist simplified
dependences. For gramschmidt it yields slower code.

Signed-off-by: Tobias Grosser <tobias@grosser.es>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_tab_basic_set_non_trivial_lexmin: do not add cuts for all variables
Tobias Grosser [Sat, 18 Feb 2012 15:13:23 +0000 (16:13 +0100)]
isl_tab_basic_set_non_trivial_lexmin: do not add cuts for all variables

We found a test case where adding cuts for all variables can increase
the time of the scheduling algorithm significantly (from ms to never
seen terminating). Even though we did not understand the problem
entirely, the intuition we have is the following:

When using isl_tab_basic_set_non_trivial_lexmin, the ILP problem is
(usually) not empty, but infinite. Adding all cuts, seems to increase
the objective function only in very small steps. Adding only one cut at
a time allows us to increase it with a step as big as possible.

This patch does not change the behaviour when calling
check_integer_feasible.  In case we're calling cut_to_integer_lexmin
from check_integer_feasible, if it takes a long time to compute, then
it's "usually" because the context is empty and it's just taking a long
time for us to discover that it's empty. We believe in that case it
could still make sense to add all cuts.

This patch was tested on polybench 2.0 with Polly and minimal fusion.
On the larger kernels, it could not solve any infinite scheduling time
problems, but it was able to speed up one test case by 2x and did not
yield increased scheduling time for the other test cases.

Signed-off-by: Tobias Grosser <tobias@grosser.es>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_map_fix_si: drop basic maps that have become empty
Sven Verdoolaege [Tue, 14 Feb 2012 23:18:59 +0000 (00:18 +0100)]
isl_map_fix_si: drop basic maps that have become empty

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_map.c: extract out common remove_if_empty
Sven Verdoolaege [Tue, 14 Feb 2012 23:18:21 +0000 (00:18 +0100)]
isl_map.c: extract out common remove_if_empty

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_map_coalesce: drop empty parts before checking the number of parts
Sven Verdoolaege [Tue, 14 Feb 2012 23:15:55 +0000 (00:15 +0100)]
isl_map_coalesce: drop empty parts before checking the number of parts

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoMerge branch 'maint'
Sven Verdoolaege [Tue, 14 Feb 2012 23:22:47 +0000 (00:22 +0100)]
Merge branch 'maint'

12 years agoisl_convex_hull.c: uset_convex_hull_wrap_bounded: remove redundant constraints
Sven Verdoolaege [Tue, 14 Feb 2012 22:55:22 +0000 (23:55 +0100)]
isl_convex_hull.c: uset_convex_hull_wrap_bounded: remove redundant constraints

The caller compute_facet expects the result to not contain any redundant
constraints.  The input to uset_convex_hull_wrap_bounded, however, might
contain redundant constraints so if this input consists of only one
basic set, we need to make sure redundant constraints are removed.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_basic_map_realign: preserve (most) flags
Sven Verdoolaege [Tue, 14 Feb 2012 19:02:43 +0000 (20:02 +0100)]
isl_basic_map_realign: preserve (most) flags

In particular, preserve the ISL_BASIC_MAP_EMPTY flag.
Many parts of the code check for this flags so they do not have
to deal with the "1 = 0" constraint.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoadd isl_printer_print_pw_aff test
Sven Verdoolaege [Tue, 7 Feb 2012 11:26:43 +0000 (12:26 +0100)]
add isl_printer_print_pw_aff test

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_basic_map_foreach_lexopt: normalize isl_affs
Sven Verdoolaege [Tue, 7 Feb 2012 11:26:20 +0000 (12:26 +0100)]
isl_basic_map_foreach_lexopt: normalize isl_affs

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_pw_*_union_add: don't subtract non-overlapping cells
Sven Verdoolaege [Tue, 7 Feb 2012 11:19:20 +0000 (12:19 +0100)]
isl_pw_*_union_add: don't subtract non-overlapping cells

