On Wed, 20 May 2020 16:52:02 +0200
Thomas Schwinge wrote:
> Hi!
>
> On 2019-12-17T22:03:47-0800, Julian Brown
> wrote:
> > --- a/libgomp/oacc-mem.c
> > +++ b/libgomp/oacc-mem.c
>
> > static int
> > -find_group_last (int pos, size_t mapnum, unsigned
On Fri, 8 May 2020 16:18:34 +0200
Thomas Schwinge wrote:
> >> Can you please explain (textually?) how this checking (design per
> >> your textual description below) is working in context of mixed
> >> OpenACC structured ("S") and dynamic ("D") reference counts? For
> >> example:
> >>
> >>
#pragma acc exit data delete ([data]) // no-op; S: 1, D: 0
> } // copyout; S: 0, D: 0
>
> assert (!acc_is_present ([data]));
>
> (Haven't compiled but I'm reasonably sure that the nesting and my
> manual "[action]; [S], [D]" annotations are corre
On Mon, 24 Feb 2020 15:03:21 +0300 (MSK)
Alexander Monakov wrote:
> On Mon, 24 Feb 2020, Andreas Schwab wrote:
>
> > On Feb 24 2020, Petr Tesarik wrote:
> >
> > > On Mon, 24 Feb 2020 12:29:40 +0100
> > > Andreas Schwab wrote:
> > >
> > >> On Feb 24 2020, Petr Tesarik wrote:
> > >>
> >
lable for the time being, so won't be
able to use his rubber-stamp powers. I added the offending libgomp code
to start with though, so I think I can go ahead and commit the patch.
I'll hold off for 24 hours though in case there are any objections
(Jakub?).
Thanks,
Julian
> On 1/10/20 2:49 A
This patch adds a new function to logically decrement the "dynamic
reference counter" for a mapped OpenACC variable, and handles some cases
in which that counter drops to zero inside a structured data
block. Previously, it's likely that at least in some cases, ending a
dynamic data lifetime in
es. Tested with
offloading to NVPTX, also with a version of my refcount-verification patch
(not currently on trunk).
I believe this (at least the first two parts) fixes a regression (for
the pr92843-1.c test case), so OK for stage 4?
Thanks,
Julian
Julian Brown (3):
Introduce dynamic data mappin
This patch prevents "exit data" directives from copying back data that
was mapped with an acc_map_data API call. This matches the behaviour
expected by the pr92843-1.c test, and together with the previous patch
in this series, allows that test to pass (with no other regressions).
Tested alongside
This patch provides a way to distinguish target_mem_descs that arise from
"enter data" operations from those that arise from structured OpenACC
data blocks. In that way, we can implement the equivalent of the "no-op"
behaviour of decrementing a dynamic reference count that is already zero
for some
On Tue, 7 Jan 2020 11:21:38 +0100
Tobias Burnus wrote:
> Hi Julian,
>
> please hold back the commit.
>
> Actually, it does not seem to work if one modifies the test
> case a bit. The following code compiles – but I think it
> should not:
>
> type one
>integer i, j
> end type
> type two
>
test.
libgomp/
* oacc-mem.c (goacc_exit_data_internal): Remove special (no-copyback)
behaviour for GOMP_MAP_STRUCT.
commit 5e9d8846bbaa33a9bdb08adcf1ee9f224a8e8fc0
Author: Julian Brown
Date: Wed Jan 8 15:57:46 2020 -0800
Fix component mappings with derived types for Ope
-04 Julian Brown
PR fortran/93025
gcc/fortran/
* openmp.c (resolve_omp_clauses): Check array references for contiguity.
gcc/testsuite/
* gfortran.dg/goacc/mapping-tests-2.f90: New test.
* gfortran.dg/goacc/subarrays.f95: Expect rejection of non
Hi,
This test (by Tobias Burnus, mildly edited) adds a test to check whether
the final component of a derived-type access has pointer or allocatable
type for manual deep copy attach/detach operations. This is just checking
existing behaviour.
This arose from discussion of the manual deep copy
patch adds the same restriction to the Fortran front-end,
and adjusts some existing tests to match. A test is also added for
detection of duplicate derived-type member accesses within a directive.
