Extend the remove_redundant_vector pass to handle vector broadcasts from
constant and variable scalars. When broadcasting from constants and
function arguments, we can place a single widest vector broadcast at
entry of the nearest common dominator for basic blocks with all uses
since constants and
Hi!
The test PASSes for C, but FAILs for C++:
.../libgomp.c-c++-common/omp_target_memset-3.c: In function 'void
test_it(void*, int, size_t)':
.../libgomp.c-c++-common/omp_target_memset-3.c:31:7: warning: pointer of type
'void *' used in arithmetic [-Wpointer-arith]
.../libgomp.c-c++-common/omp_t
Hi Joseph,
> I don't see tests for the various special cases from Annex F (for example,
> "tanpi(n) returns +0, for positive even and negative odd integers n." and
> "tanpi(n) returns -0, for positive odd and negative even integers n.").
> In such cases the sign of zero would need to be checked sp
> Am 03.06.2025 um 05:59 schrieb Andrew Pinski :
>
> This testcase was fixed by r16-906-g8da568c885dc90. Since
> this is a C testcase, it would be useful to have a C testcase besides
> a C++ one too.
>
> Tested for x86_64-linux-gnu.
Ok
Richard
>PR tree-optimization/116824
>
> gcc/tes
> Am 03.06.2025 um 04:51 schrieb Andrew Pinski :
>
> When we have a smallish CSWTCH, it could be placed in the rodata.cst16
> section so it can be merged with other constants across TUs.
>
> The fix is simple; just mark the decl as mergable (DECL_MERGEABLE).
> DECL_MERGEABLE was added with r1
When we have a smallish CSWTCH, it could be placed in the rodata.cst16
section so it can be merged with other constants across TUs.
The fix is simple; just mark the decl as mergable (DECL_MERGEABLE).
DECL_MERGEABLE was added with r14-1500-g4d935f52b0d5c0 specifically
to improve these kind of decls
This testcase was fixed by r16-906-g8da568c885dc90. Since
this is a C testcase, it would be useful to have a C testcase besides
a C++ one too.
Tested for x86_64-linux-gnu.
PR tree-optimization/116824
gcc/testsuite/ChangeLog:
* gcc.dg/tree-ssa/phiprop-2.c: New test.
Signed-off-b
I am a little hesitant about whether to let -march implicitly set
-mtune, because this is equivalent to making -march become another
-mcpu with higher priority, and we cannot avoid discussing the
priority of -march and -mtune again, and no matter what the priority
is, it will introduce more complic
On 6/2/25 7:34 PM, Kito Cheng wrote:
LGTM, and will commit once CI happy, BTW, next time you could name the
testcase into arch-.c e.g. arch-svbare.c, that could prevent
potential filename conflict. :)
Yea, I nearly made a similar suggestion. Not sure why we're using
numeric suffixes here. T
I add a ChangeLog and update testname in
https://gcc.gnu.org/pipermail/gcc-patches/2025-June/685401.html
Thanks
Dongyan Chen
在 2025/6/3 3:34, Jeff Law 写道:
On 5/29/25 7:27 AM, Dongyan Chen wrote:
This patch support svbare extension, which is an extension in RVA23
profile.
To enable GCC to
Okay, thanks, I get that.
Dongyan Chen
在 2025/6/3 9:34, Kito Cheng 写道:
LGTM, and will commit once CI happy, BTW, next time you could name the
testcase into arch-.c e.g. arch-svbare.c, that could prevent
potential filename conflict. :)
On 6/2/25 6:28 PM, Iain Sandoe wrote:
Tested on x86_64-darwin and powerpc64le so far, how does this look?
thanks
Iain
--- 8< ---
Using lookup_template_class () directly on the coroutine_handle identifier
fails in the reported test because the using TYPE_DECL is found. This is
because
lookup i
On 6/2/25 5:27 PM, Iain Sandoe wrote:
On 2 Jun 2025, at 21:17, Jason Merrill wrote:
On 6/1/25 3:30 AM, Iain Sandoe wrote:
Updated. I realised we no longer need to refer to
initial_await_resume_called in the ramp at all, v2 removes the setting
of the variable from there and puts it into the
LGTM, and will commit once CI happy, BTW, next time you could name the
testcase into arch-.c e.g. arch-svbare.c, that could prevent
potential filename conflict. :)
On Tue, Jun 3, 2025 at 9:22 AM Dongyan Chen
wrote:
>
> This patch support svbare extension, which is an extension in RVA23 profile.
