On Tue, Jul 28, 2015 at 3:27 AM, Jonathan Wakely wrote:
> On 27/07/15 19:40 -0700, Tim Shen wrote:
>>
>> Done by s/_M_add_collating_element/_M_add_collate_element/.
>
>
> Great, thanks. OK for trunk and gcc-5-branch.
Committed. Is there no need for gcc-4_9-branch? What's the general
policy for ba
-Original Message-
From: Richard Biener [mailto:richard.guent...@gmail.com]
Sent: Thursday, July 16, 2015 4:30 PM
To: Ajit Kumar Agarwal
Cc: l...@redhat.com; GCC Patches; Vinod Kathail; Shail Aditya Gupta; Vidhumouli
Hunsigida; Nagaraju Mekala
Subject: Re: [Patch,tree-optimization]: Add
Sorry about the slow response on IRC today, I got distracted onto
another issue and forgot to check back. What I started to write:
I'm exploring your suggestion to see if the back end could emit the
diagnostics. But I'm not sure it has sufficient context (location information)
to point to th
On Tue, Jul 28, 2015 at 09:24:17PM +0100, Richard Sandiford wrote:
> Trevor Saunders writes:
> > On Mon, Jul 27, 2015 at 09:05:08PM +0100, Richard Sandiford wrote:
> >> Alternatively we could have a new target_globals structure that is
> >> initialised with the result of calling the hook. If we d
Currently, #pragma GCC diagnostic is handled entirely by the FE. This
has several drawbacks:
* PR c++/53431 - C++ preprocessor ignores #pragma GCC diagnostic: The
C++ parser lexes (and preprocesses) before handling the pragmas.
* PR 53920 - "gcc -E" does not honor #pragma GCC diagnostic ignored
"
On Tue, Jul 28, 2015 at 3:10 PM, Richard Sandiford
wrote:
> Andrew Pinski writes:
>> On Tue, Jul 28, 2015 at 1:36 PM, Richard Sandiford
>> wrote:
>>> Bootstrapped & regression-tested on x86_64-linux-gnu and aarch64-linux-gnu.
>>> Also tested via config-list.mk. Committed as preapproved.
>>>
>>>
Andrew Pinski writes:
> On Tue, Jul 28, 2015 at 1:36 PM, Richard Sandiford
> wrote:
>> Bootstrapped & regression-tested on x86_64-linux-gnu and aarch64-linux-gnu.
>> Also tested via config-list.mk. Committed as preapproved.
>>
>> Thanks,
>> Richard
>>
>>
>> gcc/
>> * target-insns.def (at
On Tue, Jul 28, 2015 at 1:36 PM, Richard Sandiford
wrote:
> Bootstrapped & regression-tested on x86_64-linux-gnu and aarch64-linux-gnu.
> Also tested via config-list.mk. Committed as preapproved.
>
> Thanks,
> Richard
>
>
> gcc/
> * target-insns.def (atomic_test_and_set): New targetm inst
On Tue, Jul 28, 2015 at 1:35 PM, Richard Sandiford
wrote:
> Continuing after a break for the fr30 patch...
>
> Bootstrapped & regression-tested on x86_64-linux-gnu and aarch64-linux-gnu.
> Also tested via config-list.mk. Committed as preapproved.
>
> Thanks,
> Richard
>
>
> gcc/
> * targe
Bootstrapped & regression-tested on x86_64-linux-gnu and aarch64-linux-gnu.
Also tested via config-list.mk. Committed as preapproved.
Thanks,
Richard
gcc/
* target-insns.def (reload_load_address): New targetm instruction
pattern.
* reload1.c (gen_reload): Use it instead
Bootstrapped & regression-tested on x86_64-linux-gnu and aarch64-linux-gnu.
Also tested via config-list.mk. Committed as preapproved.
Thanks,
Richard
gcc/
* target-insns.def (atomic_test_and_set): New targetm instruction
pattern.
* optabs.c (maybe_emit_atomic_test_and_se
Bootstrapped & regression-tested on x86_64-linux-gnu and aarch64-linux-gnu.