Tested with offloading to NVPTX. OK for mainline?
Thanks,
Julian
ChangeLog
2020-01-04 Ju
Hi,
On Sun, 22 Dec 2019 00:01:10 +0100
Thomas Schwinge wrote:
> I had intentionally left out this assignment in my "In
> 'libgomp/target.c', 'struct splay_tree_key_s', use 'struct
> splay_tree_aux' for infrequently-used or API-specific data" patch,
>
Hi Maciej,
On Sun, 22 Dec 2019 00:43:54 + (GMT)
"Maciej W. Rozycki" wrote:
> On Fri, 20 Dec 2019, Mike Stump wrote:
>
> > >> This patch series addresses a problem with the testsuite
> > >> compiler being set up across libatomic, libffi, libgo, libgomp
> > >> with no correlation whatsoever
p.oacc-c-c++-common/deep-copy-10.c: Fix
unsafe async usage.
* testsuite/libgomp.oacc-fortran/lib-16-2.f90: Add async
waits.
commit 990d954e84ff393e4d07661d82979e9acce6c9a0
Author: Julian Brown
Date: Thu Jan 2 16:14:17 2020 -0800
Fix deep-copy-10.c and lib-16-2.f90 async-saf
On Wed, 18 Dec 2019 18:44:04 +0100
Thomas Schwinge wrote:
> Hi!
>
> On 2019-12-17T22:02:25-0800, Julian Brown
> wrote:
> > This patch series provides support for OpenACC 2.6's manual deep
> > copy (attach/detach) feature.
>
> Thanks.
>
>
> There is
On Wed, 18 Dec 2019 10:18:14 +0100
Thomas Schwinge wrote:
> Hi Julian!
>
> Thanks for walking me through this.
>
> On 2019-12-14T00:19:04+0000, Julian Brown
> wrote:
> > On Fri, 13 Dec 2019 16:25:25 +0100
> > Thomas Schwinge wrote:
> >> On 2019-10-29T
This patch builds on the Fortran front-end support posted earlier in
this series to enable polymorphic class pointers to be used in OpenACC
directives as well. It was last posted here:
https://gcc.gnu.org/ml/gcc-patches/2019-10/msg00541.html
This version is largely the same as the previous
This patch has been broken out of the "OpenACC 2.6 manual deep copy
support" patch, last posted here:
https://gcc.gnu.org/ml/gcc-patches/2019-11/msg02376.html
This part adds Fortran execution tests to libgomp.
Tested alongside other patches in this series with offloading to
NVPTX. OK?
This patch has been broken out of the "OpenACC 2.6 manual deep copy
support" patch, last posted here:
https://gcc.gnu.org/ml/gcc-patches/2019-11/msg02376.html
This part contains the Fortran front-end support for parsing the new
attach and detach clauses, as well as derived-type members on
This patch has been broken out of the "OpenACC 2.6 manual deep copy
support" patch, last posted here:
https://gcc.gnu.org/ml/gcc-patches/2019-11/msg02376.html
This part adds C and C++ execution tests to libgomp.
Tested alongside other patches in this series with offloading to
NVPTX. OK?
This patch has been broken out of the "OpenACC 2.6 manual deep copy
support" patch, last posted here:
https://gcc.gnu.org/ml/gcc-patches/2019-11/msg02376.html
This part contains the C and C++ changes to parse attach and detach
clauses and struct member accesses via "." or "->" on other
This patch has been broken out of the "OpenACC 2.6 manual deep copy
support" patch, last posted here:
https://gcc.gnu.org/ml/gcc-patches/2019-11/msg02376.html
This part contains the middle-end support for OpenACC 2.6 attach and
detach operations, either as standalone clauses or as
This patch has been broken out of the "OpenACC 2.6 manual deep copy
support" patch, last posted here:
https://gcc.gnu.org/ml/gcc-patches/2019-11/msg02376.html
It contains just the minimal libgomp bits necessary to support the OpenACC
runtime API routines (acc_attach, acc_detach and
This patch has been broken out of the "OpenACC 2.6 manual deep copy
support" patch, last posted here:
https://gcc.gnu.org/ml/gcc-patches/2019-11/msg02376.html
This part contains the libgomp runtime support for the GOMP_MAP_ATTACH and
GOMP_MAP_DETACH mapping kinds (etc.), as introduced by the
This is a rebased version of the reference count consistency checking
patch last posted upstream here:
https://gcc.gnu.org/ml/gcc-patches/2019-10/msg00239.html
This is not necessary for proper operation of the rest of the patches
in this series, but has proved useful in development.