>
This patch support svbare extension, which is an extension in RVA23 profile.
To enable GCC to recognize and process svbare extension correctly at compile
time.
gcc/ChangeLog:
* config/riscv/riscv-ext.def: New extension defs.
* config/riscv/riscv-ext.opt: Ditto.
* doc/risc
Tested on x86_64-darwin and powerpc64le so far, how does this look?
thanks
Iain
--- 8< ---
Using lookup_template_class () directly on the coroutine_handle identifier
fails in the reported test because the using TYPE_DECL is found. This is
because
lookup is called with default parameters that me
On Tue, Jun 3, 2025 at 6:06 AM H.J. Lu wrote:
>
> On Mon, Jun 2, 2025 at 4:45 PM Richard Sandiford
> wrote:
> >
> > "H.J. Lu" writes:
> > > On Mon, Jun 2, 2025 at 4:07 PM Richard Sandiford
> > > wrote:
> > >>
> > >> "H.J. Lu" writes:
> > >> > Since not all CALL instructions in RTL passes have
Joseph Myers wrote:
On Sun, 1 Jun 2025, Yuao Ma wrote:
For MPFR versions older than 4.2.0, we've included our own folding functions.
I think the normal practice in GCC would be to avoid the optimizations
when the MPFR support is absent, instead of working around the absence
with possibly less a
On Mon, Jun 2, 2025 at 4:45 PM Richard Sandiford
wrote:
>
> "H.J. Lu" writes:
> > On Mon, Jun 2, 2025 at 4:07 PM Richard Sandiford
> > wrote:
> >>
> >> "H.J. Lu" writes:
> >> > Since not all CALL instructions in RTL passes have a REG_CALL_DECL note,
> >> > update get_call_fndecl to also check f
> On 2 Jun 2025, at 21:17, Jason Merrill wrote:
>
> On 6/1/25 3:30 AM, Iain Sandoe wrote:
>> Updated. I realised we no longer need to refer to
>> initial_await_resume_called in the ramp at all, v2 removes the setting
>> of the variable from there and puts it into the start of the actor.
>> Tes
From: Gwenole Beauchesne
Handle '#pragma GCC optimize' earlier as the __OPTIMIZE__ macro may need
to be defined as well for certain usages. Add additional tests for the
'#pragma GCC target' case with auto-vectorization enabled and multiple
combinations of namespaces and/or class member functions.
On Thu, May 29, 2025 at 05:59:44PM +0100, Richard Sandiford wrote:
> Sorry for the slow reply.
>
> Dimitar Dimitrov writes:
> > On Fri, May 16, 2025 at 06:14:30PM +0100, Richard Sandiford wrote:
> >> Dimitar Dimitrov writes:
...
> >> It might still be worth trying to use simplify_subreg_regno an
On Mon, 02 Jun 2025 12:35:45 PDT (-0700), Robin Dapp wrote:
I don't quite follow this part. IIUC the rules before this patch were
-march=ISA: Generate code that requires the given ISA, without
changing the tuning model.
-mcpu=CPU: Generate code for the given CPU, targeting all the
libstdc++-v3/ChangeLog:
* include/std/type_traits (is_destructible, is_destructible_v):
Define using new built-in.
(is_nothrow_destructible, is_nothrow_destructible_v): Likewise.
(is_trivially_destructible, is_trivially_destructible_v):
Likewise.
---
Thanks
On 6/1/25 3:30 AM, Iain Sandoe wrote:
Updated. I realised we no longer need to refer to
initial_await_resume_called in the ramp at all, v2 removes the setting
of the variable from there and puts it into the start of the actor.
Tested on x86_64-darwin and powerp64le-linux.
