On 10/22/15 05:55, Bernd Schmidt wrote:
On 10/22/2015 10:12 AM, Jakub Jelinek wrote:
So, is the worker broadcast buffer effectively a file scope .shared
variable? My worry is that as .shared is quite limited resource, if you
compile many TUs and each allocates its own broadcast buffer you run
On 10/22/15 08:59, Jakub Jelinek wrote:
On Thu, Oct 22, 2015 at 08:50:23AM -0400, Nathan Sidwell wrote:
+ expr = build2 (TRUNC_MOD_EXPR, ivar_type, ivar,
+fold_convert (ivar_type, collapse->iters));
+ expr = build2 (MULT_EXPR, diff_type, fold_convert (diff_t
On 10/22/15 07:10, Julian Brown wrote:
On Thu, 22 Oct 2015 10:05:30 +0200
Richard Biener wrote:
So you'd need to be more precise as to what properties you are trying
to preserve by placing a single stmt somewhere.
FWIW an earlier, abandoned attempt at solving the same problem was
discussed
On 10/22/15 04:15, Jakub Jelinek wrote:
On Wed, Oct 21, 2015 at 03:13:26PM -0400, Nathan Sidwell wrote:
+/* Determine whether fork & joins are needed. */
+
+static bool
+nvptx_xform_fork_join (gcall *call, const int dims[],
+ bool ARG_UNUSED (is_fork))
Why is this
On 10/22/15 09:01, Nathan Sidwell wrote:
On 10/22/15 05:55, Bernd Schmidt wrote:
On 10/22/2015 10:12 AM, Jakub Jelinek wrote:
So, is the worker broadcast buffer effectively a file scope .shared
variable? My worry is that as .shared is quite limited resource, if you
compile many TUs and each
On 10/22/15 09:17, Jakub Jelinek wrote:
On Thu, Oct 22, 2015 at 09:08:30AM -0400, Nathan Sidwell wrote:
I agree with Richard that it would be better to write more about what kind
of IL changes are acceptable with IFN_UNIQUE in the IL and what are not.
E.g. is inlining ok (I'd hope yes)
On 10/22/15 05:37, Jakub Jelinek wrote:
And, I must say I'm at least missing testcases that check parsing but also
runtime behavior of the vector or worker clause arguments (there
is one gang (static:1) clause, but not the other clauses nor other styles of
gang arguments.
the static clause is
On 10/22/15 04:07, Richard Biener wrote:
Yeah, please make them either end or start a BB so we have to check
at most a single stmt. ECF_RETURNS_TWICE should achieve that,
it also makes it a code motion barrier.
Just so I'm clear, you're not saying that RETURNS_TWICE will stop the call being
On 10/22/15 10:05, Jakub Jelinek wrote:
On Thu, Oct 22, 2015 at 09:53:46AM -0400, Nathan Sidwell wrote:
On 10/22/15 05:37, Jakub Jelinek wrote:
And, I must say I'm at least missing testcases that check parsing but also
runtime behavior of the vector or worker clause arguments (there
i
On 10/22/15 10:04, Bernd Schmidt wrote:
+ if (par->mask & GOMP_DIM_MASK (GOMP_DIM_MAX))
+{ /* No propagation needed for a call. */ }
+ else if (par->mask & GOMP_DIM_MASK (GOMP_DIM_WORKER))
Ok that looks weird with the open brace on the line before the else. I think the
standard practice
On 10/22/15 10:26, Richard Biener wrote:
On Thu, Oct 22, 2015 at 4:01 PM, Nathan Sidwell wrote:
RETURNS_TWICE will make the invariant motion stop at UNIQUE (HEAD),
but it would have done that anyway. It will also be a CSE barrier, thus
tem = global;
UNIQUE(HEAD)
tem2 = global;
will not
On 10/22/15 10:28, Bernd Schmidt wrote:
On 10/22/2015 04:24 PM, Nathan Sidwell wrote:
+ else
+{ /* Parent will skip this parallel itself. */ }
Here too - actually no need to have an empty else at all.