Also tested via config-list.mk. Committed as preapproved.
Thanks,
Richard
gcc/
* target-insns.def (can_extend, ptr_extend): New targetm instruction
patterns.
* optabs.c (can_extend_p): Use them i
The thread_prologue_and_epilogue_insns code that used to be protected
by #ifdef HAVE_eh_return is unconditionally correct and isn't on any
kind of hot path (it's only run once per function).
Bootstrapped & regression-tested on x86_64-linux-gnu and aarch64-linux-gnu.
Also tested via config-list.mk.
Continuing after a break for the fr30 patch...
Bootstrapped & regression-tested on x86_64-linux-gnu and aarch64-linux-gnu.
Also tested via config-list.mk. Committed as preapproved.
Thanks,
Richard
gcc/
* target-insns.def (indirect_jump): New targetm instruction pattern.
* optab
"David Sherwood" writes:
> Hi,
>
> I have updated the comment above GET_MODE_INNER and while there I have
> fixed a spelling mistake in the comment above GET_MODE_UNIT_SIZE.
>
> Tested:
> aarch64 and aarch64_be - no regressions in gcc testsuite
> x86_64 - bootstrap build, no testsuite regressions
Trevor Saunders writes:
> On Mon, Jul 27, 2015 at 09:05:08PM +0100, Richard Sandiford wrote:
>> Alternatively we could have a new target_globals structure that is
>> initialised with the result of calling the hook. If we do that though,
>> it might make sense to consolidate the hooks rather than
2015-07-28 1:14 GMT+02:00 Kai Tietz :
> 2015-07-27 18:51 GMT+02:00 Jason Merrill :
>> I've trimmed this to the previously mentioned issues that still need to be
>> addressed; I'll do another full review after these are dealt with.
>
> Thanks for doing this summary of missing parts of prior review.
Hi!
> This is just an initial patch and tries to optimize integer type power 2
> constants. I wanted to get feedback on this . I bootstrapped and reg
> tested on aarch64-none-linux-gnu .
Thanks for working on it.
ChangeLog entry for the patch is missing, probably also some testcases.
> @@ -90,
I've committed this to the gomp4 branch. It cleans up the existing OACC_foo
constants defined in omp-low.h, replacing them with the ones I just defined in
gomp-constants.h
nathan
2015-07-28 Nathan Sidwell
* omp-low.h (enum oacc_loop_levels): Delete.
(OACC_LOOP_MASK): Delete.
* omp-low.c
On Tue, Jul 28, 2015 at 09:02:51AM -0700, Steve Ellcey wrote:
> Marek,
>
> I have run into a problem with this warning and building glibc.
>
> sysdeps/ieee754/s_matherr.c has:
>
> int
> weak_function
> __matherr(struct exception *x)
> {
> int n=0;
> if(x->arg1!=x->arg1) return 0;
måndag 20 juli 2015 16.32.01 skrev Magnus Granberg:
> > Patch updated and tested on x86_64-unknown-linux-gnu (Gentoo)
> >
> > Changlogs
> > /gcc
> > 2015-07-05 Magnus Granberg
> >
> > * common.opt (fstack-protector): Initialize to -1.
> > (fstack-protector-all): Likewise.
> >
On 21/04/15 06:27, Jeff Law wrote:
On 04/20/2015 01:09 AM, Shiva Chen wrote:
Hi, Jeff
Thanks for your advice.
can_replace_by.patch is the new patch to handle both cases.
pr43920-2.c.244r.jump2.ori is the original jump2 rtl dump
pr43920-2.c.244r.jump2.patch_can_replace_by is the jump2 rtl du
On Jul 27, 2015, at 7:36 PM, Eric Gallager wrote:
> On 7/27/15, Andreas Schwab wrote:
>> Eric Gallager writes:
>>
>>> Okay, I tried embedding "! -name CVS/ ! -name .svn/" into the find
>>
>> -name does an exact match, so you don't need the slash.
> Okay, attached a new version of the patch; m
On Jul 28, 2015, at 6:38 AM, Bruce Korb wrote:
> Definitely much better. I won't apply it until the weekend, so
> someone else will likely beat me to it.