Confirmed that the sani
These function templates are apparently never used or instantiated,
because they don't compile. This fixes them, but they're still unused.
I plan to make use of them in a later commit.
libstdc++-v3/ChangeLog:
* include/bits/atomic_timed_wait.h (__atomic_wait_address_until_v):
Do n
On Mon, 2025-06-02 at 17:34 +0200, Andreas Schwab wrote:
> On Jun 01 2025, John Paul Adrian Glaubitz wrote:
>
> > I was talking about the comments, not the code since NetBSD and Linux
> > disagree on what the SVR4 ABI claims.
>
> In which way?
Both NetBSD and Linux return structs differently, ye
These changes cause genapi.cc to use whichever of clock_gettime() or
gettimeofday() are available. This prevents compilation errors on
systems where clock_gettime() is not available.
gcc/cobol/ChangeLog:
PR cobol/119975
* genapi.cc (parser_intrinsic_call_0
On 6/2/25 12:15, Tobias Burnus wrote:
Not really new - but as the topic came up elsewhere (OpenMP issue 4455),
I had a second thought about this and I think it make sense to add a note.
GCC uses the OpenMP 6.0 version.
The problem is that 'int'/'int*' became 'omp_interop_rc_t ret_code'
and 'omp
Hi!
On Mon, Jun 02, 2025 at 08:37:19PM +0200, Martin Uecker wrote:
> Am Montag, dem 02.06.2025 um 13:19 -0500 schrieb Segher Boessenkool:
> > On Mon, Jun 02, 2025 at 05:50:08PM +0200, Martin Uecker wrote:
> > > According to the discussion in the bugzilla there seems to be
> > > some consensus to a
On Mon, 2025-06-02 at 21:41 +0200, Andreas Schwab wrote:
> On Jun 02 2025, John Paul Adrian Glaubitz wrote:
>
> > To me both statements seem to contradict each other.
>
> How? They comment two different things.
But it's DEFAULT_PCC_STRUCT_RETURN in both cases, no?
Adrian
--
.''`. John Paul
Hi Andre,
attached patch fixes a missing substring ref on a saved allocatable string.
The issue seems to be, that the variable is declared to be a character pointer
and not a character array. When using the latter (why not), it works as
expected and does not produce any regressions.
Regtests o
Tested x86_64-pc-linux-gnu, applying to trunk.
-- 8< --
PR c++/107600
gcc/cp/ChangeLog:
* method.cc (destructible_expr): Fix refs and arrays of unknown
bound.
gcc/testsuite/ChangeLog:
* g++.dg/ext/is_destructible2.C: Add more cases.
---
gcc/cp/method.cc
On Jun 02 2025, John Paul Adrian Glaubitz wrote:
> To me both statements seem to contradict each other.
How? They comment two different things.
--
Andreas Schwab, sch...@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510 2552 DF73 E780 A9DA AEC1
"And now for something completely di
On 5/27/25 1:32 AM, Jiawei wrote:
This patch add the RISC-V Shlcofideleg extension. It supports delegating
LCOFI interrupts(the count-overflow interrupts) to VS-mode.[1]
[1] https://riscv.github.io/riscv-isa-manual/snapshot/privileged
gcc/ChangeLog:
* config/riscv/riscv-ext.def: New
I don't quite follow this part. IIUC the rules before this patch were
-march=ISA: Generate code that requires the given ISA, without
changing the tuning model.
-mcpu=CPU: Generate code for the given CPU, targeting all the
extensions that CPU supports and using the best known tu
On Mon, 2 Jun 2025, Martin Uecker wrote:
> Am Montag, dem 02.06.2025 um 18:45 + schrieb Joseph Myers:
> > On Mon, 2 Jun 2025, Martin Uecker wrote:
> >
> > > According to the discussion in the bugzilla there seems to be
> > > some consensus to activate the warning for -Wextra (I am also
> > >
On 5/29/25 7:27 AM, Dongyan Chen wrote:
This patch support svbare extension, which is an extension in RVA23 profile.
To enable GCC to recognize and process svbare extension correctly at compile
time.
