* [Bug ipa/92535] [10/11 regression] ICF is relatively expensive and became less effective
[not found] <bug-92535-4@http.gcc.gnu.org/bugzilla/>
@ 2020-05-07 11:56 ` jakub at gcc dot gnu.org
2020-07-23 6:51 ` rguenth at gcc dot gnu.org
` (14 subsequent siblings)
15 siblings, 0 replies; 16+ messages in thread
From: jakub at gcc dot gnu.org @ 2020-05-07 11:56 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92535
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Target Milestone|10.0 |10.2
--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
GCC 10.1 has been released.
^ permalink raw reply [flat|nested] 16+ messages in thread* [Bug ipa/92535] [10/11 regression] ICF is relatively expensive and became less effective
[not found] <bug-92535-4@http.gcc.gnu.org/bugzilla/>
2020-05-07 11:56 ` [Bug ipa/92535] [10/11 regression] ICF is relatively expensive and became less effective jakub at gcc dot gnu.org
@ 2020-07-23 6:51 ` rguenth at gcc dot gnu.org
2020-11-17 19:49 ` hubicka at gcc dot gnu.org
` (13 subsequent siblings)
15 siblings, 0 replies; 16+ messages in thread
From: rguenth at gcc dot gnu.org @ 2020-07-23 6:51 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92535
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Target Milestone|10.2 |10.3
--- Comment #6 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 10.2 is released, adjusting target milestone.
^ permalink raw reply [flat|nested] 16+ messages in thread* [Bug ipa/92535] [10/11 regression] ICF is relatively expensive and became less effective
[not found] <bug-92535-4@http.gcc.gnu.org/bugzilla/>
2020-05-07 11:56 ` [Bug ipa/92535] [10/11 regression] ICF is relatively expensive and became less effective jakub at gcc dot gnu.org
2020-07-23 6:51 ` rguenth at gcc dot gnu.org
@ 2020-11-17 19:49 ` hubicka at gcc dot gnu.org
2020-11-17 20:03 ` hubicka at gcc dot gnu.org
` (12 subsequent siblings)
15 siblings, 0 replies; 16+ messages in thread
From: hubicka at gcc dot gnu.org @ 2020-11-17 19:49 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92535
Jan Hubicka <hubicka at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Assignee|unassigned at gcc dot gnu.org |hubicka at gcc dot gnu.org
--- Comment #7 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
With today trunk (after icf memory handling patches) I get
VM SIZE FILE SIZE
++++++++++++++ GROWING ++++++++++++++
+1.3% +1.21Mi .text +1.21Mi +1.3%
+5.2% +440Ki .eh_frame +440Ki +5.2%
+7.1% +126Ki .eh_frame_hdr +126Ki +7.1%
+0.4% +74.8Ki .rodata +74.8Ki +0.4%
+0.6% +61.0Ki .rela.dyn +61.0Ki +0.6%
+0.4% +15.4Ki .data.rel.ro.local +15.4Ki +0.4%
+1.1% +11.1Ki .data.rel.ro +11.1Ki +1.1%
[ = ] 0 .symtab +2.55Ki +0.0%
+0.0% +64 .data +64 +0.0%
+0.0% +24 .rela.plt +24 +0.0%
+0.0% +16 .plt +16 +0.0%
+0.2% +12 .gcc_except_table +12 +0.2%
+0.0% +8 .got.plt +8 +0.0%
-------------- SHRINKING --------------
[ = ] 0 .strtab -29.6Ki -0.1%
-0.0% -32 .bss 0 [ = ]
-0.0% -8 .got -8 -0.0%
-+-+-+-+-+-+-+ MIXED +-+-+-+-+-+-+-
+17% +44 [Unmapped] -3.74Ki -90.8%
+1.3% +1.92Mi TOTAL +1.89Mi +0.9%
time report of WPA with ICF:
ipa lto gimple in : 6.45 ( 4%) 2.65 ( 16%) 9.04 ( 5%)
810M ( 12%)
ipa lto gimple out : 1.87 ( 1%) 0.80 ( 5%) 2.72 ( 1%)
0 ( 0%)
ipa lto decl in : 18.63 ( 11%) 1.26 ( 7%) 20.02 ( 10%)
2682M ( 41%)
ipa lto decl out : 4.64 ( 3%) 0.20 ( 1%) 4.84 ( 3%)
0 ( 0%)
ipa icf : 17.58 ( 10%) 0.59 ( 4%) 18.22 ( 9%)
24M ( 0%)
TOTAL : 176.22 16.82 193.31
6617M
compared to:
ipa lto gimple in : 0.48 ( 0%) 0.11 ( 1%) 0.77 ( 1%)
2905k ( 0%)
ipa lto gimple out : 1.01 ( 1%) 0.84 ( 8%) 1.84 ( 1%)
0 ( 0%)
ipa lto decl in : 18.74 ( 14%) 1.31 ( 12%) 19.90 ( 13%)
2682M ( 47%)
ipa lto decl out : 4.34 ( 3%) 0.47 ( 4%) 4.79 ( 3%)
0 ( 0%)
TOTAL : 138.62 10.61 149.39
5715M
So still pretty bad.
