Thanks for the duplicate ping. This is fine.
So this indeed solves the discrepancy between running SMS w/ and w/o debugging?
Please include a comment next to the code stating why it's important
not to create such deps.
You may also want to store the result of "DEP_PRO (dep)" in
src_ and use it twic
Roman, Andrey,
Sorry for the delayed response.
It would indeed be good to have SMS apply to more loop patterns, still
within the realm of *countable* loops. SMS was originally designed to
handle doloops, with a specific pattern controlling the loop, easily
identified and separable from the loop's
SMS changes are ok.
>* common.opt (fmodulo-sched-reg-pressure, -fmodulo-sched-verbose):
> New flags.
We should document what the different verbosity levels are, or
at-least their range.
Thanks,
Ayal.
On Tue, Jan 10, 2012 at 7:48 PM, Vladimir Makarov wrote:
> On 01/03/2012 04:25 AM, R
On Tue, Jan 10, 2012 at 12:31 PM, Revital1 Eres wrote:
>
> Hello,
>
> The patch below fixes ICE reported in PR51794.
> It avoids creating DDG edges for register uses of class DF_REF_ARTIFICIAL
> as
> the latter does not have real instructions for them and thus calling
> BLOCK_FOR_INSN fails.
>
>
On Fri, Feb 3, 2012 at 8:07 PM, Jakub Jelinek wrote:
> Hi!
>
> On some targets e.g. sms-7.c test fails, because fprintf is called
> with %s format and NULL argument, GLIBC prints for that e.g.
> SMS loop num: 1, file: (null), line: 0
> but it isn't portable. print-rtl.c guards the locator printin
On Mon, Jan 2, 2012 at 3:30 PM, Richard Sandiford
wrote:
> Ayal Zaks writes:
>> + for (i = 0; i < ira_pressure_classes_num; i++)
>> + {
>> + enum reg_class pressure_class;
>> +
>> + pressure_class = ira_pressure_classes[i];
>> +
>>
>The attached patch prevents the creation of reg-moves for definitions
>with MODE_CC and thus solves this ICE.
>
>Currently testing and bootstrap on ppc64-redhat-linux, enabling SMS on
>loops with SC 1.
>
>OK for 4.7 once testing completes?
Yes, thanks for catching this. Shouldn't we prevent creat
On Mon, Dec 19, 2011 at 4:28 PM, Richard Sandiford
wrote:
> Hi Revital,
>
> Revital Eres writes:
>> The attached patch is a resubmission following comments made by Ayal
>> and Richard.
>>
>> Tested and bootstrap with the other patches in the series on
>> ppc64-redhat-linux, enabling SMS on loops
sure, OK, thanks for catching this leak.
Ayal.
On Mon, Dec 12, 2011 at 8:25 AM, Revital Eres wrote:
> Hello,
>
>> OK for 3.7?
>
> Sorry, I meant GCC 4.7.0...
>
> Thanks,
> Revital
These fixes to individual sms testcases are OK.
Thanks,
Ayal.
On Mon, Dec 5, 2011 at 3:07 PM, Revital Eres wrote:
> Hello,
>
> Ping: http://gcc.gnu.org/ml/gcc-patches/2011-11/msg02444.html
>
> Thanks,
> Revital
On Mon, Nov 21, 2011 at 7:07 AM, Revital Eres wrote:
> Hello,
>
> This patch support the estimation of register pressure in SMS.
> Although GCC is in stage 3 I would appreciate comments on it.
> Thanks to Richard and Ayal for discussing the implementation and their
> insights.
>
> This part of th
On Thu, Oct 27, 2011 at 4:47 PM, Revital Eres wrote:
> Hello,
>
> The attach patch fixes the current marking of SMS loops to prevent
> further scheduling as follows: it marks *all* the loop's bbs with
> BB_DISABLE_SCHEDULE which prevents them from been scheduled later.
> (with the current implemen
On Fri, Sep 30, 2011 at 5:22 PM, Roman Zhuykov wrote:
> 2011/7/21 :
>> This patch should be applied only after pending patches by Revital.
>
>
> Ping. New version is attached, it suits current trunk without
> additional patches.
Thanks for the ping.
> Also this related patch needs approval:
> h
On Mon, Oct 10, 2011 at 1:57 PM, Richard Sandiford
wrote:
> Ayal Zaks writes:
>>> I agree it's natural to schedule moves for intra-iteration dependencies
>>> in the normal get_sched_window way. But suppose we have a dependency:
>>>
>>> A --(T,N,1)
On Wed, Sep 28, 2011 at 4:49 PM, Richard Sandiford
wrote:
> Ayal Zaks writes:
>>> >> + /* The cyclic lifetime of move->new_reg starts and ends at move->def
>>> >> + (the instruction that defines move->old_reg).