Tested
This patch was previously posted here:
https://gcc.gnu.org/ml/gcc-patches/2019-11/msg00321.html
This version is the same as the last-posted version. The middle-end patch
later in the series depends on this. Tested alongside other patches in
this series with offloading to NVPTX. OK?
Thanks,
This patch was previously approved here, but I have not committed it yet
(without the other patches in this series):
https://gcc.gnu.org/ml/gcc-patches/2019-10/msg01156.html
Included for completeness. I will commit this alongside other patches
if they are approved (or it could probably go in
This patch has been broken out of the "OpenACC 2.6 manual deep copy
support" patch, last posted here:
https://gcc.gnu.org/ml/gcc-patches/2019-11/msg02376.html
This part is included for completeness. It is the same as the patch
posted by Thomas here:
PR92843 test to fail, though IMO that test relies on behaviour arising
from a rather nuanced reading of the spec. Hopefully we can resolve that
problem as a follow-up.
Tested alongside other patches in this series with offloading to
NVPTX. OK?
Julian
2019-11-22 Julian Brown
Thomas
with offloading to NVPTX. Further commentary (together with
links to previous submissions) is provided alongside individual patches,
where relevant.
Thanks,
Julian
Julian Brown (13):
Use aux struct in libgomp for infrequently-used/API-specific data
OpenACC reference count overhaul
OpenACC reference
On Sat, 14 Dec 2019 00:19:04 +
Julian Brown wrote:
> On Fri, 13 Dec 2019 16:25:25 +0100
> Thomas Schwinge wrote:
>
> > Hi Julian!
> >
> > On 2019-10-29T12:15:01+, Julian Brown
> > wrote:
> > > static int
> > > -find_poi
On Fri, 13 Dec 2019 16:25:25 +0100
Thomas Schwinge wrote:
> Hi Julian!
>
> On 2019-10-29T12:15:01+0000, Julian Brown
> wrote:
> > static int
> > -find_pointer (int pos, size_t mapnum, unsigned short *kinds)
> > +find_group_last (int pos, size_t
On Fri, 13 Dec 2019 15:13:53 +0100
Thomas Schwinge wrote:
> Hi!
>
> Julian, Tobias, regarding the following OpenACC 'exit data' 'finalize'
> handling:
>
> On 2018-05-25T13:01:58-0700, Cesar Philippidis
> wrote:
> > --- a/gcc/gimplify.c
> > +++ b/gcc/gimplify.c
>
> > @@ -10859,6 +10849,53
hangeLog
2019-12-12 Julian Brown
Thomas Schwinge
PR libgomp/92881
libgomp/
* libgomp.h (gomp_remove_var_async): Add prototype.
* oacc-mem.c (delete_copyout): Call gomp_remove_var_async instead of
gomp_remove_var.
* target.c (gomp_unref_tgt): Change return type
Hi,
This patch provides a fix for PR92888, wherein global variables mapped
using an OpenACC 'declare' directive would not be visible to
device-pointer lookups.
Tested with offloading to nvptx. OK?
Thanks,
Julian
ChangeLog
2019-12-12 Julian Brown
PR libgomp/92888
libgomp
On Mon, 9 Dec 2019 15:44:25 +0100
Thomas Schwinge wrote:
> Hi Julian!
>
> On 2019-10-03T09:35:04-0700, Julian Brown
> wrote:
> > --- a/libgomp/oacc-mem.c
> > +++ b/libgomp/oacc-mem.c
>
> > @@ -715,48 +684,34 @@ delete_copyout (unsigned f, void *h, si
Hi!