I wanted somewhere clear for the comment to go. (Actually, I think this
is the
On 10/22/15 10:30, Richard Biener wrote:
On Thu, Oct 22, 2015 at 3:24 PM, Nathan Sidwell wrote:
Essentially, yes. a set of IFN_UNIQUE form a group which must not be
separated from each other. The set is discovered implicitly by following
the CFG (though I suppose we could add an
On 10/22/15 10:32, Jakub Jelinek wrote:
There is a warning for
if (cond);
but not for
if (cond)
;
or
if (cond)
/* comment */ ;
which is the style used in various places throughout the compiler.
Sadly, that's not quite accurate. The warning occurs for all the empty if's you
On 10/22/15 10:47, Cesar Philippidis wrote:
Interesting question. The spec is unclear. It defines gang, worker and
vector as follows in section 2.7 in the OpenACC 2.0a spec:
gang [( gang-arg-list )]
worker [( [num:] int-expr )]
vector [( [length:] int-expr )]
where gang-arg is one of:
On 10/22/15 04:07, Richard Biener wrote:
Yeah, please make them either end or start a BB so we have to check
at most a single stmt. ECF_RETURNS_TWICE should achieve that,
it also makes it a code motion barrier.
I'm having a hard time making UNIQUE the end of a BB.
I'm emitting code to a gim
On 10/22/15 10:26, Richard Biener wrote:
On Thu, Oct 22, 2015 at 4:01 PM, Nathan Sidwell wrote:
On 10/22/15 04:07, Richard Biener wrote:
Yeah, please make them either end or start a BB so we have to check
at most a single stmt. ECF_RETURNS_TWICE should achieve that,
it also makes it a code
On 10/22/15 04:04, Jakub Jelinek wrote:
+ /* Ignore blocks containing non-clonable function calls. */
+ for (gsi = gsi_start_bb (CONST_CAST_BB (bb));
+ !gsi_end_p (gsi); gsi_next (&gsi))
+{
+ g = gsi_stmt (gsi);
+
+ if (is_gimple_call (g) && gimple_call_internal_p (g)
+
On 10/23/15 04:25, Jakub Jelinek wrote:
On Thu, Oct 22, 2015 at 02:06:54PM -0400, Nathan Sidwell wrote:
On 10/22/15 10:26, Richard Biener wrote:
On Thu, Oct 22, 2015 at 4:01 PM, Nathan Sidwell wrote:
On 10/22/15 04:07, Richard Biener wrote:
Yeah, please make them either end or start a BB
On 10/23/15 09:03, Richard Biener wrote:
It's a hack. I don't like hacks.
One person's hack can be another person's pragmatism :)
I think the requirement "don't duplicate me"
but inlining is ok is somewhat broken.
The requirement is that the SESE region formed by the markers remains as a
On 10/23/15 09:03, Jakub Jelinek wrote:
On Fri, Oct 23, 2015 at 08:57:17AM -0400, Nathan Sidwell wrote:
If this is during the omplower pass, then it is before cfg pass and
therefore all you need is tweak the gimple_call_initialize_ctrl_altering
function and the cfg pass will DTRT.
ok
On 10/23/15 09:16, Jakub Jelinek wrote:
On Fri, Oct 23, 2015 at 09:13:43AM -0400, Nathan Sidwell wrote:
You're correct that the SESE region could be split across a function
boundary in the manner you describe, but the complexity of dealing with
that in the backend's partitioning cod
On 10/23/15 16:17, Cesar Philippidis wrote:
Nathan, can you try out this patch with your updated patch set? I saw
some test cases getting stuck when expanding expand_GOACC_DIM_SIZE in on
the host compiler, which is wrong. I don't see that happening in
gomp-4_0-branch with this patch. Also, can y
3) IFN_UNIQUE is no longer ECF_LEAF
4) Inserted a data dependency chain to the had & tail call sequence. The 2nd
param is the result of the previous call in the chain.