^ permalink raw reply [flat|nested] 16+ messages in thread* [Bug ipa/92535] [10/11 regression] ICF is relatively expensive and became less effective
[not found] <bug-92535-4@http.gcc.gnu.org/bugzilla/>
` (2 preceding siblings ...)
2020-11-17 19:49 ` hubicka at gcc dot gnu.org
@ 2020-11-17 20:03 ` hubicka at gcc dot gnu.org
2020-11-17 20:05 ` hubicka at gcc dot gnu.org
` (11 subsequent siblings)
15 siblings, 0 replies; 16+ messages in thread
From: hubicka at gcc dot gnu.org @ 2020-11-17 20:03 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92535
--- Comment #8 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
Created attachment 49577
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49577&action=edit
Memory use of GCC trunk (11) with ICF
^ permalink raw reply [flat|nested] 16+ messages in thread* [Bug ipa/92535] [10/11 regression] ICF is relatively expensive and became less effective
[not found] <bug-92535-4@http.gcc.gnu.org/bugzilla/>
` (3 preceding siblings ...)
2020-11-17 20:03 ` hubicka at gcc dot gnu.org
@ 2020-11-17 20:05 ` hubicka at gcc dot gnu.org
2020-11-17 20:21 ` hubicka at gcc dot gnu.org
` (10 subsequent siblings)
15 siblings, 0 replies; 16+ messages in thread
From: hubicka at gcc dot gnu.org @ 2020-11-17 20:05 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92535
--- Comment #9 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
Created attachment 49578
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49578&action=edit
Memory use of GCC trunk (11) without ICF
^ permalink raw reply [flat|nested] 16+ messages in thread* [Bug ipa/92535] [10/11 regression] ICF is relatively expensive and became less effective
[not found] <bug-92535-4@http.gcc.gnu.org/bugzilla/>
` (4 preceding siblings ...)
2020-11-17 20:05 ` hubicka at gcc dot gnu.org
@ 2020-11-17 20:21 ` hubicka at gcc dot gnu.org
2020-11-17 21:37 ` hubicka at gcc dot gnu.org
` (9 subsequent siblings)
15 siblings, 0 replies; 16+ messages in thread
From: hubicka at gcc dot gnu.org @ 2020-11-17 20:21 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92535
--- Comment #10 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
Here are main reason for miscompares:
1125 libxul.so.wpa.076i.icf: false returned: 'variables types are
different' in equals at ../../gcc/ipa-icf.c:1697
1171 libxul.so.wpa.076i.icf: false returned: 'DECL_CXX_CONSTRUCTOR
mismatch' in equals_wpa at ../../gcc/ipa-icf.c:562
2211 libxul.so.wpa.076i.icf: false returned: 'different decl attributes' in
equals_wpa at ../../gcc/ipa-icf.c:662
3362 libxul.so.wpa.076i.icf: false returned: 'ctor polymorphic type
mismatch' in equals_wpa at ../../gcc/ipa-icf.c:585
3627 libxul.so.wpa.076i.icf: false returned: 'GIMPLE LHS type mismatch' in
compare_gimple_assign at ../../gcc/ipa-icf-gimple.c:695
5297 libxul.so.wpa.076i.icf: false returned: '' in compare_decl at
../../gcc/ipa-icf-gimple.c:157
5304 libxul.so.wpa.076i.icf: false returned: '' in compare_variable_decl at
../../gcc/ipa-icf-gimple.c:422
5304 libxul.so.wpa.076i.icf: false returned: '' in operand_equal_p at
../../gcc/ipa-icf-gimple.c:291
8588 libxul.so.wpa.076i.icf: false returned: 'parameter type is not
compatible' in compatible_parm_types_p at ../../gcc/ipa-icf.c:512
10188 libxul.so.wpa.076i.icf: false returned: 'inline attributes are