On Fri, 15 Nov 2019 13:41:11 -0800
Julian Brown wrote:
> This patch provides support for gang local storage allocation in
> shared memory. It is mostly identical to the version posted
> previously, with one cosmetic fix (a duplicated identical condition):
>
> https://gcc
rtitioning
patches).
Thanks,
Julian
ChangeLog
gcc/
* config/gcn/gcn.h (FIXED_REGISTERS): Make s6/s7 fixed registers.
commit a4e9b17ba2bcdc3ab206fa4424cb55bc320fd092
Author: Julian Brown
Date: Mon Dec 2 16:53:50 2019 -0800
Use fixed registers for queue ptr sgpr pair
On Tue, 3 Dec 2019 15:06:48 +0100
Thomas Schwinge wrote:
> We probably can't/shouldn't change 'amdgcn' in the target triplet now,
> but as far as I'm concerned, it's not too late to change
> 'gcc/config/gcn' etc., but I guess that won't happen: too much
> effort. (And then, I don't feel too
On Tue, 3 Dec 2019 10:32:57 +0100
Thomas Schwinge wrote:
> Hi!
>
> On 2019-12-02T14:50:42+0000, Julian Brown
> wrote:
> > On Mon, 2 Dec 2019 15:43:29 +0100
> > Thomas Schwinge wrote:
> >
> >> > --- a/libgomp/openacc.h
> >> > +++ b/l
On Mon, 2 Dec 2019 15:43:29 +0100
Thomas Schwinge wrote:
> > --- a/libgomp/openacc.h
> > +++ b/libgomp/openacc.h
> > @@ -55,6 +55,7 @@ typedef enum acc_device_t {
> >/* acc_device_host_nonshm = 3 removed. */
> >acc_device_not_host = 4,
> >acc_device_nvidia = 5,
> > + acc_device_gcn
On Thu, 28 Nov 2019 15:04:05 +
Joseph Myers wrote:
> On Thu, 28 Nov 2019, Julian Brown wrote:
>
> > Unlike e.g. the _FloatN types, when decimal floating-point types are
> > enabled, common tree nodes are created for each float type size
> > (e.g. dfloat32_type_
c/
* tree.c (build_common_tree_nodes): Use pointer type for
dfloat32_ptr_type_node, dfloat64_ptr_type_node and
dfloat128_ptr_type_node when decimal floating point support
is disabled.
commit 17119773a8a45af098364b4faafe68f2e868479a
Author: Julian Brown
Date: Wed Nov 27 18:41:56 2019
On Sat, 9 Nov 2019 01:28:51 +
Julian Brown wrote:
> On Thu, 31 Oct 2019 19:11:57 +0100
> Thomas Schwinge wrote:
>
> > So that's not related to reference counting, needs to be discussed
> > separately.
> >
> > ..., and while I do agree that t
On Fri, 22 Nov 2019 16:11:15 +
Kwok Cheung Yeung wrote:
> libgomp.oacc-fortran/lib-16.f90 in the libgomp testsuite currently
> fails with:
>
> Program aborted. Backtrace:
> #0 0x4013bf in MAIN__
> at
> src/gcc-gcn-master/libgomp/testsuite/libgomp.oacc-fortran/lib-16.f90:28
> #1
This patch flips the switch to enable worker partitioning on AMD GCN.
OK?
Thanks,
Julian
ChangeLog
gcc/
* config/gcn/gcn.c (gcn_goacc_validate_dims): Remove
no-flag_worker-partitioning assertion.
(TARGET_GOACC_WORKER_PARTITIONING): Define target hook to true.
This patch provides AMD GCN support for the parallel-dims.f90 test's
parallel-dims-aux.c helper.
OK?
Thanks,
Julian
ChangeLog
libgomp/
* testsuite/libgomp.oacc-fortran/parallel-dims-aux.c: Support AMD GCN.
---
.../testsuite/libgomp.oacc-fortran/parallel-dims-aux.c | 9
This patch checks that cfun is valid in the gcn_asm_output_symbol_ref
function. This prevents a crash when that function is called with NULL
cfun, i.e. when outputting debug symbols.
OK?
Thanks,
Julian
ChangeLog
gcc/
* config/gcn/gcn.c (gcn_asm_output_symbol_ref): Handle null
This patch provides some race-condition fixes for tests that broke
for AMD GCN.