Looks good to me as well, I checked it in.
Committed revision 226317.
[Richard wrote:]
Note the store to *pointer can be done unconditionally
Yes; if I`m mapping things correctly in my mind, this is
something that Sebastian [and Alan, via email?] and I have
already discussed and which we plan to fix in good time.
Please note that this is a minor problem at most,
Hi,
Following last patch, this patch patch prevents arm_thumb1 XPASS in
sibcall-3.c and sibcall-4.c by skipping on arm_thumb1 and arm_thumb2
respectively.
This patch also documents arm_thumb1 and arm_thumb2 effective target
options.
Is patch ok for trunk and fsf-5?
gcc/testsuite
2015-07-28
I believe the following patch fixes a problem with bootstrap failures
on some architectures with vtable verification enabled. The problem
was related to a change in name mangling, where classes with anonymous
namespaces get "" as their DECL_ASSEMBLER_NAME, rather than the
real mangled name. This
Oh, one more thing. I placed constants for the 3 launch dimensions into
gomp-constants.h, as they are needed by both library and compiler. Working on a
patch to remove the current set of constants from omp-low.h
nathan
I've committed this patch to the gomp4 branch to redo the launch API. I'll post
a version for trunk once the versioning patch gets approved & committed.
This changes the API in a number of ways, allowing device-specific knowledge to
be moved into the device compiler and out of the host compile
This patch implements a MIPS o32 ABI specific optimization called frame
header optimization. In the o32 ABI, routines allocate 16 bytes on the
stack before calling another routine. This space is used by the callee
as space to write the register arguments to if their address is taken.
The n32 and
Marek,
I have run into a problem with this warning and building glibc.
sysdeps/ieee754/s_matherr.c has:
int
weak_function
__matherr(struct exception *x)
{
int n=0;
if(x->arg1!=x->arg1) return 0;
return n;
}
And arg1 is a floating point type. I think that if the value o
Hi all,
This is a simple patch to make arm_align_max_stack_pwr.c and
arm_align_max_pwr.c compile test cases, instead of execution tests.
In my local machine, those test cases pass. However, they fail on some
systems with process memory usage restrictions. Anyway, the required
space for those
gcc/ChangeLog
2015-07-27 Martin Sebor
* c-family/c.opt (-Wbuiltin-address): New warning option.
* doc/invoke.texi (Wbuiltin-address): Document it.
* doc/extend.texi (__builtin_frame_addrress, __builtin_return_addrress):
Typoes (rr).
Fixed.
- rtx tem
- = expand
Applied the idea with python script alternative. Review, please.
2015-07-24 17:18 GMT+03:00 David Malcolm :
> On Fri, 2015-07-24 at 10:01 +0200, Jakub Jelinek wrote:
>> #!/usr/bin/python
>> import sys
>> with open(sys.argv[1],"rb") as f:
>> nextblock = f.read(12)
>> while 1:
>> bl
On 07/28/2015 02:21 AM, Thomas Schwinge wrote:
> Cesar, please address the following compiler diagnostig:
>
>> 2015-07-21 Cesar Philippidis
>>
>> gcc/
>> * omp-low.c (execute_oacc_transform): New function.
>> (class pass_oacc_transform): New function.
>> (make_pass_oacc_tra
On 26/07/15 13:38 -0700, Tim Shen wrote:
On Sat, Jul 25, 2015 at 8:31 AM, Jonathan Wakely wrote:
On 25/07/15 00:11 -0700, Tim Shen wrote:
It's not a very necessary refactoring, but simply can't resist. :)
I'm not sure of the ::memcpy calls. It looks not very idiomatic, but
std::copy on char*
Hi Richard,
For Aarch64 target, I was trying to vectorize the expression
"arr[i]=arr[i]*4;" via vector shifts instructions since they don't have
vector mults.
unsigned long int __attribute__ ((aligned (64)))arr[100];
int i;
#if 1
void test_vector_shifts()
{
for(i=0; i<=99;i++)
On Mon, Jul 27, 2015 at 09:08:34PM -0600, Martin Sebor wrote:
> >>So, my suggestion would be to warn for any call with a nonzero value.