different' in compare_referenced_symbol_properties at ../../gcc/ipa-icf.c:350
16218 libxul.so.wpa.076i.icf: false returned: 'parameter types are not
compatible' in equals_wpa at ../../gcc/ipa-icf.c:639
26076 libxul.so.wpa.076i.icf: false returned: 'references to virtual tables
cannot be merged' in compare_referenced_symbol_properties at
../../gcc/ipa-icf.c:373
28813 libxul.so.wpa.076i.icf: false returned: 'decl_or_type flags are
different' in equals_wpa at ../../gcc/ipa-icf.c:572
32478 libxul.so.wpa.076i.icf: false returned: 'different tree types' in
compatible_types_p at ../../gcc/ipa-icf-gimple.c:206
60520 libxul.so.wpa.076i.icf: false returned: 'call function types are not
compatible' in compare_gimple_call at ../../gcc/ipa-icf-gimple.c:635
106399 libxul.so.wpa.076i.icf: false returned: 'types are not compatible' in
compatible_types_p at ../../gcc/ipa-icf-gimple.c:212
120240 libxul.so.wpa.076i.icf: false returned: 'result types are different'
in equals_wpa at ../../gcc/ipa-icf.c:621
269745 libxul.so.wpa.076i.icf: false returned: 'compare_ao_refs failed
(semantic difference)' in compare_operand at ../../gcc/ipa-icf-gimple.c:336
355772 libxul.so.wpa.076i.icf: false returned: '' in compare_gimple_call at
../../gcc/ipa-icf-gimple.c:607
456913 libxul.so.wpa.076i.icf: false returned: 'THIS pointer ODR type
mismatch' in equals_wpa at ../../gcc/ipa-icf.c:677
460252 libxul.so.wpa.076i.icf: false returned: 'types are not same for ODR'
in compatible_polymorphic_types_p at ../../gcc/ipa-icf-gimple.c:197
1391954 libxul.so.wpa.076i.icf: false returned: 'GIMPLE assignment operands
are different' in compare_gimple_assign at ../../gcc/ipa-icf-gimple.c:699
1477890 libxul.so.wpa.076i.icf: false returned: 'operand_equal_p failed' in
compare_operand at ../../gcc/ipa-icf-gimple.c:356
1840986 libxul.so.wpa.076i.icf: false returned: '' in equals_private at
../../gcc/ipa-icf.c:886
So compare_ao_refs is not really a top even though it can be improved. I
suppose it is time to improve the hash.
Common problem is
OBJ_TYPE_REF(_7;(struct nsServerSocket)_1->2) (_1);
OBJ_TYPE_REF(_7;(struct nsJSProtocolHandler)_1->2) (_1);
this happens 350645 (24% of the operand_equal_p failures) so I guess one can
start from there.
here clearly we want to hash ODR name of the OTR type.
^ permalink raw reply [flat|nested] 16+ messages in thread* [Bug ipa/92535] [10/11 regression] ICF is relatively expensive and became less effective
[not found] <bug-92535-4@http.gcc.gnu.org/bugzilla/>
` (5 preceding siblings ...)
2020-11-17 20:21 ` hubicka at gcc dot gnu.org
@ 2020-11-17 21:37 ` hubicka at gcc dot gnu.org
2020-11-18 7:33 ` hubicka at gcc dot gnu.org
` (8 subsequent siblings)
15 siblings, 0 replies; 16+ messages in thread
From: hubicka at gcc dot gnu.org @ 2020-11-17 21:37 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92535
--- Comment #11 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
Created attachment 49582
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49582&action=edit
Memory use of GCC 10 release branch with ICF
^ permalink raw reply [flat|nested] 16+ messages in thread* [Bug ipa/92535] [10/11 regression] ICF is relatively expensive and became less effective
[not found] <bug-92535-4@http.gcc.gnu.org/bugzilla/>
` (6 preceding siblings ...)