OK?
Thanks,
Julian
ChangeLog
libgomp/
* testsuite/libgomp.oacc-c-c++-common/lib-94.c: Fix race condition.
* testsuite/libgomp.oacc-fortran/lib-16-2.f90: Likewise.
---
This patch corrects a possible race condition in locking for the
asynchronous queue-full condition check in the AMD GCN libgomp plugin.
OK?
Julian
ChangeLog
libgomp/
* plugin/plugin-gcn.c (wait_for_queue_nonfull): Don't lock/unlock
aq->mutex here.
It appears that the hsa_memory_copy API routine
has problems copying from read-only data: in the
libgomp/testsuite/libgomp.oacc-c-c++-common/reduction-8.c test, a "const"
variable cannot be successfully copied to the target. I think the problem
is with read-only page mappings (in the HSA runtime).
The AMD GCN runtime support appears to exercise asynchronous operations
more heavily than other offload targets. As such, latent problems with
asynchronous host-to-device copies have come to light with GCN. This
patch provides a solution to those.
Previously posted for the og9 branch (with some
This is the main patch implementing worker-partitioning support on AMD
GCN. The following description is taken from the version of the patch
submitted on the openacc-gcc-9-branch:
This patch implements worker-partitioning support in the middle end,
by rewriting gimple. The OpenACC execution model
This patch adjusts expected output for several tests after the oaccdevlow
pass is split into three by an earlier patch in this series.
OK?
Julian
ChangeLog
gcc/testsuite/
* c-c++-common/goacc/classify-kernels-unparallelized.c,
c-c++-common/goacc/classify-kernels.c,
This patch provides AMD GCN-specific parts supporting middle-end
worker partitioning. The adjust_propagation_record hook is now called
create_propagation_record. Several builtins are redefined to take an
argument in a special address space (corresponding to GPU shared memory).
Tested alongside
Reference-type private variables or reference-type variables used as
reduction targets do not work well with the scheme to implement worker
partitioning on AMD GCN. This patch (originally by Cesar Philippidis, but
modified somewhat) provides support for replacing such variables with new
GPUs.
The series has been tested as a whole with offloading to AMD GCN.
Further commentary is provided alongside individual patches.
OK for trunk?
Thanks,
Julian
Julian Brown (13):
Add support for gang local storage allocation in shared memory
Target-dependent gang-private variable decl
This patch provides support for the adjust_private_decl hook introduced
by the previous patch changing the type of its decl argument, e.g. if the
offload target forces it into another address space. Any ADDR_EXPR nodes
that have the decl as an argument will have the wrong type: this patch
This patch provides support for gang local storage allocation in shared
memory. It is mostly identical to the version posted previously, with
one cosmetic fix (a duplicated identical condition):
https://gcc.gnu.org/ml/gcc-patches/2019-11/msg00448.html
Tested alongside other patches in this
On Thu, 14 Nov 2019 12:08:45 +0100
Thomas Schwinge wrote:
> Hi!
>
> In context of reviewing Julian's "OpenACC reference count overhaul",
> I'm generally reviewing (also known as: trying to understand) the
> libgomp OpenMP 'target' "refcount"ing , and I noticed something
> strange (?):
>
> On
On Thu, 31 Oct 2019 19:11:57 +0100
Thomas Schwinge wrote:
> Hi Julian!
>
> On 2019-10-29T12:15:01+0000, Julian Brown
> wrote:
> > This is a new version of the patch which hopefully addresses all
> > review comments. Further commentary below.
>
> Thanks, great,
oacc-c-c++-common/lib-30.c: Likewise.
commit 00607b06c8e506b0f0744a230856e1e8776633c3
Author: Julian Brown
Date: Thu Nov 7 14:24:49 2019 -0800
OpenACC "present" subarrays: runtime API return value and unmapping fixes
libgomp/
* oacc-mem.c (present_create_co
bly be replaced with some lighter-weight
mechanism as a followup).
I realised I omitted to make some of the cosmetic changes Thomas
highlighted below on starting to write this email, but I can do that
(with suitable retesting) if desired before committing.