> >
> >The current documentation says that you should only use nonzero values
> >for debug purposes. A warning would help yes, how many people read the
> >manual
On 07/28/15 08:30, Thomas Schwinge wrote:
I do agree that a copy clause is wrong (expected to fail at runtime), but
why do you say an implicit copy clause is created? OpenACC 2.0a, 2.5.1
Parallel Construct, says that »[...] An array or variable of aggregate
data type referenced in the parallel
> On Jul 28, 2015, at 3:25 AM, Kyrill Tkachov wrote:
>
> Hi all,
>
> Currently we assign the wrong rtx cost to instructions of the form
> add x0, x0, x1, sxtw
>
> that is, an arith operation plus a single extend (no shifting).
> We correctly catch the cases where the extend is inside a sh
Definitely much better. I won't apply it until the weekend, so
someone else will likely beat me to it. Thank you.
On Mon, Jul 27, 2015 at 7:36 PM, Eric Gallager wrote:
> On 7/27/15, Andreas Schwab wrote:
>> Eric Gallager writes:
>>
>>> Okay, I tried embedding "! -name CVS/ ! -name .svn/" into
The instruction sequences for preparing argument for TLS descriptor
runtime resolver and the later function call to resolver can actually be
hoisted out of the loop.
Currently we can't because we have exposed the hard register X0 as
destination of "set". While GCC's RTL data flow infrastructure
On Tue, Jul 28, 2015 at 03:04:52PM +0200, Mikael Morin wrote:
> Le 21/07/2015 21:08, Steve Kargl a ?crit :
> > When C++ was injected into trans-expr.c in the form of vec,
> > it seems whomever did the conversion to vec forgot to check
> > for a NULL C++ thing. This patch seems to avoid the problem
This fixes a problem where on ARM ubsan can introduce an uninitialized variable.
It's ARM only since the ARM C++ ABI says that when creating a pointer to member
function, the LSB of ptr discriminates between the address of a non-virtual
member
function and the offset in the class's virtual table o
Le 21/07/2015 21:08, Steve Kargl a écrit :
When C++ was injected into trans-expr.c in the form of vec,
it seems whomever did the conversion to vec forgot to check
for a NULL C++ thing. This patch seems to avoid the problem,
but having zero knowledge of C++ I could be wrong.
OK for trunk?
2015-
Hi Nathan!
On Tue, 28 Jul 2015 08:19:17 -0400, Nathan Sidwell
wrote:
> On 07/28/15 06:14, Thomas Schwinge wrote:
> > On Sat, 25 Jul 2015 16:02:01 -0400, Nathan Sidwell wrote:
> >> I've committed this to gomp4 branch. It fixes some tests that were
> >> incorrect
> >
> > Hmm, I fail to see what
On 07/28/15 06:14, Thomas Schwinge wrote:
Hi Nathan!
On Sat, 25 Jul 2015 16:02:01 -0400, Nathan Sidwell wrote:
I've committed this to gomp4 branch. It fixes some tests that were incorrect
Hmm, I fail to see what you deem incorrect in the following two Fortran
test cases? Implicit present_o
On 28/07/15 09:59, Richard Biener wrote:
On Fri, Jul 24, 2015 at 4:39 PM, Tom de Vries wrote:
Hi,
this patch allows parallelization and vectorization of reduction operators
that are guaranteed to not overflow (such as min and max operators),
independent of the overflow behaviour of the type.
On 27/07/15 05:38, Andreas Schwab wrote:
> Kugan writes:
>
>> * cse.c (cse_insn): Fix missing check for STRICT_LOW_PART and minor
>> clean up.
>
> This breaks
>
> gcc.target/m68k/tls-ie-xgot.c scan-assembler jsr __m68k_read_tp
> gcc.target/m68k/tls-ie.c scan-assembler jsr __m68k_re
This implements some remaining parts of fold_comparison address
comparisons but still no complete part of it. Still it is good
enough to make fold_stmt not regress when not dispatching to fold_binary.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied to trunk.