There is no point in subtracting a cell from another if their intersection
is empty.  It can only lead to a further subdivision of the other cell.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_printer_print_pw_aff: skip constraints implied by div expression
Sven Verdoolaege [Tue, 7 Feb 2012 11:17:03 +0000 (12:17 +0100)]
isl_printer_print_pw_aff: skip constraints implied by div expression

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoadd isl_basic_set_is_div_constraint
Sven Verdoolaege [Tue, 7 Feb 2012 11:11:26 +0000 (12:11 +0100)]
add isl_basic_set_is_div_constraint

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoMerge branch 'maint'
Sven Verdoolaege [Tue, 7 Feb 2012 11:58:43 +0000 (12:58 +0100)]
Merge branch 'maint'

Conflicts:
isl_output.c

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_printer_print_pw_aff: fix printing in C format
Sven Verdoolaege [Tue, 7 Feb 2012 11:13:23 +0000 (12:13 +0100)]
isl_printer_print_pw_aff: fix printing in C format

In 3280c05 (make isl_pw_* object live in a map space,
Tue Aug 30 16:47:59 2011 +0200), the space in which objects live was
changed from that of its domain to that of a map, but
isl_printer_print_pw_aff was not changed accordingly.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_stream_read_pw_aff: call isl_pw_aff_union_add instead of isl_pw_aff_add
Sven Verdoolaege [Tue, 7 Feb 2012 11:09:03 +0000 (12:09 +0100)]
isl_stream_read_pw_aff: call isl_pw_aff_union_add instead of isl_pw_aff_add

We missed this in 7a86dd8 (rename isl_pw_aff_add to isl_pw_aff_union_add,
Thu Oct 13 12:16:45 2011 +0200).

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_vertices.c: compute_chambers: avoid access to freed data structure
Sven Verdoolaege [Tue, 7 Feb 2012 11:55:43 +0000 (12:55 +0100)]
isl_vertices.c: compute_chambers: avoid access to freed data structure

The problem was introduced in 328c78f (isl_tab_from_basic_map: preserve all
constraints in input when tracking, Mon Jan 16 16:55:44 2012 +0100).

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoexport isl_pw_multi_aff_coalesce
Sven Verdoolaege [Sun, 29 Jan 2012 15:52:15 +0000 (16:52 +0100)]
export isl_pw_multi_aff_coalesce

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoexport isl_pw_multi_aff_domain
Sven Verdoolaege [Sun, 29 Jan 2012 17:15:51 +0000 (18:15 +0100)]
export isl_pw_multi_aff_domain

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_aff_normalize: remove unused divs
Sven Verdoolaege [Mon, 30 Jan 2012 12:18:09 +0000 (13:18 +0100)]
isl_aff_normalize: remove unused divs

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_pw_multi_aff_from_map: add a special case where result can be read off
Sven Verdoolaege [Thu, 2 Feb 2012 11:01:53 +0000 (12:01 +0100)]
isl_pw_multi_aff_from_map: add a special case where result can be read off

If all output dimensions are uniquely defined in terms of the parameters
and input dimensions, we can directly read off the desired isl_pw_multi_aff.
This usually results in much simpler expressions.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoprivately export isl_basic_map_plain_is_single_valued
Sven Verdoolaege [Thu, 2 Feb 2012 09:29:43 +0000 (10:29 +0100)]
privately export isl_basic_map_plain_is_single_valued

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_printer_print_pw_aff: simplify domain expression in C format
Sven Verdoolaege [Thu, 2 Feb 2012 09:45:48 +0000 (10:45 +0100)]
isl_printer_print_pw_aff: simplify domain expression in C format

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_morph.c: fix typo in comment
Sven Verdoolaege [Wed, 1 Feb 2012 16:11:38 +0000 (17:11 +0100)]
isl_morph.c: fix typo in comment

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoMerge branch 'maint'
Sven Verdoolaege [Wed, 1 Feb 2012 21:50:00 +0000 (22:50 +0100)]
Merge branch 'maint'

12 years agoisl_basic_set_full_compression: detect equalities in input first
Sven Verdoolaege [Wed, 1 Feb 2012 17:01:31 +0000 (18:01 +0100)]
isl_basic_set_full_compression: detect equalities in input first

Without this explicit detection of equalities, some equalities
may be discovered in the middle of the computation, possibly
leading to the introduction of existentially quantified variables,
while callers of isl_basic_set_full_compression typically do not
want such variables.