>>> >
>>> > So
On Wed, Sep 28, 2011 at 4:53 PM, Richard Sandiford
wrote:
> Ayal Zaks writes:
>>>> Only request is to document that the register moves are
>>>> placed/assigned-id's in a specific order.
>>>
>>>I suppose this is the downside of splitting the patc
On Fri, Sep 30, 2011 at 10:03 AM, Revital Eres wrote:
> Hello,
>
>> This
>> + /* Skip instructions that do not set a register. */
>> + if (set && !REG_P (SET_DEST (set)))
>> + continue;
>> is ok. Can you also prevent !set insns from having reg_moves? (To be updated
>> once auto_i
On Tue, Sep 27, 2011 at 9:47 AM, Revital Eres wrote:
> Hello,
>
>> ok, so if we have an auto-inc'ing insn which defines (auto-inc's) an
>> addr register and another (say, result) register, we want to allow the
>> result register to have life ranges in excess of ii (by eliminating
>> anti-dep edges
On Tue, Sep 27, 2011 at 10:47 AM, Revital Eres wrote:
> Hello,
>
>> This
>> + /* Skip instructions that do not set a register. */
>> + if (set && !REG_P (SET_DEST (set)))
>> + continue;
>> is ok. Can you also prevent !set insns from having reg_moves? (To be updated
>> once auto_i
On Mon, Sep 26, 2011 at 7:31 AM, Revital Eres wrote:
> Hello,
>
> This patch extends the implementation to support instructions with
> REG_INC notes.
> It addresses the comments from the previous submission:
> http://gcc.gnu.org/ml/gcc-patches/2011-08/msg01299.html.
>
ok, so if we have an auto-in
>OK for mainline?
Doh, hard to believe we never checked that an insn defines a register
before spitting out reg_moves for it ... nice catch.
This
+ /* Skip instructions that do not set a register. */
+ if (set && !REG_P (SET_DEST (set)))
+ continue;
is ok. Can you also prevent !
2011/9/23 Richard Sandiford
>
> Thanks as always for the review.
Sure, always a pleasure.
>
> Ayal Zaks writes:
> >> Richard Sandiford wrote on 30/08/2011
> >> 03:29:26 PM:
> >>
> >> > From: Richard Sandiford
> >> > T
> Richard Sandiford wrote on 30/08/2011
> 03:29:26 PM:
>
> > From: Richard Sandiford
> > To: gcc-patches@gcc.gnu.org
> > Cc: Ayal Zaks/Haifa/IBM@IBMIL
> > Date: 30/08/2011 03:29 PM
> > Subject: [4/4] Make SMS schedule register moves
> >
> > This
e = SCHED_FIRST_REG_MOVE (u) + (i_reg_moves - 1);
Ayal.
2011/9/22 Richard Sandiford
>
> Ayal Zaks writes:
> > Richard Sandiford wrote on 30/08/2011
> > 03:10:50 PM:
> >
> >> From: Richard Sandiford
> >> To: gcc-patches@gcc.gnu.org
> >&
Richard Sandiford wrote on 30/08/2011
03:10:50 PM:
> From: Richard Sandiford
> To: gcc-patches@gcc.gnu.org
> Cc: Ayal Zaks/Haifa/IBM@IBMIL
> Date: 30/08/2011 03:10 PM
> Subject: [3/4] SMS: Record moves in the partial schedule
>
> This patch adds infrastructure that will
2011/9/13 Richard Sandiford
>
> Ayal Zaks writes:
> > So instead of navigating directly from
> > ps_insn->ddg_node->node_sched_params, we now use indices and lookup
> > pointees in ddg_node and node_sched_params arrays. A bit of a
> > nuisance, but it&
Richard Sandiford wrote on 30/08/2011
03:03:59 PM:
> From: Richard Sandiford
> To: gcc-patches@gcc.gnu.org
> Cc: Ayal Zaks/Haifa/IBM@IBMIL
> Date: 30/08/2011 03:05 PM
> Subject: [2/4] SMS: Use ids to represent ps_insns
>
> Instructions in a partial schedule are curre
Copying the lists..
-- Forwarded message --
From: Ayal Zaks
Date: 2011/9/11
Subject: Re: [PATCH, SMS] Minor misc. fixes
To: Revital Eres revital.e...@linaro.org
2011/9/8 Revital Eres
>
> Hello,
>
> The attached patch contains minor fixes.