Richard.
2015-07-28 R
For types with just two values max - 1 is equal to min and thus
we fail to optimize some cases of comparisons. With fold-const.c
bool < 0 needed the abs(x) < 0 simplification to trigger it
(same issue with mis-ordered if / else-ifs).
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
On 07/27/15 11:21, Tom de Vries wrote:
On 26/07/15 19:09, Nathan Sidwell wrote:
I've committed this update to my earlier breakout of default handling.
After complaining about something because of 'none', we should fall
through to the default handling, to prevent ICEing later (on patch
seriesI'm
This documents the change to arm_neon_fp16_ok in the first patch; the addition
of arm_neon_fp16_hw_ok in the last patch; and corrects a cross-reference.
(I tried using an @ref instead of "Implies previous." but the page ref looked
very out-of-place in PDF when I am referring to the previous ite
This is a respin of https://gcc.gnu.org/ml/gcc-patches/2015-07/msg00488.html,
fixing up the testsuite for float16 vectors. Relative to the previous version,
most of the additions to the tests are now within #if..#endif such that they are
only compiled if we have a scalar __fp16 type (the excepti
The fix here (as noted https://gcc.gnu.org/ml/gcc-patches/2015-04/msg01345.html)
is that this changes the vector indices present in the RTL on bigendian for
float vec_unpacks, to be the same as for integer vec_unpacks. This appears
consistent with the usage of VEC_UNPACK_(FLOAT_)?EXPR in tree-ve
gcc/testsuite/ChangeLog:
* gcc.target/aarch64/advsimd-intrinsics/advsimd-intrinsics.exp:
set additional flags for neon-fp16 support.
* gcc.target/aarch64/advsimd-intrinsics/vcvt_f16.c: New.
commit e6cc7467ddf5702d3a122b8ac4163621d0164b37
Author: Alan Lawrence
Date: Wed
gcc/ChangeLog:
* config/aarch64/arm_neon.h (vreinterpret_p8_f16, vreinterpret_p16_f16,
vreinterpret_f16_f64, vreinterpret_f16_s8, vreinterpret_f16_s16,
vreinterpret_f16_s32, vreinterpret_f16_s64, vreinterpret_f16_f32,
vreinterpret_f16_u8, vreinterpret_f16_u16, vrei
gcc/ChangeLog:
* config/aarch64/aarch64-simd.md (aarch64_float_truncate_lo_v2sf):
Reparameterize to...
(aarch64_float_truncate_lo_): ...this, for both V2SF and V4HF.
(aarch64_float_truncate_hi_v4sf): Reparameterize to...
(aarch64_float_truncate_hi_): ...thi
gcc/ChangeLog:
* config/aarch64/aarch64.c (aarch64_vector_mode_supported_p): Support
V4HFmode and V8HFmode.
(aarch64_split_simd_move): Add case for V8HFmode.
* config/aarch64/aarch64-builtins.c (v4hf_UP, v8hf_UP): Define.
(aarch64_simd_builtin_std_type): Ha
gcc/ChangeLog:
* config/aarch64/aarch64.c (aarch64_split_simd_combine): Add V4HFmode.
* config/aarch64/aarch64-builtins.c (VAR13, VAR14): New.
(aarch64_scalar_builtin_types, aarch64_init_simd_builtin_scalar_types):
Add __builtin_aarch64_simd_hf.
* config/aa
gcc/testsuite/ChangeLog:
* gcc.target/aarch64/fp16/fp16.exp: New.
* gcc.target/aarch64/fp16/f16_convs_1.c: New.
* gcc.target/aarch64/fp16/f16_convs_2.c: New.
commit bc5045c0d3dd34b8cb94910281384f9ab9880325
Author: Alan Lawrence
Date: Thu May 7 10:08:12 2015 +0100
(
gcc/ChangeLog:
* config/aarch64/aarch64-builtins.c (aarch64_fp16_type_node): New.
(aarch64_init_builtins): Make aarch64_fp16_type_node, use for __fp16.