On Wed, 12 Jun 2019 20:42:16 +0100
Julian B
ig_base' supplied), and the other caller takes the second branch
> ('orig_base' not supplied). If you tell me that's not feasible, then
> I shall again try to understand that code in the form you've proposed.
...is left outside the refactored function, as you suggest here.
Hi!
This is a new version of the patch which hopefully addresses all review
comments. Further commentary below.
On Mon, 21 Oct 2019 16:14:11 +0200
Thomas Schwinge wrote:
> On 2019-10-03T09:35:04-0700, Julian Brown
> wrote:
> > This patch has been broken out of the patch supporting
during OpenACC shutdown.
commit bbf990a35ab5a7b5100b03c5c5aca11ef0729e7b
Author: Julian Brown
Date: Fri Oct 18 11:47:37 2019 -0700
Remove vars before TLS data during OpenACC shutdown
libgomp/
* oacc-init.c (acc_shutdown_1): Remove variable mappings before TLS
On Tue, 15 Oct 2019 17:30:06 +0200
Thomas Schwinge wrote:
> Hi Julian!
>
> On 2019-10-03T09:35:04-0700, Julian Brown
> wrote:
> > This patch has been broken out of the patch supporting OpenACC 2.6
> > manual deep copy last posted here:
> >
> > https:
This patch (for the openacc-gcc-9-branch) reworks how the partitioning
level for OpenACC "private" variables is calculated and represented in
the compiler. There have been two previous approaches:
- The first (by Chung-Lin Tang) recorded which variables should be
made private per-gang in each
This patch adds support for AMD GCN offloading to the
libgomp.oacc-c-c++-common/serial-dims.c test case.
I will apply to the og9 branch shortly.
Thanks,
Julian
ChangeLog
libgomp/
* testsuite/libgomp.oacc-c-c++-common/serial-dims.c: Support AMD GCN.
---
On Wed, 9 Oct 2019 14:40:42 +0100
Jozef Lawrynowicz wrote:
> Constants generated for modes with fewer bits than in HOST_WIDE_INT
> must be sign extended to full width (e.g., with gen_int_mode). For
> constants for modes with more bits than in HOST_WIDE_INT the implied
> high order bits of that
This patch provides basic support for Fortran (2003) polymorphic class
pointers. Such pointers have a descriptor that is somewhat like an array
descriptor, so I re-used the GOMP_MAP_TO_PSET mapping to transfer such
class descriptors from the host to the target. That seems to work well,
though I
Julian Brown (4):
Add function for pretty-printing OpenACC clause names
Use gomp_map_val for OpenACC host-to-device address translation
Factor out duplicate code in gimplify_scan_omp_clauses
OpenACC 2.6 manual deep copy support (attach/detach)
gcc/c-family/c-common.h
This patch factors out some code in gimplify_scan_omp_clauses into two
new outlined functions. Previously approved here:
https://gcc.gnu.org/ml/gcc-patches/2018-12/msg01309.html
FAOD, OK for trunk?
Thanks,
Julian
ChangeLog
gcc/
* gimplify.c (insert_struct_comp_map,
This patch uses gomp_map_val for OpenACC host-to-device address
translation instead of open-coding the device address calculation.
OK for trunk?
Thanks,
Julian
ChangeLog
libgomp/
* libgomp.h (gomp_map_val): Add prototype.
* oacc-parallel.c (GOACC_parallel_keyed): Use
This patch adds a function to pretty-print OpenACC clause names from
OMP_CLAUSE_MAP_KINDs, for error output. The function is used by subsequent
patches.
Previously approved as part of:
https://gcc.gnu.org/ml/gcc-patches/2018-12/msg01292.html
FAOD, OK for trunk?
Julian
ChangeLog
to NVPTX. OK for trunk?
Thanks,
Julian
2019-10-02 Julian Brown
libgomp/
* libgomp.h (RC_CHECKING): New macro, disabled by default, guarding all
hunks in this patch.
(target_mem_desc): Add refcount_chk, mark fields.
(splay_tree_key_s): Add refcount_chk field
anticipate
that support.
Tested with offloading to NVPTX, with good results (though a couple of
tests need fixing also).
