Hi Honza,
I experimented building Coremark with both PGO and LTO at -O3 level on
Aarch64 machine. First I generated profiles using the recommended
seeds in Coremark's readme.txt. Then compiled again with -O3 -flto and
-fprofile-use.
I tried using GCC Linaro compiler (september) which is based on
On Wed, 5 Nov 2014, Andrew Pinski wrote:
> Hi,
> I was trying to hook up tree-ssa-phiopt to match-and-simplify using
> either gimple_build (or rather using gimple_simplify depending on if
> we want to produce cond_expr for conditional move). I ran into a
> problem.
> With the pattern below:
> /
On Thu, Nov 6, 2014 at 8:02 AM, Tobias Grosser wrote:
> On 06.11.2014 07:04, Roman Gareev wrote:
>>>
>>> CLooG is not necessarily needed. You can run graphite just with ISL. The
>>> main reason that ISL code generation is not enabled by default is that we
>>> did not yet get extensive testing and
On 06.11.2014 10:05, Richard Biener wrote:
On Thu, Nov 6, 2014 at 8:02 AM, Tobias Grosser wrote:
On 06.11.2014 07:04, Roman Gareev wrote:
CLooG is not necessarily needed. You can run graphite just with ISL. The
main reason that ISL code generation is not enabled by default is that we
did not
> Hi Honza,
Hello,
>
> I experimented building Coremark with both PGO and LTO at -O3 level on
> Aarch64 machine. First I generated profiles using the recommended seeds in
> Coremark's readme.txt. Then compiled again with -O3 -flto and -fprofile-use.
>
> I tried using GCC Linaro compiler (septemb
On 11/6/14, Tobias Grosser wrote:
> On 06.11.2014 10:05, Richard Biener wrote:
>> On Thu, Nov 6, 2014 at 8:02 AM, Tobias Grosser wrote:
>>> On 06.11.2014 07:04, Roman Gareev wrote:
>
> CLooG is not necessarily needed. You can run graphite just with ISL.
> The
> main reason that IS
On 06.11.2014 11:15, Richard Biener wrote:
On 11/6/14, Tobias Grosser wrote:
On 06.11.2014 10:05, Richard Biener wrote:
On Thu, Nov 6, 2014 at 8:02 AM, Tobias Grosser wrote:
On 06.11.2014 07:04, Roman Gareev wrote:
CLooG is not necessarily needed. You can run graphite just with ISL.
The
ma
On 11/6/14, Tobias Grosser wrote:
> On 06.11.2014 11:15, Richard Biener wrote:
>> On 11/6/14, Tobias Grosser wrote:
>>> On 06.11.2014 10:05, Richard Biener wrote:
On Thu, Nov 6, 2014 at 8:02 AM, Tobias Grosser
wrote:
> On 06.11.2014 07:04, Roman Gareev wrote:
>>>
>>> CLooG
On Wed, 5 Nov 2014, Andrew Pinski wrote:
> Hi,
> I was trying to hook up tree-ssa-phiopt to match-and-simplify using
> either gimple_build (or rather using gimple_simplify depending on if
> we want to produce cond_expr for conditional move). I ran into a
> problem.
> With the pattern below:
> /
On Thu, 6 Nov 2014, Richard Biener wrote:
> On Wed, 5 Nov 2014, Andrew Pinski wrote:
>
> > Hi,
> > I was trying to hook up tree-ssa-phiopt to match-and-simplify using
> > either gimple_build (or rather using gimple_simplify depending on if
> > we want to produce cond_expr for conditional move).
On Thu, Nov 06, 2014 at 10:43:16AM +0100, Tobias Grosser wrote:
> is there any chance we can add the deprecated isl_int includes back into isl
> 0.14.1. This would unblock the testing and we could remove them as soon as
> gcc 4.8/4.9 has been phased out.
I prefer Richard's solution.