* config/aarch64/aarch64-modes.def: Add HFmode.
* config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define
This is a respin of https://gcc.gnu.org/ml/gcc-patches/2015-07/msg00477.html.
The only change is to publish float16x8_t only if we actually have a scalar
__fp16 type.
gcc/ChangeLog:
* config/arm/arm.h (VALID_NEON_QREG_MODE): Add V8HFmode.
* config/arm/arm.c (arm_vector_mode_supported_
This is a respin of https://gcc.gnu.org/ml/gcc-patches/2015-07/msg00478.html ,
again making the intrinsics available only if we have a scalar __fp16 type.
(This covers the intrinsics whose implementation is entirely within arm_neon.h;
those requiring .md changes follow in the next patch).
gcc/
This is a respin of https://gcc.gnu.org/ml/gcc-patches/2015-07/msg00479.html,
again to make the intrinsics available only if we have a scalar __fp16 type.
This does not fix existing indentation issues in neon.md but rather keeps the
affected lines consistent with those around them.
gcc/Change
This is a respin of https://gcc.gnu.org/ml/gcc-patches/2015-07/msg00476.html.
The change is to provide all the new float16 intrinsics only if we actually have
an scalar __fp16 type. (This covers the intrinsics whose implementation is
entirely within arm_neon.h; those requiring .md changes follow
This makes the existing float16 vector intrinsics available only when we have an
__fp16 type (i.e. when one of the ARM_FP16_FORMAT_... macros is defined).
Thus, we also rearrange the float16x[48]_t types to use the same type as __fp16
for the element type (ACLE says that __fp16 should be an ali
All AArch64 patches are unchanged from previous version. However, in response to
discussion, the ARM patches are changed (much as I suggested
https://gcc.gnu.org/ml/gcc-patches/2015-07/msg02249.html); this version:
* Hides the existing vcvt_f16_f32 and vcvt_f32_f16 intrinsics, and
float16x4
Hi,
I have updated the comment above GET_MODE_INNER and while there I have
fixed a spelling mistake in the comment above GET_MODE_UNIT_SIZE.
Tested:
aarch64 and aarch64_be - no regressions in gcc testsuite
x86_64 - bootstrap build, no testsuite regressions
arm-none-eabi - no regressions in gcc te
Ping.
This is a pre-requisite for:
https://gcc.gnu.org/ml/gcc-patches/2015-07/msg02347.html
On 22/07/15 14:56, Sebastian Huber wrote:
Move the thread pool initialization from the team start to the team
creation. This eliminates one conditional expression. In addition this
is a preparation pa
libgomp/ChangeLog
2015-07-28 Sebastian Huber
* config/posix/pool.h (gomp_adjust_thread_attr): New.
* config/rtems/pool.h (gomp_adjust_thread_attr): Likewise.
(gomp_thread_pool_reservoir): Add priority member.
* confi/rtems/proc.c (allocate_thread_pool_reservoir):
In RTEMS we may have multiple scheduler instances with different
scheduling algorithms. In addition we have a single process environment
so all threads run in one address space. In order to support work
stealing applications it is important to limit the number of thread
pools used for OpenMP sinc
libgomp/ChangeLog
2015-07-28 Sebastian Huber
* config/rtems/bar.c: New.
* config/rtems/bar.h: Likewise.
* config/rtems/mutex.c: Likewise.
* config/rtems/mutex.h: Likewise.
* config/rtems/sem.c: Likewise.
* config/rtems/sem.h: Likewise.
* c
On Tue, Jun 2, 2015 at 1:14 AM, Kugan wrote:
>
Sorry for replying so late...
> On 08/05/15 22:48, Richard Biener wrote:
>> You compute which promotions are unsafe, like sources/sinks of memory
>> (I think you miss call arguments/return values and also asm operands here).
>> But instead of simply
On 28/07/15 11:25, Kyrill Tkachov wrote:
> Hi all,
>
> Currently we assign the wrong rtx cost to instructions of the form
>add x0, x0, x1, sxtw
>
> that is, an arith operation plus a single extend (no shifting).