Since the detection of equalities can in some cases be fairly expensive,
we may have to find a better way of dealing with this problem at some point.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_stream_read_pw_qpolynomial: accept products of integer divisions
Sven Verdoolaege [Wed, 1 Feb 2012 11:34:30 +0000 (12:34 +0100)]
isl_stream_read_pw_qpolynomial: accept products of integer divisions

Before, we called accept_affine to parse an integer division that appears
inside a function, but accept_affine also parses a subsequent '*' and
then expects this '*' to be followed by a constant.
Call accept_div directly intead.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_union_set_product: call isl_set_product on entries
Sven Verdoolaege [Tue, 31 Jan 2012 18:18:23 +0000 (19:18 +0100)]
isl_union_set_product: call isl_set_product on entries

Before, we would simply delegate control to isl_union_map_product,
which calls isl_map_product on the entries.  This doesn't work
on isl_sets because they only have a "range" and no "domain".

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_tab_pip.c: sol_add: skip empty contexts
Sven Verdoolaege [Tue, 31 Jan 2012 14:09:12 +0000 (15:09 +0100)]
isl_tab_pip.c: sol_add: skip empty contexts

If the context becomes non-empty inside find_solutions, we stop
looking for solutions since a89841a (isl_tab_pip.c: find_solutions: break
when context becomes empty, Sun Jan 31 18:29:12 2010 +0100), but
we still call sol_add.  We need to return from sol_add early in
such cases because there is nothing to do and, more importantly,
when a basic set is marked empty, its existentials are removed,
resulting in a mismatch in the number of dimensions.  In particular,
the basic set returned by sol_domain may not satisfy all assumptions
made by the remainder of sol_add.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_space_set_dim_id: also set id of parameter in nested spaces
Sven Verdoolaege [Tue, 31 Jan 2012 11:27:09 +0000 (12:27 +0100)]
isl_space_set_dim_id: also set id of parameter in nested spaces

A nested space is supposed to have the same parameters as
the nesting space, so if the id of a parameter in the nesting
space space changes, it has to be changed in the nested spaces too.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_space_set_dim_id: rename "dim" variable to "space"
Sven Verdoolaege [Tue, 31 Jan 2012 11:24:38 +0000 (12:24 +0100)]
isl_space_set_dim_id: rename "dim" variable to "space"

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoscheduler: replace split_parallel by split_scaled option
Sven Verdoolaege [Sun, 29 Jan 2012 13:42:16 +0000 (14:42 +0100)]
scheduler: replace split_parallel by split_scaled option

The split_parallel option was ill-conceived.
The condition split_parallel looked for was too specific
and the implementation was wrong.  Part of this has been fixed
in 1b2fdb6 (schedule.c: split_parallel: avoid invalid memory accesses,
Mon Jan 30 15:23:18 2012 +0100), but the code could in principle
still produce an incorrect schedule because it didn't check the sizes
of the constant terms.

The new option essentially applies strip-mining and should therefore
be safe.  In particular, it is applied when the linear parts of the
schedules have a non-trivial common divisor.  The strip-mining is
then performed with respect to this common divisor.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoadd isl_mat_scale_down_row
Sven Verdoolaege [Sun, 29 Jan 2012 13:42:01 +0000 (14:42 +0100)]
add isl_mat_scale_down_row

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoAdd isl_union_map_gist_range
Tobias Grosser [Mon, 30 Jan 2012 12:19:38 +0000 (13:19 +0100)]
Add isl_union_map_gist_range

Signed-off-by: Tobias Grosser <tobias@grosser.es>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoscheduler: add an option to configure the level of fusing
Sven Verdoolaege [Sat, 26 Nov 2011 14:52:07 +0000 (15:52 +0100)]
scheduler: add an option to configure the level of fusing

We currently only support maximal (the old behavior) and minimal fusion.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoMerge branch 'maint'
Sven Verdoolaege [Mon, 30 Jan 2012 14:28:04 +0000 (15:28 +0100)]
Merge branch 'maint'

12 years agoschedule.c: split_parallel: avoid invalid memory accesses
Sven Verdoolaege [Mon, 30 Jan 2012 14:23:18 +0000 (15:23 +0100)]
schedule.c: split_parallel: avoid invalid memory accesses

The original code incorrectly assumed that the schedules for
all statements have the same number of columns, i.e., that the
dimensions of the iteration domains of all statements are the same.