2020-11-17 21:37 ` hubicka at gcc dot gnu.org
@ 2020-11-18 7:33 ` hubicka at gcc dot gnu.org
2020-11-18 8:26 ` hubicka at gcc dot gnu.org
` (7 subsequent siblings)
15 siblings, 0 replies; 16+ messages in thread
From: hubicka at gcc dot gnu.org @ 2020-11-18 7:33 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92535
--- Comment #12 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
With ODR name hashing fix and fix to streaming the access types we now get:
957 false returned: 'different references' in compare_symbol_references
at ../../gcc/ipa-icf.c:465
961 false returned: 'size mismatch' in equals_wpa at
../../gcc/ipa-icf.c:1651
1171 false returned: 'DECL_CXX_CONSTRUCTOR mismatch' in equals_wpa at
../../gcc/ipa-icf.c:562
1360 false returned: '' in compare_gimple_call at
../../gcc/ipa-icf-gimple.c:607
2209 false returned: 'different decl attributes' in equals_wpa at
../../gcc/ipa-icf.c:662
3362 false returned: 'ctor polymorphic type mismatch' in equals_wpa at
../../gcc/ipa-icf.c:585
8399 false returned: 'parameter type is not compatible' in
compatible_parm_types_p at ../../gcc/ipa-icf.c:512
10161 false returned: 'inline attributes are different' in
compare_referenced_symbol_properties at ../../gcc/ipa-icf.c:350
16217 false returned: 'parameter types are not compatible' in equals_wpa at
../../gcc/ipa-icf.c:639
26071 false returned: 'references to virtual tables cannot be merged' in
compare_referenced_symbol_properties at ../../gcc/ipa-icf.c:373
28812 false returned: 'decl_or_type flags are different' in equals_wpa at
../../gcc/ipa-icf.c:572
29308 false returned: 'different tree types' in compatible_types_p at
../../gcc/ipa-icf-gimple.c:206
99308 false returned: 'types are not compatible' in compatible_types_p at
../../gcc/ipa-icf-gimple.c:212
119994 false returned: 'result types are different' in equals_wpa at
../../gcc/ipa-icf.c:621
128276 false returned: 'compare_ao_refs failed (semantic difference)' in
compare_operand at ../../gcc/ipa-icf-gimple.c:336
242454 false returned: 'operand_equal_p failed' in compare_operand at
../../gcc/ipa-icf-gimple.c:356
369360 false returned: 'GIMPLE assignment operands are different' in
compare_gimple_assign at ../../gcc/ipa-icf-gimple.c:699
370952 false returned: '' in equals_private at ../../gcc/ipa-icf.c:886
456907 false returned: 'THIS pointer ODR type mismatch' in equals_wpa at
../../gcc/ipa-icf.c:677
460246 false returned: 'types are not same for ODR' in
compatible_polymorphic_types_p at ../../gcc/ipa-icf-gimple.c:197
so ODR types of THIS pointers come next.
^ permalink raw reply [flat|nested] 16+ messages in thread* [Bug ipa/92535] [10/11 regression] ICF is relatively expensive and became less effective
[not found] <bug-92535-4@http.gcc.gnu.org/bugzilla/>
` (7 preceding siblings ...)