Preparing updated trunk patches now ...
nathan
2015-10-25 Nathan Sidwell
* internal-fn.def (IFN_UNIQUE): N
On 10/23/15 17:25, Nathan Sidwell wrote:
On 10/23/15 16:17, Cesar Philippidis wrote:
Nathan, can you try out this patch with your updated patch set? I saw
some test cases getting stuck when expanding expand_GOACC_DIM_SIZE in on
the host compiler, which is wrong. I don't see that happeni
h shortly.
comments?
nathan
2015-10-25 Nathan Sidwell
* internal-fn.c (expand_UNIQUE): New.
* internal-fn.h (enum ifn_unique_kind): New.
* internal-fn.def (IFN_UNIQUE): New.
* gimple.h (gimple_call_internal_unique_p): New.
* gimple.c (gimple_call_same_target_p): Check internal fn
uni
mply does a src->dst move, which will eventually be deleted as
unnecessary.
comments?
nathan
2015-10-25 Nathan Sidwell
* internal-fn.def (IFN_GOACC_LOOP): New.
* internal-fn.h (enum ifn_unique_kind): Add IFN_UNIQUE_OACC_FORK,
IFN_UNIQUE_OACC_JOIN, IFN_UNIQUE_OACC_HEAD_MARK,
IFN_UNIQUE
riable 'axis'
[-Werror=unused-variable]
rtx axis = expand_normal (gimple_call_arg (stmt, 2));
Fixed thusly.
nathan
2015-10-26 Nathan Sidwell
* internal-fn.c (expand_UNIQUE): Protect fork & join inside
co
On 10/26/15 07:36, Richard Biener wrote:
+ {
#ifdef HAVE_oacc_fork
(etc.) can you use target-insn.def and targetm.have_oacc_fork () instead?
I can try ...
nathan
On 10/26/15 08:13, Jakub Jelinek wrote:
It won't convert them into such representations.
Can you fix that incrementally? I'd expect that code marked with acc loop
vector
can't have loop carried backward lexical dependencies, at least not within
the adjacent number of iterations specified in
On 10/26/15 09:51, Jakub Jelinek wrote:
If not, I think the only thing remaining is the IFN_UNIQUE patch, which
(At least) needs an update to use targetm.have... conversion.
Ok, will wait till you make those changes then?
Hope to have that later today.
nathan
On 10/26/15 07:36, Richard Biener wrote:
Looks better now.
+ {
#ifdef HAVE_oacc_fork
(etc.) can you use target-insn.def and targetm.have_oacc_fork () instead?
I've committed this to gomp4. Will port and update the patches for trunk.
nathan
2015-10-26 Nathan Si
se a difficulty with inlining, we simply identify them with an offset
determined when generating the lowering. This offset is useful in our case for
worker level reductions, but could be used for other purposes by different backends.
nathan
2015-10-26 Nathan Sidwell
* doc/tm.texi: Re
parate function. That's fine -- it'll copy
the whole CFG of interest.
ok?
nathan
2015-10-26 Nathan Sidwell
* internal-fn.c (expand_UNIQUE): New.
* internal-fn.h (enum ifn_unique_kind): New.
* internal-fn.def (IFN_UNIQUE): New.
* target-insns.def (unique): Define.
* gimple.h (gimp
e PTX case simply does a src->dst move, which will eventually be deleted as
unnecessary.
ok?
nathan
2015-10-26 Nathan Sidwell
* internal-fn.def (IFN_GOACC_LOOP): New.
* internal-fn.h (enum ifn_unique_kind): Add IFN_UNIQUE_OACC_FORK,
IFN_UNIQUE_OACC_JOIN, IFN_UNIQUE_O
On 10/27/15 01:18, Jakub Jelinek wrote:
LGTM, though could I ask you to try to try to move the
struct oacc_collapse
expand_oacc_collapse_init
expand_oacc_collapse_vars
expand_oacc_for
additions somewhere else
(e.g. in between expand_omp_taskreg and expand_omp_for_init_counts),
ok, I wasn't su
On 10/27/15 06:45, Richard Biener wrote:
On Tue, Oct 27, 2015 at 9:03 AM, Jakub Jelinek wrote:
Ok for me, but please wait for Richi's ack too.