We fix this up by simply bailing out if we find this assumption not to hold.
We will improve the applicability of the transformation performed by
split_parallel in a subsequent patch.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoupdate for change in clang's HandleTopLevelDecl return type
Sven Verdoolaege [Wed, 18 Jan 2012 13:18:26 +0000 (14:18 +0100)]
update for change in clang's HandleTopLevelDecl return type

In recent versions of clang, HandleTopLevelDecl should return a boolean
indicating whether parsing should continue.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoconfigure.ac: only link against libraries from LLVM analysis component
Sven Verdoolaege [Wed, 18 Jan 2012 15:47:38 +0000 (16:47 +0100)]
configure.ac: only link against libraries from LLVM analysis component

Before, we would link against all LLVM components, which would make
us depend on google-test.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoput annotations after "struct"
Sven Verdoolaege [Wed, 18 Jan 2012 15:49:19 +0000 (16:49 +0100)]
put annotations after "struct"

Recent versions of clang require the annotations to be placed after "struct".

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoMerge branch 'maint'
Sven Verdoolaege [Wed, 18 Jan 2012 17:41:37 +0000 (18:41 +0100)]
Merge branch 'maint'

12 years agoisl_pw_aff_non_zero_set: don't assume isl_pw_aff is total
Sven Verdoolaege [Tue, 17 Jan 2012 12:52:09 +0000 (13:52 +0100)]
isl_pw_aff_non_zero_set: don't assume isl_pw_aff is total

Before, we would return the domain where the function is not zero
or undefined, whereas we should only return the domain where the function
is not zero.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_aff_nonneg_basic_set and isl_aff_zero_basic_set: simplify result
Sven Verdoolaege [Wed, 18 Jan 2012 17:36:53 +0000 (18:36 +0100)]
isl_aff_nonneg_basic_set and isl_aff_zero_basic_set: simplify result

The constructed constraint may be a tautology in which case it needs
to be simplified away, especially if it is an equality.

We probably want to perform this simplification in
isl_basic_set_from_constraint instead, but CLooG currently depends
on the resulting basic set not being simplified.
We would have to fix CLooG first before we can change this behavior.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_tab_from_basic_map: preserve all constraints in input when tracking
Sven Verdoolaege [Mon, 16 Jan 2012 15:55:44 +0000 (16:55 +0100)]
isl_tab_from_basic_map: preserve all constraints in input when tracking

If the user wants to keep track of a basic map representation of the tableau,
then we need to make sure that the initial tableau corresponds exactly to
the input.  In particular, we need to make sure that no (redundant) constraints
are removed.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_tab_track_bmap: drop sanity checks in case of empty tab
Sven Verdoolaege [Mon, 16 Jan 2012 15:46:55 +0000 (16:46 +0100)]
isl_tab_track_bmap: drop sanity checks in case of empty tab

If tab is empty, then we may not have added all constraints of "bmap"
and so the number of constraints in tab and bmap may be different.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_tab_track_bmap: make sure we have a unique copy of the basic map
Sven Verdoolaege [Mon, 16 Jan 2012 15:21:49 +0000 (16:21 +0100)]
isl_tab_track_bmap: make sure we have a unique copy of the basic map

The basic map will be modified in place by isl_tab_add_ineq.
Some callers were passing a unique copy to isl_tab_track_bmap
already, but some others apparently were not.
It's safer to just let isl_tab_track_bmap take care of this.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_tab_track_bmap: fix memory leak on error path
Sven Verdoolaege [Mon, 16 Jan 2012 11:59:33 +0000 (12:59 +0100)]
isl_tab_track_bmap: fix memory leak on error path