Needs a ChangeLog entry and the testname needs to be updated now that
arch-59 is taken. Wit
On 5/29/25 5:44 AM, Dongyan Chen wrote:
This patch support smcntrpmf extension[1].
To enable GCC to recognize and process smcntrpmf extension correctly at compile
time.
[1]https://github.com/riscvarchive/riscv-smcntrpmf
gcc/ChangeLog:
* config/riscv/riscv-ext.def: New extension def
Am Montag, dem 02.06.2025 um 18:45 + schrieb Joseph Myers:
> On Mon, 2 Jun 2025, Martin Uecker wrote:
>
> > According to the discussion in the bugzilla there seems to be
> > some consensus to activate the warning for -Wextra (I am also
> > looking into implementing the suggested improvements t
On 6/1/25 1:12 AM, H.J. Lu wrote:
Move get_call_rtx_from to final.c and call call_from_call_insn.
PR other/120493
* final.cc (call_from_call_insn): Change the argument type to
const rtx_call_insn *.
(get_call_rtx_from): New.
* rtl.h (is_a_helper ::test): New.
(get_call_rtx_from): Moved to the
On Sun, 01 Jun 2025 22:18:21 PDT (-0700), Robin Dapp wrote:
This rule clearly applies to directly related options like -ffoo and
-fno-foo, but it’s less obvious for unrelated pairs like -ffoo and
-fbar especially when there is traditionally strong specifics.
In many cases, the principle of "
Tested x86_64-pc-linux-gnu, applying to trunk.
-- 8< --
In r16-57 I thought it was unnecessary to mention incomplete initialization
after another diagnostic, but actually it's useful elaboration.
PR c++/120506
gcc/cp/ChangeLog:
* constexpr.cc (cxx_eval_outermost_constant_expr):
Hi Thomas,
thanks for the ok. Unfortunately does the patch regress in gomp (test case
gomp/pr104382 when I am not mistaken ; the one with the lone 'save'
statement). This was reported by the regression testing host at first for
arm, but also occurs on x86_64. Since when are proposed patches ch
On Mon, 2 Jun 2025, Martin Uecker wrote:
> I *believe* TYPE_PACKED should be propagated to existing main
> variants.
>
>
> Bootstrapped and regression tested for x86_64.
>
> Martin
>
>
> c: fix ICE with enum completed with packed attribute after forward decl
> [PR116892]
>
> Aft
On Mon, 2 Jun 2025, Martin Uecker wrote:
> According to the discussion in the bugzilla there seems to be
> some consensus to activate the warning for -Wextra (I am also
> looking into implementing the suggested improvements that may
> make it suitable fo r-Wall). When making this change, I also
On Sun, 1 Jun 2025, Peter Frost wrote:
> Ping https://gcc.gnu.org/pipermail/gcc-patches/2025-January/672568.html
This needs various coding style fixes. Lines should be broken before
binary operators such as && or || rather than after, and there should be a
space before '(' in function and macr
Am Montag, dem 02.06.2025 um 13:19 -0500 schrieb Segher Boessenkool:
> Hi!
>
> On Mon, Jun 02, 2025 at 05:50:08PM +0200, Martin Uecker wrote:
> > According to the discussion in the bugzilla there seems to be
> > some consensus to activate the warning for -Wextra (I am also
> > looking into impleme
We don't use this GCC coding convention in libstdc++.
libstdc++-v3/ChangeLog:
* include/bits/basic_string.h (basic_string::size): Remove space
before parameter list.
(basic_string::capacity): Likewise.
* include/bits/stl_deque.h (deque::size): Likewise.
* i
Not really new - but as the topic came up elsewhere (OpenMP issue 4455),
I had a second thought about this and I think it make sense to add a note.
GCC uses the OpenMP 6.0 version.
The problem is that 'int'/'int*' became 'omp_interop_rc_t ret_code'
and 'omp_interop_rc_t *ret_code'.
For C++, pas
On Sun, 1 Jun 2025, Martin Uecker wrote:
> This patch now moves the checking assertion out of the recursion,
> which seems better anyhow. I could not do the check unconditionally
> for all types. because we sometimes call composite_type for with
> mismatching qualifiers.