+ /* An IFN_UNIQUE call must be duplicated as part of its group,
+or not at all. */
+ if (is_gimple_call (g) && gimple_call_inter
On 10/27/15 01:03, Jakub Jelinek wrote:
On Mon, Oct 26, 2015 at 03:32:45PM -0700, Nathan Sidwell wrote:
to break out the else part into a separate function. That's fine -- it'll
copy the whole CFG of interest.
The question is if some UNIQUE call could be ever considered as p
This is the patch that was committed.
nathan
2015-10-27 Nathan Sidwell
* internal-fn.c (expand_UNIQUE): New.
* internal-fn.h (enum ifn_unique_kind): New.
* internal-fn.def (IFN_UNIQUE): New.
* target-insns.def (unique): Define.
* gimple.h (gimple_call_internal_unique_p): New
pidis
Thomas Schwinge
James Norris
Joseph Myers
Julian Brown
Nathan Sidwell
Bernd Schmidt
gcc/cp/
* parser.c (cp_parser_omp_clause_name): Add auto, gang, seq,
vector, worker.
(cp_parser_oacc_simple_clause): New.
(cp_parser_oacc_shape_clause)
athan
2015-10-27 Cesar Philippidis
Thomas Schwinge
James Norris
Joseph Myers
Julian Brown
Nathan Sidwell
Bernd Schmidt
gcc/cp/
* parser.c (cp_parser_omp_clause_name): Add auto, gang, seq,
vector, worker.
(cp_parser_oacc_simple_cl
On 10/22/15 02:32, Jakub Jelinek wrote:
On Wed, Oct 21, 2015 at 03:50:31PM -0400, Nathan Sidwell wrote:
This patch is obvious, but included for completeness. We always want to run
the device lowering pass (when openacc is enabled), in order to delete the
marker and loop functions that should
On 10/22/15 01:15, Jakub Jelinek wrote:
On Wed, Oct 21, 2015 at 03:13:26PM -0400, Nathan Sidwell wrote:
2015-10-20 Nathan Sidwell
* target.def (fork_join): New GOACC hook.
* targhooks.h (default_goacc_fork_join): Declare.
* omp-low.c (default_goacc_forkjoin): New
On 10/22/15 01:58, Jakub Jelinek wrote:
On Wed, Oct 21, 2015 at 03:24:13PM -0400, Nathan Sidwell wrote:
2015-10-20 Nathan Sidwell
* omp-low.c (oacc_init_rediction_array): New.
(oacc_initialize_reduction_data): Initialize array.
Ok.
Committed.
On 10/26/15 08:13, Jakub Jelinek wrote:
On Wed, Oct 21, 2015 at 03:49:08PM -0400, Nathan Sidwell wrote:
This patch is the device-specific half of the previous patch. It processes
the partition head & tail markers and loop abstraction functions inserted
during omp lowering.
This is
On 10/27/15 07:02, Nathan Sidwell wrote:
yeah, I noticed diff got confused. (I'm not sure the above suggestion will
resolve it, but we can give it a go.
This is what I've committed. This breaks the libgomp reduction tests on nvidia,
because there's now a discrepancy between
This is the patch I've just committed.
It includes the new target hook overriding, which was originally in patch 3.
nathan
2015-10-28 Nathan Sidwell
* config/nvptx/nvptx.h (struct machine_function): Add
axis_predicate.
* config/nvptx/nvptx-protos.h (nvptx_expand_oacc
On 10/21/15 13:16, Nathan Sidwell wrote:
On 10/21/15 16:14, Ilya Verbin wrote:
<11-trunk-tests.patch>
Does the testcase with offload IR appear here accidentally?
D'oh! yup, fixed.
Now all applied, Thanks for everybody's help.
nathan
I've committed this to trunk, cleaning up a few bits and bobs I noticed
comparing with the gomp4 branch.
nathan
2015-10-28 Nathan Sidwell
* config/nvptx/nvptx.c (nvptx_print_operand): Remove 'd' case.
(struct parallel): Update comment.
(nvptx_reorg): Likewise.