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_basic_map_remove_unknown_divs: restart whenever div has been removed
Sven Verdoolaege [Sun, 15 Jan 2012 15:12:49 +0000 (16:12 +0100)]
isl_basic_map_remove_unknown_divs: restart whenever div has been removed

The call to isl_basic_map_remove_dims may end up calling
isl_basic_map_simplify, which may in turn remove additional divs
if they no longer occur in any constraint.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_map_simple_hull: improve error handling
Sven Verdoolaege [Sun, 15 Jan 2012 14:21:17 +0000 (15:21 +0100)]
isl_map_simple_hull: improve error handling

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoexport isl_multi_aff_drop_dims
Sven Verdoolaege [Thu, 12 Jan 2012 14:49:38 +0000 (15:49 +0100)]
export isl_multi_aff_drop_dims

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_multi_aff_drop_dims: handle removal of output dimensions
Sven Verdoolaege [Thu, 12 Jan 2012 14:49:28 +0000 (15:49 +0100)]
isl_multi_aff_drop_dims: handle removal of output dimensions

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_multi_aff_set_aff: improve error handling
Sven Verdoolaege [Thu, 12 Jan 2012 14:48:27 +0000 (15:48 +0100)]
isl_multi_aff_set_aff: improve error handling

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoadd isl_multi_aff_set_tuple_id
Sven Verdoolaege [Thu, 12 Jan 2012 14:24:01 +0000 (15:24 +0100)]
add isl_multi_aff_set_tuple_id

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_pw_aff_on_shared_domain: perform gist on results wrt shared domains
Sven Verdoolaege [Thu, 12 Jan 2012 10:10:04 +0000 (11:10 +0100)]
isl_pw_aff_on_shared_domain: perform gist on results wrt shared domains

The shared domain may have extra constraints compared to the original
domains, which may help to simplify some of the integer divisions involved.
As a side effect, the gist operation may also simplify integer divisions
that could have been simplified even on a universe domain.
We should probably handle such cases much earlier during the construction
of isl_affs.

Reported-by: Tobias Grosser <tobias@grosser.es>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoadd isl_aff_project_domain_on_params
Sven Verdoolaege [Tue, 3 Jan 2012 09:35:43 +0000 (10:35 +0100)]
add isl_aff_project_domain_on_params

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoadd isl_aff_align_params
Sven Verdoolaege [Tue, 3 Jan 2012 09:42:33 +0000 (10:42 +0100)]
add isl_aff_align_params

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_printer_print_aff: only print () when needed in C format
Sven Verdoolaege [Tue, 3 Jan 2012 10:30:04 +0000 (11:30 +0100)]
isl_printer_print_aff: only print () when needed in C format

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_aff_gist: normalize result
Sven Verdoolaege [Tue, 3 Jan 2012 10:25:47 +0000 (11:25 +0100)]
isl_aff_gist: normalize result

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoMerge branch 'maint'
Sven Verdoolaege [Tue, 3 Jan 2012 10:43:51 +0000 (11:43 +0100)]
Merge branch 'maint'

12 years agoisl_schedule.c: mark row constructed by compute_split_schedule as non-zero
Sven Verdoolaege [Tue, 3 Jan 2012 09:05:50 +0000 (10:05 +0100)]
isl_schedule.c: mark row constructed by compute_split_schedule as non-zero

compute_split_schedule is only called when no zero-distance schedule row
could be found on the entire graph.
We missed this part in ee1ca56 (isl_schedule.c: keep track of parallel loops,
Sun May 29 19:41:55 2011 +0200).