Calling composite_type wi
Hi!
On Mon, Jun 02, 2025 at 05:50:08PM +0200, Martin Uecker wrote:
> According to the discussion in the bugzilla there seems to be
> some consensus to activate the warning for -Wextra (I am also
> looking into implementing the suggested improvements that may
> make it suitable fo r-Wall). When m
This replaces two implicit conversions from ptrdiff_t to size_t with
explicit conversions that include unreachable hints for the ptrdiff_t
value not being negative.
libstdc++-v3/ChangeLog:
* include/bits/stl_vector.h (~_Vector_base): Add unreachable
hint for negative capacity and
On Mon, 2025-06-02 at 19:44 +0200, Andreas Schwab wrote:
> On Jun 02 2025, John Paul Adrian Glaubitz wrote:
>
> > Both NetBSD and Linux return structs differently, yet the comments in the
> > source code claim their way is according to the m68k SVR4 specification.
>
> Where does it claim that?
l
On Jun 02 2025, John Paul Adrian Glaubitz wrote:
> Both NetBSD and Linux return structs differently, yet the comments in the
> source code claim their way is according to the m68k SVR4 specification.
Where does it claim that?
--
Andreas Schwab, sch...@linux-m68k.org
GPG Key fingerprint = 7578 E
On Sun, 1 Jun 2025, Yuao Ma wrote:
> For MPFR versions older than 4.2.0, we've included our own folding functions.
I think the normal practice in GCC would be to avoid the optimizations
when the MPFR support is absent, instead of working around the absence
with possibly less accurate implementa
> On 2 Jun 2025, at 17:10, Andrew Pinski wrote:
>
> On Mon, Jun 2, 2025 at 8:17 AM Jason Merrill wrote:
>>
>> On 6/2/25 5:13 AM, Iain Sandoe wrote:
>>> Tested on x86_64-darwin, OK for trunk?
>>> thanks
>>> Iain
>>>
>>> --- 8< ---
>>>
>>> Using lookup_template_class () directly on the corou
This suppresses some -Wsign-conversion warnings from Clang when
compiling with -Wsystem-headers.
libstdc++-v3/ChangeLog:
* include/std/bit (__rotl, __rotr): Use static_cast for
conversion from int to unsigned.
---
Tested x86_64-linux.
libstdc++-v3/include/std/bit | 4 ++--
1 fi
__cpp_lib_any and __cpp_lib_chrono are defined unconditionally in C++20
and __cpp_lib_three_way_comparison and __cpp_lib_concepts depend on
front-end features which are definitely supported by GCC trunk.
libstdc++-v3/ChangeLog:
* src/c++23/std.cc.in: Remove redundant checks for feature te
The __detail::__wait_until function has a comment that should have been
removed when r16-1000-g225622398a9631 changed the return type from a
std::pair to a struct with three members.
__atomic_wait_address_until_v returns __res.first which should have been
changed when __res was no longer a std::pa
Hi Richard,
On 6/2/25 01:27, Richard Sandiford wrote:
> Vineet Gupta writes:
>> +CC gcc-patches
>>
>> On 5/30/25 14:04, Vineet Gupta wrote:
>>> Hi Jeff, Richard
>>>
>>> As part of RISC-V FRM mode switching improvements, I'm running into a
>>> behavior
>>> in late_combine2 where it is eliminating
On 02/06/2025 11:06, Spencer Abson wrote:
This patch introduces expanders for FP<-FP conversions that levarage
partial vector modes. We also extend the INT<-FP and FP<-INT conversions
using the same approach.
The ACLE enables vectorized conversions like the following:
fcvt z0.h, p7/m, z1.s
Mo
> Umm, I don't think so. I've been building crosses with gcc for decades.
> It should not be necessary, though it may sometimes be convenient.
>
> If you feel there's a strong need, then you're going to have to make a
> better case than what you've done above. Specifically you'd need to
> start wi
On Jun 01 2025, John Paul Adrian Glaubitz wrote:
> I was talking about the comments, not the code since NetBSD and Linux
> disagree on what the SVR4 ABI claims.