A follow-up patch provides self-checking of the reference count
implementation.
OK for trunk?
Thanks,
Julian
2019-10-02 Julian Brown
libgomp/
* libgomp.h
by Thomas in:
https://gcc.gnu.org/ml/gcc-patches/2018-12/msg01723.html
This patch comprises the remaining parts. Tested (alongside other
patches) with offloading to NVPTX. OK for trunk?
Thanks,
Julian
2019-10-02 Julian Brown
Cesar Philippidis
libgomp/
* libgomp.h
23b6eb1d74 100644
--- a/gcc/ChangeLog.openacc
+++ b/gcc/ChangeLog.openacc
@@ -1,3 +1,8 @@
+2019-09-20 Julian Brown
+
+ * gimplify.c (localize_reductions): Rewrite references for
+ OMP_CLAUSE_PRIVATE also.
+
2019-09-17 Tobias Burnus
* config/gcn/gcn.c (gcn_expand_scalar_to_vect
943a9e4a933..7813760e642 100644
--- a/libgomp/ChangeLog.openacc
+++ b/libgomp/ChangeLog.openacc
@@ -1,3 +1,8 @@
+2019-09-19 Julian Brown
+
+ * plugin/plugin-nvptx.c (GOMP_OFFLOAD_openacc_async_host2dev):
+ Add EPHEMERAL parameter, and FIXME function comment.
+
2019-09-18 Tobias Burnus
insertions(+), 88 deletions(-)
diff --git a/libgomp/ChangeLog.openacc b/libgomp/ChangeLog.openacc
index 41e05e9c61b..5f39fae6f51 100644
--- a/libgomp/ChangeLog.openacc
+++ b/libgomp/ChangeLog.openacc
@@ -1,3 +1,29 @@
+2019-09-17 Julian Brown
+
+ * oacc-host.c (host_openacc_async_queu
cc b/libgomp/ChangeLog.openacc
index 5f39fae6f51..1006b8149c8 100644
--- a/libgomp/ChangeLog.openacc
+++ b/libgomp/ChangeLog.openacc
@@ -1,3 +1,36 @@
+2019-09-17 Julian Brown
+
+ * libgomp-plugin.h (GOMP_OFFLOAD_openacc_async_host2dev): Update
+ prototype.
+ * libgomp.h (gomp_copy_host2de
-git a/gcc/testsuite/ChangeLog.openacc b/gcc/testsuite/ChangeLog.openacc
index 899b9cf1783..6faedb6866a 100644
--- a/gcc/testsuite/ChangeLog.openacc
+++ b/gcc/testsuite/ChangeLog.openacc
@@ -1,3 +1,16 @@
+2019-09-17 Julian Brown
+
+ * c-c++-common/goacc/classify-kernels-unparallelized
/libgomp/ChangeLog.openacc
@@ -1,3 +1,8 @@
+2019-09-17 Julian Brown
+
+ * target.c (gomp_map_vars_internal): Remove read of uninitialised
+ data.
+
2019-09-17 Julian Brown
* testsuite/libgomp.oacc-c-c++-common/async_queue-1.c: Only run
diff --git a/libgomp/target.c b/libgomp
/libgomp/ChangeLog.openacc
@@ -1,3 +1,14 @@
+2019-09-17 Julian Brown
+
+ * testsuite/libgomp.oacc-c-c++-common/async_queue-1.c: Only run
+ NVidia-specific test on NVidia hardware.
+ * testsuite/libgomp.oacc-c-c++-common/asyncwait-nop-1.c (main):
+ Initialise
+1,9 @@
+2019-09-10 Julian Brown
+
+ * config/gcn/mkoffload.c (process_asm): Remove omp_data_size,
+ gridified_kernel_p, kernel_dependencies_count, kernel_dependencies
+ from emitted hsa_kernel_description struct array.
+
2019-09-10 Andrew Stubbs
* config/gcn/gcn.c
/ChangeLog.openacc
index 59605cd85bf..a15d4a0ed49 100644
--- a/libgomp/ChangeLog.openacc
+++ b/libgomp/ChangeLog.openacc
@@ -1,3 +1,8 @@
+2019-09-10 Julian Brown
+
+ * plugin/plugin-gcn.c (GOMP_OFFLOAD_openacc_async_host2dev): Enqueue
+ copy from src_copy not src.