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_map_project_out: properly handle named or nested spaces
Sven Verdoolaege [Mon, 2 Jan 2012 13:10:50 +0000 (14:10 +0100)]
isl_map_project_out: properly handle named or nested spaces

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_arg_parse: reset all flags with same option name whenever a flag is set
Sven Verdoolaege [Sun, 1 Jan 2012 17:24:58 +0000 (18:24 +0100)]
isl_arg_parse: reset all flags with same option name whenever a flag is set

Otherwise, there is no way to turn off a flag that is set by default.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_ctx_peek_options: special case isl_options_args
Sven Verdoolaege [Sun, 1 Jan 2012 17:00:49 +0000 (18:00 +0100)]
isl_ctx_peek_options: special case isl_options_args

The isl_options_args options are always available, even
when they do not appear in ctx->user_opt.
Just return them instead of looking through ctx->user_opt.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_basic_set_compute_vertices: set activity domain on single vertex
Sven Verdoolaege [Fri, 30 Dec 2011 23:19:47 +0000 (00:19 +0100)]
isl_basic_set_compute_vertices: set activity domain on single vertex

isl_basic_set_compute_vertices always sets the activity domains of
the vertices, except when there is only a single vertex.
Chnage vertices_0D to also set the activity domain in this case.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoMerge branch 'maint'
Sven Verdoolaege [Fri, 30 Dec 2011 23:28:50 +0000 (00:28 +0100)]
Merge branch 'maint'

12 years agoisl_morph_{dom,ran}_params: add missing isl_morph_cow
Sven Verdoolaege [Fri, 30 Dec 2011 23:18:11 +0000 (00:18 +0100)]
isl_morph_{dom,ran}_params: add missing isl_morph_cow

Usually, isl_morph_remove_{dom,ran}_dims will call isl_morph_cow for us,
but not when n == 0.

Reported-by: Tobias Grosser <tobias@grosser.es>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_access_info: allow user interaction to restrict potential sources
Sven Verdoolaege [Mon, 28 Nov 2011 14:28:37 +0000 (15:28 +0100)]
isl_access_info: allow user interaction to restrict potential sources

In particular, the user can register a callback that will be called
any time right before the last potential source is computed.
This should be useful for fuzzy array dataflow analysis.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_access_info_compute_flow: store sink domain map in isl_access_info
Sven Verdoolaege [Mon, 26 Dec 2011 15:07:48 +0000 (16:07 +0100)]
isl_access_info_compute_flow: store sink domain map in isl_access_info

We will be able to reuse this map in the next commit.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_access_info_alloc: use isl_calloc to allocate structure
Sven Verdoolaege [Mon, 26 Dec 2011 14:59:57 +0000 (15:59 +0100)]
isl_access_info_alloc: use isl_calloc to allocate structure

This makes it easier to add new (zero-by-default) fields.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_access_info_compute_flow: don't pass ownership of "acc" down
Sven Verdoolaege [Mon, 26 Dec 2011 14:28:46 +0000 (15:28 +0100)]
isl_access_info_compute_flow: don't pass ownership of "acc" down

This slightly simplifies the error handling.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl 0.09 isl-0.09
Sven Verdoolaege [Sat, 17 Dec 2011 17:19:52 +0000 (18:19 +0100)]
isl 0.09

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoavoid "errno" argument name
Sven Verdoolaege [Sun, 18 Dec 2011 08:41:15 +0000 (09:41 +0100)]
avoid "errno" argument name

On mingw, errno is defined as (*_errno()) in stdlib.h

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoadd isl_morph_dump
Sven Verdoolaege [Sat, 17 Dec 2011 09:14:59 +0000 (10:14 +0100)]
add isl_morph_dump

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agorename isl_morph_dump to isl_morph_print_internal
Sven Verdoolaege [Sat, 17 Dec 2011 09:13:43 +0000 (10:13 +0100)]
rename isl_morph_dump to isl_morph_print_internal

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_basic_set_compute_vertices: fix handling of lower-dimensional input
Sven Verdoolaege [Fri, 16 Dec 2011 08:20:34 +0000 (09:20 +0100)]
isl_basic_set_compute_vertices: fix handling of lower-dimensional input

Support was broken by 9043b59 (explicitly differentiate between spaces
of maps, sets and parameter sets, Thu Aug 25 19:21:05 2011 +0200).
This fix should have gone in before that commit.