In which way?
--
Andreas Schwab, sch...@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510 2552 DF73 E780 A9DA AEC1
"And n
From: Pan Li
This patch would like to combine the vec_duplicate + vdiv.vv to the
vdiv.vx. From example as below code. The related pattern will depend
on the cost of vec_duplicate from GR2VR. Then the late-combine will
take action if the cost of GR2VR is zero, and reject the combination
if the
On Mon, Jun 2, 2025 at 8:17 AM Jason Merrill wrote:
>
> On 6/2/25 5:13 AM, Iain Sandoe wrote:
> > Tested on x86_64-darwin, OK for trunk?
> > thanks
> > Iain
> >
> > --- 8< ---
> >
> > Using lookup_template_class () directly on the coroutine_handle identifier
> > fails in the reported test because
According to the discussion in the bugzilla there seems to be
some consensus to activate the warning for -Wextra (I am also
looking into implementing the suggested improvements that may
make it suitable fo r-Wall). When making this change, I also
noticed that it was not working for -Wc++-compat
On 6/2/25 11:43 AM, Iain Sandoe wrote:
On 2 Jun 2025, at 16:33, Iain Sandoe wrote:
Hi Jason
On 2 Jun 2025, at 16:20, Jason Merrill wrote:
On 5/31/25 3:19 PM, Iain Sandoe wrote:
Some small cleanups found while working on other changes, tested
on x86_64-darwin, OK for trunk?
thanks
Iain
-
I *believe* TYPE_PACKED should be propagated to existing main
variants.
Bootstrapped and regression tested for x86_64.
Martin
c: fix ICE with enum completed with packed attribute after forward decl
[PR116892]
After forward declaration of an enum and when completing it with the
In the comment trail for PR119966, I'd said that the validate_subreg
condition:
/* The outer size must be ordered wrt the register size, otherwise
we wouldn't know at compile time how many registers the outer
mode occupies. */
if (!ordered_p (osize, regsize))
return false;
"is
> On 2 Jun 2025, at 16:33, Iain Sandoe wrote:
>
> Hi Jason
>
>> On 2 Jun 2025, at 16:20, Jason Merrill wrote:
>>
>> On 5/31/25 3:19 PM, Iain Sandoe wrote:
>>> Some small cleanups found while working on other changes, tested
>>> on x86_64-darwin, OK for trunk?
>>> thanks
>>> Iain
>>> --- 8<
Hi Jason
> On 2 Jun 2025, at 16:20, Jason Merrill wrote:
>
> On 5/31/25 3:19 PM, Iain Sandoe wrote:
>> Some small cleanups found while working on other changes, tested
>> on x86_64-darwin, OK for trunk?
>> thanks
>> Iain
>> --- 8< ---
>> We were incorrectly guarding all the frame cleanups on the
Tested x86_64-pc-linux-gnu, applying to trunk.
-- 8< --
We don't want the new call to get_dtor to cause function instantiation.
PR c++/107600
gcc/cp/ChangeLog:
* semantics.cc (trait_expr_value) [CPTK_HAS_TRIVIAL_DESTRUCTOR]:
Add cp_unevaluated.
gcc/testsuite/ChangeLog:
Tested x86_64-pc-linux-gnu, applying to trunk.
-- 8< --
destructible_expr was wrongly assuming that TO is a class type.
When is_xible_helper was added in r8-742 it returned early for abstract
class types, which is correct for __is_constructible, but not
__is_assignable or (now) __is_destructible
On 5/30/25 5:38 PM, Pietro Monteiro wrote:
> The autoconf macro PKG_CHECK_MODULES defined in config/pkg.m4 is used in
> binutils/gdb, but not in GCC. That macro uses the obsolete AC_TRY_LINK[1].
> Update the code to use AC_LINK_IFELSE as documented by autoconf.