> We correctly catch the cases where the extend is inside a shift, but
> not
On Thu, Jul 23, 2015 at 4:45 PM, Yuri Rumyantsev wrote:
> Hi Richard,
>
> I checked that both test-cases from 23855 are sucessfully unswitched
> by proposed patch. I understand that it does not catch deeper loop
> nest as
>for (i=0; i<10; i++)
> for (j=0;j for (k=0;k<20;k++)
> .
On 27/07/15 19:40 -0700, Tim Shen wrote:
Done by s/_M_add_collating_element/_M_add_collate_element/.
Great, thanks. OK for trunk and gcc-5-branch.
Hi all,
Currently we assign the wrong rtx cost to instructions of the form
add x0, x0, x1, sxtw
that is, an arith operation plus a single extend (no shifting).
We correctly catch the cases where the extend is inside a shift, but
not the simple case.
This patch fixes that oversight by catc
On Fri, Jul 17, 2015 at 10:12 PM, Abe wrote:
> Dear all,
>
> Another benefit of the new if converter that perhaps I neglected to
> mention/explain...
>
> TLDR: for some source code which can reasonably be expected to exist in
> "real-world code",
> when the false/true values of the condition bits
On 24/07/15 12:27, Kyrill Tkachov wrote:
On 24/07/15 12:10, Szabolcs Nagy wrote:
(-a)*b should not be compiled to vnmul a,b with -frounding-math.
Added a new -(a*b) pattern for vnmul and the old one is only
used if !flag_rounding_math. Updated the costs too.
This is the ARM version of
https://
Hi Jeff,
On 27/07/15 17:40, Kyrill Tkachov wrote:
On 27/07/15 17:09, Jeff Law wrote:
On 07/27/2015 04:17 AM, Kyrill Tkachov wrote:
I experimented with resource.c and the roadblock I hit is that it
seems to have an assumption that it operates on hard regs (in fact
the struct it uses to describe
Hi Nathan!
On Sat, 25 Jul 2015 16:02:01 -0400, Nathan Sidwell wrote:
> I've committed this to gomp4 branch. It fixes some tests that were incorrect
Hmm, I fail to see what you deem incorrect in the following two Fortran
test cases? Implicit present_or_copy clauses should be added by the
compi
On Fri, Jul 24, 2015 at 12:10 PM, Tom de Vries wrote:
> On 20/07/15 15:04, Tom de Vries wrote:
>>
>> On 16/07/15 12:15, Richard Biener wrote:
>>>
>>> On Thu, Jul 16, 2015 at 11:39 AM, Tom de Vries
>>> wrote:
On 16/07/15 10:44, Richard Biener wrote:
>
>
> On Wed, Jul 15, 2015
Hi,
For now, SCEV may compute iv base in the form of "(signed T)((unsigned
T)base + step))". This complicates other optimizations/analysis depending
on SCEV because it's hard to dive into type conversions. For many cases,
such type conversions can be simplified with additional range information
i
Hi,
Loop niter computes inaccurate bound information for different loops. This
patch is to improve it by using loop initial condition in
determine_value_range. Generally, loop niter is computed by subtracting
start var from end var in loop exit condition. Moreover, loop bound is
computed using v
>
> On 07/27/2015 04:25 AM, David Sherwood wrote:
> > Hi,
> >
> > Part 1 of this change is a clean-up. I have changed calls to GET_MODE_INNER
> > (m)
> > so that it returns m in cases where there is no inner mode. This simplifies
> > some
> > of the calling code by removing the need to check for
Hi!
On Tue, 21 Jul 2015 10:15:05 -0700, Cesar Philippidis
wrote:
> Jakub,
>
> Nathan pointed out that I should make the fold_oacc_reductions pass that
> I introduced in my reduction patch more generic so that other openacc
> transformations may use it. This patch introduces an empty skeleton pa
Hi,
This is an obvious change to bypass following iterations if expr has already
been simplified into constant in function
simplify_using_loop_initial_conditions.
Is it OK?