Reported-by: Tobias Grosser <tobias@grosser.es>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agodoc: improve consistency in list of schedule options
Sven Verdoolaege [Wed, 14 Dec 2011 10:45:51 +0000 (11:45 +0100)]
doc: improve consistency in list of schedule options

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoadd Feautrier's scheduling algorithm
Sven van Haastregt [Wed, 14 Dec 2011 09:55:18 +0000 (10:55 +0100)]
add Feautrier's scheduling algorithm

Offer Feautrier's algorithm as an alternative to isl's default scheduling
algorithm.  The schedule is extended to a full-dimensional schedule.
To select the scheduling algorithm to use, we introduce isl option
schedule-algorithm.

Signed-off-by: Sven van Haastregt <svhaastr@liacs.nl>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoadd option to bound the constant scheduling coefficients
Tobias Grosser [Mon, 12 Dec 2011 16:07:18 +0000 (17:07 +0100)]
add option to bound the constant scheduling coefficients

If larger coefficients appear as part of the input dependences, the
schedule calculation can take a very long time. We observed that the
main overhead in this calculation is due to optimizing the constant
coefficients. They are misused to increase locality by merging several
unrelated dimensions into a single dimension. This unwanted optimization
increases the complexity of the code and slows down the generated code.

We introduce a new option that bounds the values in the constant
dimension by a user defined value. If the right value is choosen,
costly overoptimization can be prevented.

This solution works, but requires the user to specify the value by
which the constants are bound. For the moment, this is our best
solution, but we hope to to find a more generic one later on.

Signed-off-by: Tobias Grosser <tobias@grosser.es>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoIntroduce new ISL_CTX_(GET|SET)_INT_DEF macro
Tobias Grosser [Sun, 11 Dec 2011 23:21:55 +0000 (00:21 +0100)]
Introduce new ISL_CTX_(GET|SET)_INT_DEF macro

This macro uses the same implementation as the corresponding
BOOL and CHOICE macros. Hence, we simplify BOOL and CHOICE by
forwarding to the new INT macro.

Signed-off-by: Tobias Grosser <tobias@grosser.es>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoFix typo in documentation
Tobias Grosser [Sun, 11 Dec 2011 23:21:53 +0000 (00:21 +0100)]
Fix typo in documentation

Signed-off-by: Tobias Grosser <tobias@grosser.es>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoAdd interface to set schedule_maximize_band_depth
Tobias Grosser [Sun, 11 Dec 2011 23:21:52 +0000 (00:21 +0100)]
Add interface to set schedule_maximize_band_depth

Previously, schedule_maximize_band_depth could only be set with the
corresponding command line option. This patch adds two functions
to change this option when using isl as a library.

int isl_options_set_schedule_maximize_band_depth(
isl_ctx *ctx, int val);
int isl_options_get_schedule_maximize_band_depth(
isl_ctx *ctx);

Signed-off-by: Tobias Grosser <tobias@grosser.es>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agodoc/SubmittingPatches: fix typo in email address
Sven van Haastregt [Mon, 12 Dec 2011 15:28:32 +0000 (16:28 +0100)]
doc/SubmittingPatches: fix typo in email address

Signed-off-by: Sven van Haastregt <svhaastr@liacs.nl>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoexport schedule_split_parallel option
Sven van Haastregt [Mon, 12 Dec 2011 10:45:17 +0000 (11:45 +0100)]
export schedule_split_parallel option

Signed-off-by: Sven van Haastregt <svhaastr@liacs.nl>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoadd isl_pw_multi_aff_get_pw_aff
Sven Verdoolaege [Sun, 11 Dec 2011 12:12:29 +0000 (13:12 +0100)]
add isl_pw_multi_aff_get_pw_aff

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoMerge branch 'maint'
Sven Verdoolaege [Sat, 10 Dec 2011 18:43:38 +0000 (19:43 +0100)]
Merge branch 'maint'

12 years agoisl_map_from_pw_aff: handle NULL input
Sven Verdoolaege [Fri, 9 Dec 2011 18:51:06 +0000 (19:51 +0100)]
isl_map_from_pw_aff: handle NULL input