(nvptx_n
I've committed this as obvious. There were a couple of bits of now-redundant
code I'd missed in my recent merge to trunk.
nathan
2015-10-28 Nathan Sidwell
* omp-low.c (MASK_GANG, MASK_WORKER, MASK_VECTOR): Delete.
(extract_omp_for_data): Remove OpenACC special handling of
e POS must be zero.
ok?
nathan
2015-10-28 Nathan Sidwell
* gimple-fold.c: Include omp-low.h.
(fold_internal_goacc_dim): New.
(gimple_fold_call): Call it.
Index: gcc/gimple-fold.c
===
--- gcc/gimple-fold.c (revision 229488)
++
I've committed this to gomp4 branch. It reduces the delta from trunk by
removing unneeded code and porting some changes that occurred on the route to my
recent merge of code to trunk.
nathan
2015-10-28 Nathan Sidwell
* target.def (goacc.fork_join): Invert sense of result.
* doc/tm
:
if (offloaded || data_region)
{ A; B; }
If that's incorrect, is the first '|| data_region' wrong?
nathan
2015-10-28 Nathan Sidwell
* omp-low.c (lower_omp_target): Remove unreachable code & merge
ifs
g part of
openacc.h in the test directory, and this had diverged from the openacc.h we
actually have. I deleted those tests and inserted one in the libgomp testsuite,
which correctly picks up the openacc.h of the tool under test, (rather than one
in system includes).
nathan
2015-10-29 Nathan
I've committed this to gomp4. It removes a no-longer needed field from
omp_context & simplifies the dummy head/tail generation needed for reductions
at the outermost level. Also incorporates the simplification I committed to
trunk earlier today.
nathan
2015-10-29 Natha
lementation anyway. I added enumeration values to acc_device_t
to enforce that, just in case someone tries to compile their openacc code with
-fshort-enums.
Committed to trunk.
nathan
2015-10-29 Nathan Sidwell
gcc/
* openacc.h (enum acc_device_t): Reformat. Ensure layout
compatibili
t have
to deal with version skew between libgomp and its plugins, this doesn't require
versioning changes etc.
Jakub, not sure if this patch falls under your review or not. Given ptx is the
only current plugin, and this doesn't break backwards compatibility I figure it ok.
nat
This bit of trunk code in cgraph_node::create at around line 500 of cgraph.c
looks wrong. Specifically the contents of the #ifdef -- it's uncompilable as
there's no 'g'.
if ((flag_openacc || flag_openmp)
&& lookup_attribute ("omp declare target", DECL_ATTRIBUTES (decl)))
{
no
On 10/30/15 13:54, Jeff Law wrote:
On 10/30/2015 02:52 PM, Nathan Sidwell wrote:
This bit of trunk code in cgraph_node::create at around line 500 of
cgraph.c looks wrong. Specifically the contents of the #ifdef -- it's
uncompilable as there's no 'g'.
if ((flag_op
.
nathan
2015-10-30 Jeff Law
Nathan Sidwell
* config/nvptx/nvptx.h (HARD_REGNO_NREGS): Avoid warning on unused
args.
(MOVE_MAX): Set to 8.
Index: config/nvptx/nvptx.h
===
--- config/nvptx/nvptx.h (revision 229595)
+++
On 10/30/15 15:16, Nathan Sidwell wrote:
Testing attached ...
... with parens on void cast args ...
nathan
On 10/30/15 15:30, Jeff Law wrote:
On 10/30/2015 04:16 PM, Nathan Sidwell wrote:
Cool. I'll go ahead and add nvptx-elf to the list shortly on the assumption
this stuff will get fixed on way or another.
I've just committed nvptx-none
Index:
On 10/28/15 14:40, Nathan Sidwell wrote:
Richard,
this patch adds folding for the new GOACC_DIM_POS and GOACC_DIM_SIZE internal
functions. IIUC gimple_fold_call is the right place to add this.
