public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug rtl-optimization/34826]  New: branch probability is not reserved with do-loop optimization
@ 2008-01-17  7:21 eres at il dot ibm dot com
  2008-01-17  9:20 ` [Bug rtl-optimization/34826] " eres at il dot ibm dot com
                   ` (4 more replies)
  0 siblings, 5 replies; 6+ messages in thread
From: eres at il dot ibm dot com @ 2008-01-17  7:21 UTC (permalink / raw)
  To: gcc-bugs

Running the attached testcase with trunk r131577 on SPU
with -dm -fmodulo-sched -O2 flags (which enables also do-loop optimization)
demonstartes the fact that the branch probability is not reserved with do-loop
optimization.
This causes a major preformance regression (of about 40%) when tested with some
similar benchmark as there is no hardware branch prediction on SPU.


-- 
           Summary: branch probability is not reserved with do-loop
                    optimization
           Product: gcc
           Version: 4.3.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: rtl-optimization
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: eres at il dot ibm dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34826


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [Bug rtl-optimization/34826] branch probability is not reserved with do-loop optimization
  2008-01-17  7:21 [Bug rtl-optimization/34826] New: branch probability is not reserved with do-loop optimization eres at il dot ibm dot com
@ 2008-01-17  9:20 ` eres at il dot ibm dot com
  2008-01-17  9:33 ` steven at gcc dot gnu dot org
                   ` (3 subsequent siblings)
  4 siblings, 0 replies; 6+ messages in thread
From: eres at il dot ibm dot com @ 2008-01-17  9:20 UTC (permalink / raw)
  To: gcc-bugs



------- Comment #1 from eres at il dot ibm dot com  2008-01-17 07:21 -------
Created an attachment (id=14955)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14955&action=view)
The testcase


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34826


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [Bug rtl-optimization/34826] branch probability is not reserved with do-loop optimization
  2008-01-17  7:21 [Bug rtl-optimization/34826] New: branch probability is not reserved with do-loop optimization eres at il dot ibm dot com
  2008-01-17  9:20 ` [Bug rtl-optimization/34826] " eres at il dot ibm dot com
@ 2008-01-17  9:33 ` steven at gcc dot gnu dot org
  2008-01-17  9:53 ` [Bug rtl-optimization/34826] branch probability is not updated " eres at il dot ibm dot com
                   ` (2 subsequent siblings)
  4 siblings, 0 replies; 6+ messages in thread
From: steven at gcc dot gnu dot org @ 2008-01-17  9:33 UTC (permalink / raw)
  To: gcc-bugs



------- Comment #2 from steven at gcc dot gnu dot org  2008-01-17 07:31 -------
Can you show in an RTL dump why you are sure that the branch probabilities are
lost?


-- 

steven at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rakdver at gcc dot gnu dot
                   |                            |org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34826


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [Bug rtl-optimization/34826] branch probability is not updated with do-loop optimization
  2008-01-17  7:21 [Bug rtl-optimization/34826] New: branch probability is not reserved with do-loop optimization eres at il dot ibm dot com
  2008-01-17  9:20 ` [Bug rtl-optimization/34826] " eres at il dot ibm dot com
  2008-01-17  9:33 ` steven at gcc dot gnu dot org
@ 2008-01-17  9:53 ` eres at il dot ibm dot com
  2008-01-17 18:43 ` eres at il dot ibm dot com
  2008-01-20  6:02 ` pinskia at gcc dot gnu dot org
  4 siblings, 0 replies; 6+ messages in thread
From: eres at il dot ibm dot com @ 2008-01-17  9:53 UTC (permalink / raw)
  To: gcc-bugs



------- Comment #3 from eres at il dot ibm dot com  2008-01-17 08:17 -------
(In reply to comment #2)
> Can you show in an RTL dump why you are sure that the branch probabilities are
> lost?

Sure, I am currently testing a patch we have to fix this problem,
the following '-' lines are from the SMS dump generated with current trunk and
the '+' lines are generated from trunk containing the fix.

@@ -1955,7 +1955,7 @@
         (if_then_else (ne:SI (reg:SI 204 [ n ])
                 (const_int 0 [0x0]))
             (label_ref:SI 145)
-            (pc))) 356 {*spu.md:3450} (expr_list:REG_BR_PROB (const_int 5000
[0x1388])
+            (pc))) 356 {*spu.md:3450} (expr_list:REG_BR_PROB (const_int 9100
[0x238c])
         (nil)))
 ;; End of basic block 8 -> ( 10 9)
 ;; lr  out      1 [$sp] 127 [$127] 128 [$vfp] 129 [$vap] 176 177 178 179 180
181 183 201 204
@@ -2015,8 +2015,8 @@
 ;;      reg 204 { d110(bb 8 insn 104) }


-;; Succ edge  10 [50.0%]  (dfs_back)
-;; Succ edge  9 [50.0%]  (fallthru,loop_exit)
+;; Succ edge  10 [91.0%]  (dfs_back)
+;; Succ edge  9 [9.0%]  (fallthru,loop_exit)


-- 

eres at il dot ibm dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|branch probability is not   |branch probability is not
                   |reserved with do-loop       |updated with do-loop
                   |optimization                |optimization


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34826


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [Bug rtl-optimization/34826] branch probability is not updated with do-loop optimization
  2008-01-17  7:21 [Bug rtl-optimization/34826] New: branch probability is not reserved with do-loop optimization eres at il dot ibm dot com
                   ` (2 preceding siblings ...)
  2008-01-17  9:53 ` [Bug rtl-optimization/34826] branch probability is not updated " eres at il dot ibm dot com
@ 2008-01-17 18:43 ` eres at il dot ibm dot com
  2008-01-20  6:02 ` pinskia at gcc dot gnu dot org
  4 siblings, 0 replies; 6+ messages in thread
From: eres at il dot ibm dot com @ 2008-01-17 18:43 UTC (permalink / raw)
  To: gcc-bugs



------- Comment #4 from eres at il dot ibm dot com  2008-01-17 18:28 -------
A patch was committed to trunk -r131604


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34826


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [Bug rtl-optimization/34826] branch probability is not updated with do-loop optimization
  2008-01-17  7:21 [Bug rtl-optimization/34826] New: branch probability is not reserved with do-loop optimization eres at il dot ibm dot com
                   ` (3 preceding siblings ...)
  2008-01-17 18:43 ` eres at il dot ibm dot com
@ 2008-01-20  6:02 ` pinskia at gcc dot gnu dot org
  4 siblings, 0 replies; 6+ messages in thread
From: pinskia at gcc dot gnu dot org @ 2008-01-20  6:02 UTC (permalink / raw)
  To: gcc-bugs



------- Comment #5 from pinskia at gcc dot gnu dot org  2008-01-20 04:52 -------
Fixed.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |FIXED
   Target Milestone|---                         |4.3.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34826


^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2008-01-20  4:53 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2008-01-17  7:21 [Bug rtl-optimization/34826] New: branch probability is not reserved with do-loop optimization eres at il dot ibm dot com
2008-01-17  9:20 ` [Bug rtl-optimization/34826] " eres at il dot ibm dot com
2008-01-17  9:33 ` steven at gcc dot gnu dot org
2008-01-17  9:53 ` [Bug rtl-optimization/34826] branch probability is not updated " eres at il dot ibm dot com
2008-01-17 18:43 ` eres at il dot ibm dot com
2008-01-20  6:02 ` pinskia at gcc dot gnu dot org

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).