+
2019-09-10 Kwok
/ChangeLog.openacc
@@ -1,3 +1,24 @@
+2019-09-10 Julian Brown
+
+ * plugin/plugin-gcn.c (struct placeholder, struct asyncwait_info,
+ enum entry_type): New.
+ (queue_entry): Use entry_type enum for tag. Add asyncwait and
+ placeholder event type fields
/gcc/ChangeLog.openacc
@@ -1,3 +1,8 @@
+2019-09-06 Julian Brown
+
+ * gimplify.c (gimplify_omp_for): Use for_stmt in call to
+ localize_reductions.
+
2019-09-06 Julian Brown
* config/nvptx/nvptx.c (omp-sese.h): Include.
diff --git a/gcc/gimplify.c b/gcc/gimplify.c
index
This patch adds profiling support to the AMD GCN libgomp plugin, modeled
after the equivalent support in the NVPTX plugin. This gives a positive
test delta in AMD GCN offload testing.
I will apply to the openacc-gcc-9-branch shortly.
Julian
2019-09-06 Julian Brown
libgomp
100644
--- a/libgomp/ChangeLog.openacc
+++ b/libgomp/ChangeLog.openacc
@@ -1,3 +1,8 @@
+2019-09-06 Julian Brown
+
+ * config/gcn/target.c (omp_pause_resource, omp_pause_resource_all): New
+ functions, plus ialiases.
+
2019-09-05 Julian Brown
* plugin/plugin-gcn.c (gcn_exec
On Thu, 5 Sep 2019 16:01:19 +0100
Julian Brown wrote:
> On Thu, 5 Sep 2019 21:52:00 +0800
> Chung-Lin Tang wrote:
>
> > On 2019/9/5 9:45 AM, Julian Brown wrote:
> > > Much of omp-sese.c originates from code written for NVPTX by
> > > Nathan Sidwell (adapted
/ChangeLog.openacc
@@ -1,3 +1,8 @@
+2019-09-06 Julian Brown
+
+ * gimplify.c (gimplify_omp_workshare): Use OMP_CLAUSES, OMP_BODY
+ instead of OMP_TARGET_CLAUSES, OMP_TARGET_BODY.
+
2019-09-05 Andrew Stubbs
Backport from mainline:
diff --git a/gcc/gimplify.c b/gcc
9e7893aa11e..ffe19bc5809 100644
--- a/gcc/ChangeLog.openacc
+++ b/gcc/ChangeLog.openacc
@@ -1,3 +1,25 @@
+2019-09-06 Julian Brown
+
+ * config/nvptx/nvptx.c (omp-sese.h): Include.
+ (bb_pair_t, bb_pair_vec_t, pseudo_node_t, bracket, bracket_vec_t,
+ bb_sese, bb_sese::~bb_sese
On Thu, 5 Sep 2019 21:52:00 +0800
Chung-Lin Tang wrote:
> On 2019/9/5 9:45 AM, Julian Brown wrote:
> > Much of omp-sese.c originates from code written for NVPTX by Nathan
> > Sidwell (adapted to work on gimple instead of RTL) -- though at
> > present, only the p
insertions(+), 37 deletions(-)
diff --git a/gcc/ChangeLog.openacc b/gcc/ChangeLog.openacc
index 0d068ac8ae2..2b7f616810d 100644
--- a/gcc/ChangeLog.openacc
+++ b/gcc/ChangeLog.openacc
@@ -1,3 +1,19 @@
+2019-09-05 Cesar Philippidis
+ Julian Brown
+
+ * gimplify.c (privatize_reduction
@@
+2019-09-05 Julian Brown
+
+ * config/gcn/gcn.c (gcn_goacc_validate_dims): Remove
+ no-flag_worker-partitioning assertion.
+ (TARGET_GOACC_WORKER_PARTITIONING): Define target hook to true.
+ * config/gcn/gcn.opt (flag_worker_partitioning): Change default to 1.
+
2019-09
401 - 500 of 814 matches
Mail list logo