>
> Cur
Resending; didn't seem to go through.
-- Forwarded message --
From: Ayal Zaks
Date: 2011/9/11
Subject: gcc-patches@gcc.gnu.org
To: Richard Sandiford richard.sandif...@linaro.org
Richard Sandiford wrote on 30/08/2011
02:58:22 PM:
> From: Richard Sandiford
&
Ok, so this extends the infrastructure to support insns which set an
arbitrary number of registers, but currently specifically handles only
REG_INC situations (which set two registers). I'm not against
{0,1,infinity}, but wonder if this case really deserves the
complexity: post/pre-inc/decrementing
> [PATCH, SMS] Fix calculation of issue_rate
> http://gcc.gnu.org/ml/gcc-patches/2011-05/msg01344.html
This is ok (with the updated Changelog). Alternatively, we can have a
local variable for holding the issue_rate.
Ayal.
2011/7/20 Revital Eres :
> Hello,
>
> [PATCH, SMS 3/4] Optimize stage coun
>[PATCH, SMS 4/4] Misc. fixes
http://gcc.gnu.org/ml/gcc-patches/2011-05/msg01342.html
Sure, this is fine.
(Sorry for all the previous '?'s..).
Thanks,
Ayal.
2011/7/20 Revital Eres
>
> Hello,
>
> [PATCH, SMS 3/4] Optimize stage count
> http://gcc.gnu.org/ml/gcc-patches/2011-05/msg01341.html
>
> [P
> [PATCH, SMS 3/4] Optimize stage count
> http://gcc.gnu.org/ml/gcc-patches/2011-05/msg01341.html
This patch for minimizing the stage count (which also refactors and
cleans up the code) is approved. Have some minor comments below,
followed by some thoughts for possible follow-up improvements.
Than
Revital Eres wrote on 14/06/2011 09:27:32 AM:
> From: Revital Eres
> To: Ayal Zaks/Haifa/IBM@IBMIL
> Cc: gcc-patches@gcc.gnu.org, Patch Tracking
> Date: 14/06/2011 09:27 AM
> Subject: Re: [PATCH, SMS 1/4] Fix calculation of row_rest_count
>
> Hello,
>
> > Ple
Revital Eres wrote on 15/06/2011 11:45:15 AM:
> From: Revital Eres
> To: Ayal Zaks/Haifa/IBM@IBMIL
> Cc: gcc-patches@gcc.gnu.org, Patch Tracking
> Date: 15/06/2011 11:45 AM
> Subject: Re: [PATCH, SMS] Fix violation of memory dependence
>
> Hello,
>
> >>>
Revital Eres wrote on 13/06/2011 10:29:06 AM:
> From: Revital Eres
> To: Ayal Zaks/Haifa/IBM@IBMIL
> Cc: gcc-patches@gcc.gnu.org, Patch Tracking
> Date: 13/06/2011 10:29 AM
> Subject: [PATCH, SMS] Fix violation of memory dependence
>
> Hello,
>
> The attached patc
Revital Eres wrote on 19/05/2011 07:44:23 AM:
> From: Revital Eres
> To: Ayal Zaks/Haifa/IBM@IBMIL
> Cc: gcc-patches@gcc.gnu.org, Patch Tracking
> Date: 19/05/2011 07:44 AM
> Subject: [PATCH, SMS 2/4] Move the creation of anti-dep edge
>
> Hello,
>
> The attached
ial_schedule): Likewise.
> (create_ps_insn): Remove rest_count argument.
> (remove_node_from_ps): Update rows_length.
> (add_node_to_ps): Update rows_length and call create_ps_insn
> without passing row_rest_count.
> [attachment "patch_row_rest_count_17_5.txt" deleted by Ayal
Zaks/Haifa/IBM]
> OK for mainline?
Yes, this is pretty obvious. (You don't have to change to
prev_nondebug_insn btw).
Ayal.
From: Revital Eres
To: Ayal Zaks/Haifa/IBM@IBMIL
Cc: gcc-patches@gcc.gnu.org, Patch Tracking
Date: 08/05/2011 07:37 AM
Subject:[PATCH, SMS 2/3] Skip DEB
resting
to share comparative measurements when available.
Ayal.
Revital Eres wrote on 08/05/2011 07:37:07 AM:
> From: Revital Eres
> To: Ayal Zaks/Haifa/IBM@IBMIL
> Cc: gcc-patches@gcc.gnu.org, Patch Tracking
> Date: 08/05/2011 07:37 AM
> Subject: [PATCH, SMS 1/3] Support
40 matches
Mail list logo