The size of a compute dimension is very often a compile-time constant. On the
host, in particular
Jakub,
the following patch series implements the reduction handling for OpenACC:
01-trunk-reductions-core-1102.patch Core execution changes
02-trunk-reductions-ptx-1102.patch PTX backend bits
03-trunk-reductions-tests-1102.patch Testcases
The reduction mechanism relies on a new internal bui
tion we can do there following the openmp merge. This only
affected one testcase (which was (a) broken anyway and (b) working by accident).
nathan
2015-11-02 Nathan Sidwell
Cesar Philippidis
* internal-fn.def (GOACC_REDUCTION): New.
* internal-fn.h (enum ifn_goacc_reduction_kind): N
to 32 in this manner, which is what
other compilers appear to do anyway.
Bernd, I think the builtin fn bits you originally wrote got completely rewritten
by me several times.
nathan
2015-11-02 Nathan Sidwell
Cesar Philippidis
* config/nvptx/nvptx.c: Include gimple headers.
reference variable. Reference handling will be a later
patch.
nathan
2015-11-02 Nathan Sidwell
libgomp/
* libgomp.oacc-c-c++-common/loop-red-g-1.c: New.
* libgomp.oacc-c-c++-common/loop-red-gwv-1.c: New.
* libgomp.oacc-c-c++-common/loop-red-v-1.c: New.
* libgomp.oacc-c-c++-common/loop-red-v-2
The following pair of patches add support for routines.
01-trunk-routine-code-1102.patch: Compiler changes
02-trunk-routine-tests-1102.patch: Tests
The changes are pretty straight forwards, as the execution model patch set laid
the groundwork. Routines are decorated with the 'oacc routine' pr
sets
TREE_PURPOSE of such dimensions to zero. while those dimensions over which it
must not partition a loop have TREE_PURPOSE set to non-zero. the handling for
validating this in the lower_oacc_device pass, and in the PTX backend is already
there.
ok?
nathan
2015-11-02 Nat
ests.
ok?
nathan
2015-11-02 Nathan Sidwell
gcc/testsuite/
* c-c++-common/goacc/routine-1.c: New.
* c-c++-common/goacc/routine-2.c: New.
* c-c++-common/goacc/routine-3.c: New.
* c-c++-common/goacc/routine-4.c: New.
libgomp/
* testsuite/libgomp.oacc-c-c++-common/routine-g-1.c: New.
* tests
On 11/02/15 14:41, Jakub Jelinek wrote:
Does this work even with -O0? I mean, the assembler is invalid
for any target other than PTX, so you are relying on aggressively folding
this away.
Correct. As thread identification is inherently target-specific, I don't see
how to do otherwise.
We
On 11/03/15 10:35, Jakub Jelinek wrote:
On Mon, Nov 02, 2015 at 02:21:43PM -0500, Nathan Sidwell wrote:
--- gcc/c/c-parser.c(revision 229667)
+++ gcc/c/c-parser.c(working copy)
@@ -1160,7 +1160,8 @@ enum c_parser_prec {
static void c_parser_external_declaration (c_parser *);
static
On 11/03/15 10:46, Jakub Jelinek wrote:
On Mon, Nov 02, 2015 at 11:18:37AM -0500, Nathan Sidwell wrote:
This is the core execution bits of OpenACC reductions.
We have a new internal fn 'IFN_GOACC_REDUCTION' and a new target hook
goacc.reduction, to lower it on the target compiler.
On 11/03/15 10:38, Jakub Jelinek wrote:
On Mon, Nov 02, 2015 at 02:23:19PM -0500, Nathan Sidwell wrote:
Here are the tests for the routine support. The compiler tests check
invalid combinations of gang, worker, vector & seq. The libgomp execution
tests check the expected partioning oc
an
2015-11-03 Nathan Sidwell
* target.def (goacc.dim_limit): New hook.
* targhooks.h (default_goacc_dim_limit): Declare.
* doc/tm.texi.in (TARGET_GOACC_DIM_LIMIT): Add.
* doc/tm.texi: Rebuilt.
* omp-low.c (default_goacc_dim_limit): New.