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoadd isl_space_map_from_domain_and_range
Sven Verdoolaege [Tue, 6 Dec 2011 08:17:03 +0000 (09:17 +0100)]
add isl_space_map_from_domain_and_range

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoIntroduce option 'on_error'
Tobias Grosser [Sat, 3 Dec 2011 15:52:35 +0000 (16:52 +0100)]
Introduce option 'on_error'

The option on_error defines the behaviour of isl in case of an error.
By default, the option is set to 'warn'. This is the current behaviour
and means isl prints a warning and continues the execution of the
program. We also introduce two new modes: 'continue' and 'abort'.
'continue' does not print any warning, but continues execution and 'abort'
abort the program from within isl.  'abort' is useful for debuggers as
they automatically break at the place of the error condition.

Signed-off-by: Tobias Grosser <tobias@grosser.es>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoexport isl_map_eliminate
Sven Verdoolaege [Thu, 1 Dec 2011 15:02:22 +0000 (16:02 +0100)]
export isl_map_eliminate

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoImprove message when ctx is freed, but still referenced
Tobias Grosser [Thu, 1 Dec 2011 14:15:45 +0000 (15:15 +0100)]
Improve message when ctx is freed, but still referenced

Instead of a cryptic 'Assertion "ctx->ref == 0" failed' we report
'isl_ctx freed, but some objects still reference it'.

Signed-off-by: Tobias Grosser <tobias@grosser.es>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoadd isl_local_space_range
Tobias Grosser [Thu, 1 Dec 2011 14:15:44 +0000 (15:15 +0100)]
add isl_local_space_range

Signed-off-by: Tobias Grosser <tobias@grosser.es>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoMerge branch 'maint'
Sven Verdoolaege [Fri, 2 Dec 2011 15:39:58 +0000 (16:39 +0100)]
Merge branch 'maint'

12 years agoisl_pw_aff_scale: don't special case multiplication with zero
Sven Verdoolaege [Thu, 1 Dec 2011 10:34:31 +0000 (11:34 +0100)]
isl_pw_aff_scale: don't special case multiplication with zero

isl_pw_*_scale returns an empty result if a multiplication with zero
is being performed.  We do this because zero is the default value
for polynomials.  Affine expressions, by contrast, have no default value
and so we should not return an empty result in this case.

Reported-by: Tomofumi Yuki <tomofumi.yuki@gmail.com>
Tested-by: Tomofumi Yuki <tomofumi.yuki@gmail.com>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoadd isl_set_find_dim_by_name
Sven Verdoolaege [Mon, 28 Nov 2011 15:45:16 +0000 (16:45 +0100)]
add isl_set_find_dim_by_name

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_basic_map_involves_dims: double-check divs
Sven Verdoolaege [Tue, 29 Nov 2011 09:17:46 +0000 (10:17 +0100)]
isl_basic_map_involves_dims: double-check divs

In principle, the extra test is not needed at this moment, but
this may change in future.

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoisl_schedule.c: fix typo in comment
Sven Verdoolaege [Sat, 26 Nov 2011 09:40:31 +0000 (10:40 +0100)]
isl_schedule.c: fix typo in comment

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoconfigure.ac: only create interface/Makefile if user has clang
Sven Verdoolaege [Fri, 25 Nov 2011 10:00:17 +0000 (11:00 +0100)]
configure.ac: only create interface/Makefile if user has clang

This allows users without clang to perform a "make dist".

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoadd isl_pw_aff_has_dim_id
Sven Verdoolaege [Mon, 21 Nov 2011 21:27:28 +0000 (22:27 +0100)]
add isl_pw_aff_has_dim_id

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoadd missing namespace qualification on llvm::ArrayRef
Sven Verdoolaege [Tue, 22 Nov 2011 22:28:39 +0000 (23:28 +0100)]
add missing namespace qualification on llvm::ArrayRef

Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
12 years agoMerge branch 'maint'
Sven Verdoolaege [Wed, 23 Nov 2011 09:33:26 +0000 (10:33 +0100)]
Merge branch 'maint'