>
> Regenerating all autotool files
On 5/31/25 3:19 PM, Iain Sandoe wrote:
Some small cleanups found while working on other changes, tested
on x86_64-darwin, OK for trunk?
thanks
Iain
--- 8< ---
We were incorrectly guarding all the frame cleanups on the
basis of frame_needs_free (which is always set for the present
code-gen since
On 5/30/25 3:50 PM, Iain Sandoe wrote:
Tested on x86_64-darwin (but does need the addition of the coroutine
keywords to dump_expr), OK for trunk?
Thanks,
Iain
OK.
--- 8< ---
We checked that the coroutine expressions were not suitable for
constexpr, but did not emit and error when needed.
On 5/30/25 3:48 PM, Iain Sandoe wrote:
Tested on x86_64-darwin, OK for trunk?
thanks
Iain
OK.
--- 8< ---
These were omitted there as an oversight, most of the error handling
for the coroutines code is specific rather than using generic %qE etc.
gcc/cp/ChangeLog:
* error.cc (dump_ex
On 6/2/25 5:13 AM, Iain Sandoe wrote:
Tested on x86_64-darwin, OK for trunk?
thanks
Iain
--- 8< ---
Using lookup_template_class () directly on the coroutine_handle identifier
fails in the reported test because the using TYPE_DECL is found.
Hmm, this seems like a longstanding (since the implem
On 02/06/2025 15:40, Tobias Burnus wrote:
Hi Andrew,
Andrew Stubbs wrote:
The hsa_memory_copy API is known to be slow, so for smaller data sizes
it's probably better to have one hsa_memory_copy replace the whole
memset than use three API calls, even with setting up some host-side
memory to co
On 6/2/25 2:23 AM, Pengfei Li wrote:
Hi all,
I would like to bring attention back to this patch:
https://inbox.sourceware.org/gcc-patches/20250508164950.5646-1-pengfei@arm.com/
The patch improves auto-vectorization for loops with known small trip counts by
introducing a new target ho
On 6/2/25 3:01 AM, Arijit Kumar Das wrote:
Hi,
When compiling GCC for targets like nvptx-none that require newlib, we
need to put newlib-cygwin/newlib in the root directory of the source
tree (either a copy or a symlink), which is then built by GCC when
targeting offload devices like the a
From: Pan Li
Add asm dump check test for vec_duplicate + vdiv.vv combine to vdiv.vx,
with the GR2VR cost is 0, 2 and 15.
The below test suites are passed for this patch.
* The rv64gcv fully regression test.
gcc/testsuite/ChangeLog:
* gcc.target/riscv/rvv/autovec/vx_vf/vx-1-i16.c: Add a
From: Pan Li
Add asm dump check test for vec_duplicate + vdiv.vv combine to vdiv.vx,
with the GR2VR cost is 0, 1 and 2.
The below test suites are passed for this patch.
* The rv64gcv fully regression test.
gcc/testsuite/ChangeLog:
* gcc.target/riscv/rvv/autovec/vx_vf/vx-4-i16.c: Add as
From: Pan Li
Some existing vdiv related test need some adjust for the
asm check.
gcc/testsuite/ChangeLog:
* gcc.target/riscv/rvv/autovec/binop/vdiv-rv32gcv-nofm.c: Adjust
the asm check for vdiv.
* gcc.target/riscv/rvv/autovec/binop/vdiv-rv32gcv.c: Ditto.
* gcc.ta
From: Pan Li
This patch would like to introduce the combine of vec_dup + vdiv.vv into
vdiv.vx on the cost value of GR2VR. The late-combine will take place if
the cost of GR2VR is zero, or reject the combine if non-zero like 1, 15
in test. There will be two cases for the combine:
Case 0:
| .