In any case, y
It appears to me that both loop invariant motion passes (tree/rtl) don't
look at basic block frequencies and will gladly hoist invariant code
from a cold block within a loop. This can impact performance by
executing (possibly costly) code that would otherwise not be executed,
adds another regis
On Thu, Nov 6, 2014 at 7:08 PM, Pat Haugen wrote:
> It appears to me that both loop invariant motion passes (tree/rtl) don't
> look at basic block frequencies and will gladly hoist invariant code from a
> cold block within a loop. This can impact performance by executing (possibly
> costly) code t
> As the ISL code generator has been default since a while and we did not get
> many bug reports, the actual switch seems to have worked well. We could
> probably still need some testing, but in this case it is most likely time to
> drop the CLooG support entirely. Are you interested to provide the
On 06.11.2014 20:08, Roman Gareev wrote:
As the ISL code generator has been default since a while and we did not get
many bug reports, the actual switch seems to have worked well. We could
probably still need some testing, but in this case it is most likely time to
drop the CLooG support entirely
On 11/06/2014 01:00 PM, Richard Biener wrote:
Shouldn't we never hoist anything from a bb with lower execution frequency
to a bb with higher one? It seems LIM simply assumes that inside a loop
is always higher frequency than outside of it.
So - why artificially have that factor of 0.1 instead o
Richard Biener wrote:
On Thu, Nov 6, 2014 at 8:02 AM, Tobias Grosser wrote:
Also, as Tobias suggested we should raise the minimal supported isl level to
0.14 to be sure PR 62289 is fixed.
As I am testing with system isl/cloog that would be unfortunate as it means
I'd either drop testing graphi
Snapshot gcc-4.8-20141106 is now available on
ftp://gcc.gnu.org/pub/gcc/snapshots/4.8-20141106/
and on various mirrors, see http://gcc.gnu.org/mirrors.html for details.
This snapshot has been generated from the GCC 4.8 SVN branch
with the following options: svn://gcc.gnu.org/svn/gcc/branches
That's what I assumed. However, can reload spill GPRs into FPRs as LRA
does? For even after specifying -mno-lra, I still see excessive slots in
FPRs.
Thank you,
--
Evandro Menezes Austin, TX
> -Original Message-
> From: Richard Earnshaw [mailto:rearn...@a
On 11/04/14 20:29, Bin.Cheng wrote:
Hi,
The parameter MAX_PENDING_LIST_LENGTH is set to 32 by default. It
seems to me the length of pending list can't be larger than 32. But
in sched-deps.c, below code is used:
/* Pending lists can't get larger with a readonly context. */
if (!de
On Thu, Nov 6, 2014 at 3:03 PM, Evandro Menezes wrote:
> That's what I assumed. However, can reload spill GPRs into FPRs as LRA
> does? For even after specifying -mno-lra, I still see excessive slots in
> FPRs.
Not fully. What is happening most likely is IRA is deciding to use
some FPRs for so
On Fri, Nov 7, 2014 at 7:10 AM, Jeff Law wrote:
> On 11/04/14 20:29, Bin.Cheng wrote:
>>
>> Hi,
>> The parameter MAX_PENDING_LIST_LENGTH is set to 32 by default. It
>> seems to me the length of pending list can't be larger than 32. But
>> in sched-deps.c, below code is used:
>>/* Pending
On Thu, Nov 6, 2014 at 2:40 AM, Richard Biener wrote:
> On Thu, 6 Nov 2014, Richard Biener wrote:
>
>> On Wed, 5 Nov 2014, Andrew Pinski wrote:
>>
>> > Hi,
>> > I was trying to hook up tree-ssa-phiopt to match-and-simplify using
>> > either gimple_build (or rather using gimple_simplify depending
On November 7, 2014 5:03:19 AM CET, Andrew Pinski wrote:
>On Thu, Nov 6, 2014 at 2:40 AM, Richard Biener
>wrote:
>> On Thu, 6 Nov 2014, Richard Biener wrote:
>>
>>> On Wed, 5 Nov 2014, Andrew Pinski wrote:
>>>
>>> > Hi,
>>> > I was trying to hook up tree-ssa-phiopt to match-and-simplify
>using
> On Nov 6, 2014, at 11:24 PM, Richard Biener wrote:
>
>> On November 7, 2014 5:03:19 AM CET, Andrew Pinski wrote:
>> On Thu, Nov 6, 2014 at 2:40 AM, Richard Biener
>> wrote:
>>> On Thu, 6 Nov 2014, Richard Biener wrote:
>>>
> On Wed, 5 Nov 2014, Andrew Pinski wrote:
>
> Hi,
>
25 matches
Mail list logo