Thanks,
bin
2015-07-28 Bin Cheng
* tree-ssa-loop-niter.c (simplify_using_initial_conditions): Break
Loo
Hi,
other bits. Tested x86_64-linux.
Thanks,
Paolo.
/
2015-07-28 Paolo Carlini
* call.c (build_op_delete_call, convert_like_real, build_over_call):
Use Use DECL_SOURCE_LOCATION and "%qD" in inform and pedwarn instead
of "%q+D".
* constexpr
Hi,
this patch adds a test-case with a double reduction in an oacc kernels
region.
In order to get it in the proper shape for parloops to deal with, I
needed to repeat the pass_lim/pass_copy_prop sequence.
Bootstrapped and reg-tested on x86_64.
Committed to gomp-4_0-branch.
Thanks,
- Tom
Hi,
I couldn't reproduce the need for pass_fre in the oacc kernels pass
group, so I've replaced it with an instance of pass_copy_prop.
Bootstrapped and reg-tested on x86_64.
Committed to gomp-4_0-branch.
Thanks,
- Tom
Replace pass_fre with pass_copy_prop in kernels pass group
2015-07-28 To
Hi
I've backported these parloops-related patches to gomp-4_0-branch.
- Simplify gather_scalar_reductions
- Update outer-4.c and uns-outer-4.c
- Handle double reduction in parloops
Thanks,
- Tom
Simplify gather_scalar_reductions
2015-07-27 Tom de Vries
backport from trunk:
2015-07-27 To
Hi DJ,
There is no need to assert these just to say "not supported" and gcc
may rarely generate addresses from valid code which trigger these
asserts. Ok?
OK - please apply.
Cheers
Nick
On Fri, Jul 24, 2015 at 4:39 PM, Tom de Vries wrote:
> Hi,
>
> this patch allows parallelization and vectorization of reduction operators
> that are guaranteed to not overflow (such as min and max operators),
> independent of the overflow behaviour of the type.
>
> Bootstrapped and reg-tested on x
On Tue, Jul 28, 2015 at 12:32 AM, Tom de Vries wrote:
> On 24/07/15 12:30, Tom de Vries wrote:
>>
>> On 13/07/15 16:55, Tom de Vries wrote:
>>>
>>> Hi,
>>>
>>> this patch fixes PR66851.
>>>
>>> In parloops, we manage to parallelize outer loops, but not if the inner
>>> loop contains a reduction. T
On Mon, Jul 27, 2015 at 6:20 PM, Jeff Law wrote:
> On 07/27/2015 03:25 AM, Richard Biener wrote:
>>
>> On Mon, Jul 27, 2015 at 5:41 AM, Michael Collison
>> wrote:
>>>
>>> This patch is designed to optimize end of loop conditions involving of
>>> the
>>> form
>>> i < x && i < y into i < min (x,
On 26/07/15 18:53, Tom de Vries wrote:
On 26/07/15 18:49, Tom de Vries wrote:
On 24/07/15 16:39, Tom de Vries wrote:
Hi,
this patch allows parallelization and vectorization of reduction
operators that are guaranteed to not overflow (such as min and max
operators), independent of the overflow b
On 24/07/15 16:39, Tom de Vries wrote:
Hi,
this patch allows parallelization and vectorization of reduction
operators that are guaranteed to not overflow (such as min and max
operators), independent of the overflow behaviour of the type.
Bootstrapped and reg-tested on x86_64.
OK for trunk?
On 22/07/15 20:15, Tom de Vries wrote:
On 13/07/15 13:02, Tom de Vries wrote:
Hi,
this patch fixes PR46193.
It handles min and max reductions of pointer type in parloops.
Bootstrapped and reg-tested on x86_64.
OK for trunk?
Ping.
Committed to gomp-4_0-branch.
Thanks,
- Tom
0001-Hand
This moves/merges the equality folding of decl addresses from
fold_comparison with that from fold_binary in match.pd.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied to trunk.
Richard.
2015-07-28 Richard Biener
* fold-const.c (fold_comparison): Remove equality folding
1 - 100 of 102 matches
Mail list logo