> -Original Message-
> From: Richard Biener
> Sent: Monday, May 26, 2025 2:56 PM
> To: Tamar Christina
> Cc: gcc-patches@gcc.gnu.org; nd
> Subject: RE: [PATCH 1/2]middle-end: Apply loop->unroll directly in vectorizer
>
> On Mon, 19 May 2025, Tamar Christina wrote:
>
> > > >/* Compl
On Mon, Jun 2, 2025 at 4:45 PM Richard Sandiford
wrote:
>
> "H.J. Lu" writes:
> > On Mon, Jun 2, 2025 at 4:07 PM Richard Sandiford
> > wrote:
> >>
> >> "H.J. Lu" writes:
> >> > Since not all CALL instructions in RTL passes have a REG_CALL_DECL note,
> >> > update get_call_fndecl to also check f
"H.J. Lu" writes:
> On Mon, Jun 2, 2025 at 4:07 PM Richard Sandiford
> wrote:
>>
>> "H.J. Lu" writes:
>> > Since not all CALL instructions in RTL passes have a REG_CALL_DECL note,
>> > update get_call_fndecl to also check function symbol for function
>> > declaration so that it can be used on CA
On Thu, 29 May 2025 at 07:49, Tomasz Kamiński wrote:
>
> This patch adjust the passing of parameters for the move_only_function,
> copyable_function and function_ref. For types that are declared as being
> passed
> by value in signature template argument, they are passed by value to the
> invoke
On Mon, Jun 2, 2025 at 1:21 PM Tomasz Kamiński wrote:
> This patch fixes the handle multiple digits values for the month, day,
> weekday
> and hour, when used with the %m, %d, %e, %m, %u, %w, %H, and %D, %F
> specifiers.
> The values are now printed unmodified. This patch also fixes printing
> n
This patch fixes the handle multiple digits values for the month, day, weekday
and hour, when used with the %m, %d, %e, %m, %u, %w, %H, and %D, %F specifiers.
The values are now printed unmodified. This patch also fixes printing negative
year with %F, where the values was not padded to four digits
On Tue, 27 May 2025 at 13:36, Tomasz Kamiński wrote:
>
> From: Jonathan Wakely
>
> This patch implements C++26 std::polymorphic as specified in P3019 with
> amendment to move assignment from LWG 4251.
>
> The implementation always allocate stored object on the heap. The manager
> function (_M_man
Extend the binary op/UNSPEC_SEL combiner patterns from SVE_FULL_F/
SVE_FULL_F_B16B16 to SVE_F/SVE_F_B16B16, where the strictness value
is SVE_RELAXED_GP.
gcc/ChangeLog:
* config/aarch64/aarch64-sve.md (*cond__2_relaxed):
Extend from SVE_FULL_F_B16B16 to SVE_F_B16B16.
(*con
This patch extends the expander for fma, fnma, fms, and fnms to support
partial SVE FP modes.
We add the missing BF16 tests, which we can now trigger for having
implemented the conditional expander.
We also add tests for the 'merging with multiplicand' case, which this
expander canonicalizes (alb
Tested on x86_64-darwin, OK for trunk?
thanks
Iain
--- 8< ---
Using lookup_template_class () directly on the coroutine_handle identifier
fails in the reported test because the using TYPE_DECL is found. Fix this
by looking up the std::coroutine_handle template specifically and then
instantiating
This patch extends the expander for unconditional fma, fnma, fms, and
fnms, so that it supports partial SVE FP modes.
gcc/ChangeLog:
* config/aarch64/aarch64-sve.md (4): Extend from
SVE_FULL_F_B16B16 to SVE_F_B16B16. Use sve_fp_pred instead
of aarch64_ptrue_reg.
(
This series incrementally adds support for operations on unpacked vectors
of floating-point values. By "unpacked", we're referring to the in-register
layout of partial SVE vector modes. For example, the elements of a VNx4HF
are stored as:
... | X | HF | X | HF | X | HF | X | HF |
Where 'X' deno
Extend the ternary op/UNSPEC_SEL combiner patterns from SVE_FULL_F/
SVE_FULL_F_BF to SVE_F/SVE_F_BF, where the strictness value is
SVE_RELAXED_GP.
We can only reliably test the 'merging with the third input' (addend)
and 'independent value' patterns at this stage as the canocalisation that
reorder
Define new iterators for partial floating-point modes, and cover these
in some existing mode_attrs. This patch serves as a starting point for
a series that extends support for unpacked floating-point operations.
To differentiate between BFloat mode iterators that need to test
TARGET_SSVE_B16B16,
1 - 100 of 120 matches
Mail list logo