* config/nvptx/nvptx.c (PTX_VECTOR_LENGTH, PTX_WOR
On 11/04/15 05:26, Richard Biener wrote:
On Tue, Nov 3, 2015 at 7:11 PM, Nathan Sidwell wrote:
Richard,
this patch implements VRP for the 2 openacc axis internal fns I've added.
We know the position within a dimension cannot exceed that dimensions
extend. Further, if the extend is dy
On 11/04/15 05:01, Jakub Jelinek wrote:
On Mon, Nov 02, 2015 at 11:35:34AM -0500, Nathan Sidwell wrote:
2015-11-02 Nathan Sidwell
Cesar Philippidis
* config/nvptx/nvptx.c: Include gimple headers.
(worker_red_size, worker_red_align, worker_red_name,
...
I
On 11/04/15 05:31, Jakub Jelinek wrote:
On Tue, Nov 03, 2015 at 11:01:57AM -0500, Nathan Sidwell wrote:
On 11/03/15 10:46, Jakub Jelinek wrote:
On Mon, Nov 02, 2015 at 11:18:37AM -0500, Nathan Sidwell wrote:
This is the core execution bits of OpenACC reductions.
We have a new internal fn
On 11/04/15 08:27, Bernd Schmidt wrote:
On 11/02/2015 05:35 PM, Nathan Sidwell wrote:
There are two such switch statements, and it's possible to write this more
compactly:
if (!INTEGRAL_MODE_P (...))
code = VIEW_CONVERT_EXPR;
if (GET_MODE_SIZE (...) == 8)
fn = CMP_S
On 11/04/15 08:27, Bernd Schmidt wrote:
Adjust and applied, thanks!
nathan
2015-11-04 Nathan Sidwell
Cesar Philippidis
* config/nvptx/nvptx.c: Include gimple headers.
(worker_red_size, worker_red_align, worker_red_name,
worker_red_sym): New.
(nvptx_option_override): Initialize
On 11/04/15 05:26, Richard Biener wrote:
On Tue, Nov 3, 2015 at 7:11 PM, Nathan Sidwell wrote:
Richard,
this all seems a little bit fragile and relying on implementation details?
Is the attribute always present? Is the call argument always a constant
that fits in a HOST_WIDE_INT (or even
).
Also a new testcase to exercise the error handling.
Committed to trunk.
nathan
2015-11-04 Nathan Sidwell
gcc/
* config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Add checking.
libgomp/
* testsuite/libgomp.oacc-fortran/reduction-1.f90: Fix dimensions
and reduction copy.
* testsui
I've committed this patch to trunk as obvious. I'd missed some code obsoleted
by the new reduction implementation. This code was simply creating a splay
tree, not adding anything to it and then processing the splay tree. Nuked from
orbit.
nathan
2015-11-04 Nathan Sidwell
*
On 11/05/15 10:29, Jakub Jelinek wrote:
Hi!
I've merged the current state of gomp-4_5-branch into trunk, after
bootstrapping/regtesting it on x86_64-linux and i686-linux.
There are
+FAIL: gfortran.dg/goacc/private-3.f95 -O (test for excess errors)
+FAIL: libgomp.oacc-c/../libgomp.oacc-c-c++-
On 11/04/15 05:02, Bernd Schmidt wrote:
On 11/02/2015 02:56 PM, Nathan Sidwell wrote:
On 10/28/15 14:40, Nathan Sidwell wrote:
Richard,
this patch adds folding for the new GOACC_DIM_POS and GOACC_DIM_SIZE
internal
functions. IIUC gimple_fold_call is the right place to add this.
The size of a
e allowed to be specified with the kernels
construct for what it's worth, but that doesn't mean we couldn't use it
internally, of course, if so required.)
I think interpretation #2 or #3 make sense, and I suspect result in the same
emitted code.
nathan
--
Nathan Sidwell
#2 (a smaller patch) may not make it today, as I
have to rebase it on the reworking of #1 I did to remove the two enums I
disucssed with Jakub.
nathan
--
Nathan Sidwell
n removed yesterday, feel free to re-add it,
but of course never set it for OpenMP, just OpenACC constructs
(so for OpenMP keep the gimplifier assertion, for OpenACC set it).
FWIW not noticed a problem with my firstprivate reworking, rebased ontop
yesterday's openmp merge
nathan
--
Nathan Sidwell
s fixes a few xfails.
I'll post the default handling patch, which is much more localized.
nathan
2015-11-06 Nathan Sidwell
Cesar Philippidis
gcc/
* gcc/gimplify.c (enum omp_region_type): Add ORT_ACC,
ORT_ACC_DATA, ORT_ACC_PARALLEL, ORT_ACC_KERNELS. Adjust ORT_NONE.
(new_omp_
On 11/09/15 08:46, Jakub Jelinek wrote:
On Sat, Nov 07, 2015 at 08:50:28AM -0500, Nathan Sidwell wrote:
Index: gcc/gimplify.c
===
If you want to switch to hexadecimal, you should change all values
in the enum to hexadecimal for
On 11/09/15 08:59, Nathan Sidwell wrote:
On 11/09/15 08:46, Jakub Jelinek wrote:
On Sat, Nov 07, 2015 at 08:50:28AM -0500, Nathan Sidwell wrote:
Say
int arr[64];
// initialize arr
#pragma acc data copyin (arr)
{
// modify arr on the host
# pragma acc parallel firstprivate (arr
On 11/09/15 09:10, Jakub Jelinek wrote:
On Mon, Nov 09, 2015 at 08:59:15AM -0500, Nathan Sidwell wrote:
This I'm afraid performs often two copies rather than just one (one to copy
the host value to the present_copyin mapped value, another one in the
region),
I don't think that can
I've committed this to gomp4, the relevant handling is in gimple-fold now.
nathan
2015-11-09 Nathan Sidwell
* omp-low.c (oacc_xform_dim): Delete.
(execute_oacc_device_lower): Remove IFN_GOACC_DIM_POS,
IFN_GOACC_DIM_SIZE handling.
Index: omp-
I've committed this to trunk. It nuke the now unused GOACC_GET_NUM_THREADS and
GOACC_GET_THREAD_NUM calls. Also fixed up some comment typos I noticed
nathan
2015-11-09 Nathan Sidwell
* omp-low.c: Fix some OpenACC comment typos.
(lower_reduction_clauses): R
On 11/03/15 10:35, Jakub Jelinek wrote:
On Mon, Nov 02, 2015 at 02:21:43PM -0500, Nathan Sidwell wrote:
--- gcc/c/c-parser.c(revision 229667)
+++ gcc/c/c-parser.c(working copy)
@@ -1160,7 +1160,8 @@ enum c_parser_prec {
static void c_parser_external_declaration (c_parser *);
static
And these are the new tests. Cesar, c-c++-common/goacc/routine-5.c will need
adjusting with your C++ parser patch. You'll see the two cases I've #if'd out.
nathan
2015-11-09 Nathan Sidwell
gcc/testsuite/
* c-c++-common/goacc/routine-1.c: New.
* c-c++-common/goacc/routine
I've removed a testcase I inadvertently committed along with the routine
testcases:
2015-11-09 Nathan Sidwell
* testsuite/libgomp.oacc-c-c++-common/firstprivate-1.c: Remove
inadvertent commit.
nathan
On 11/09/15 09:46, Nathan Sidwell wrote:
I'm going to try and get clarification, but I think the intent is to initialize
with the value seen on the device. Consider:
My thinking is that the intent of the firstprivate is to initialize with the
value known on the device (and behave
I've committed this to backport a bunch of random bits from trunk to gomp4, and
thereby reduce divergence.
nathan
2015-11-10 Nathan Sidwell
* omp-low.c: Backport whitespace & formatting changes from trunk.
(is_atomic_compatible_p): Delete.
(lower_reduction_clauses): Don&
I've committed this to gomp4. In preparing the reworked firstprivate patch
changes for gomp4's gimplify.c I discovered these testcases were passing by
accident, and lacked a data clause.
nathan
2015-11-10 Nathan Sidwell
* testsuite/libgomp.oacc-fortran/parallel-reduction.f90
501 - 600 of 2551 matches
Mail list logo