2020-11-18 7:33 ` hubicka at gcc dot gnu.org
@ 2020-11-18 8:26 ` hubicka at gcc dot gnu.org
2020-11-19 13:57 ` hubicka at gcc dot gnu.org
` (6 subsequent siblings)
15 siblings, 0 replies; 16+ messages in thread
From: hubicka at gcc dot gnu.org @ 2020-11-18 8:26 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92535
--- Comment #13 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
Actually, I did not wait long enough for ICF to finish dumping. Here is the
correct output. Still nice improvement for OBJ_TYPE_REF
8399 false returned: 'parameter type is not compatible' in
compatible_parm_types_p at ../../gcc/ipa-icf.c:512
10187 false returned: 'inline attributes are different' in
compare_referenced_symbol_properties at ../../gcc/ipa-icf.c:350
16217 false returned: 'parameter types are not compatible' in equals_wpa at
../../gcc/ipa-icf.c:639
26071 false returned: 'references to virtual tables cannot be merged' in
compare_referenced_symbol_properties at ../../gcc/ipa-icf.c:373
28812 false returned: 'decl_or_type flags are different' in equals_wpa at
../../gcc/ipa-icf.c:572
32287 false returned: 'different tree types' in compatible_types_p at
../../gcc/ipa-icf-gimple.c:206
60520 false returned: 'call function types are not compatible' in
compare_gimple_call at ../../gcc/ipa-icf-gimple.c:635
103684 false returned: 'types are not compatible' in compatible_types_p at
../../gcc/ipa-icf-gimple.c:212
119994 false returned: 'result types are different' in equals_wpa at
../../gcc/ipa-icf.c:621
134888 false returned: '' in compare_gimple_call at
../../gcc/ipa-icf-gimple.c:607
264624 false returned: 'compare_ao_refs failed (semantic difference)' in
compare_operand at ../../gcc/ipa-icf-gimple.c:336
456907 false returned: 'THIS pointer ODR type mismatch' in equals_wpa at
../../gcc/ipa-icf.c:677
460246 false returned: 'types are not same for ODR' in
compatible_polymorphic_types_p at ../../gcc/ipa-icf-gimple.c:197
881974 false returned: 'operand_equal_p failed' in compare_operand at
../../gcc/ipa-icf-gimple.c:356
1011296 false returned: 'GIMPLE assignment operands are different' in
compare_gimple_assign at ../../gcc/ipa-icf-gimple.c:699
1239444 false returned: '' in equals_private at ../../gcc/ipa-icf.c:886
^ permalink raw reply [flat|nested] 16+ messages in thread* [Bug ipa/92535] [10/11 regression] ICF is relatively expensive and became less effective
[not found] <bug-92535-4@http.gcc.gnu.org/bugzilla/>
` (8 preceding siblings ...)
2020-11-18 8:26 ` hubicka at gcc dot gnu.org
@ 2020-11-19 13:57 ` hubicka at gcc dot gnu.org
2020-11-19 18:10 ` hubicka at gcc dot gnu.org
` (5 subsequent siblings)
15 siblings, 0 replies; 16+ messages in thread
From: hubicka at gcc dot gnu.org @ 2020-11-19 13:57 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92535
--- Comment #14 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
I fixed some issues
1) merging of OBJ_TYPE_REF was broken
2) last version of my COMPONENT_REF patch clears incorrectly OEP_ADDRESS_OF
3) gimple clobbers mismatches for no good reason
4) volatile memory references can are never considered equal
5) alignment of memory accesses is not hashed and may cause late mismatch
(common in vector code)
With this I get
141 false returned: 'Declaration mismatch' in equals at
../../gcc/ipa-icf.c:1817
162 false returned: 'DECL_CXX_DESTRUCTOR mismatch' in equals_wpa at
../../gcc/ipa-icf.c:565
169 false returned: '' in compare_decl at ../../gcc/ipa-icf-gimple.c:162
246 false returned: 'final flag mismatch' in
compare_referenced_symbol_properties at ../../gcc/ipa-icf.c:401
380 false returned: '' in compare_phi_node at ../../gcc/ipa-icf.c:1595
380 false returned: 'PHI node comparison returns false' in equals_private
at ../../gcc/ipa-icf.c:916
513 false returned: 'different decl attributes' in equals_wpa at
../../gcc/ipa-icf.c:662
571 false returned: 'compare_ao_refs failed (dependence clique
difference)' in compare_operand at ../../gcc/ipa-icf-gimple.c:368
811 false returned: 'GIMPLE call operands are different' in
compare_gimple_call at ../../gcc/ipa-icf-gimple.c:676
912 false returned: 'different references' in compare_symbol_references
at ../../gcc/ipa-icf.c:465
961 false returned: 'size mismatch' in equals_wpa at
../../gcc/ipa-icf.c:1666
1125 false returned: 'variables types are different' in equals at
../../gcc/ipa-icf.c:1712
1129 false returned: 'DECL_CXX_CONSTRUCTOR mismatch' in equals_wpa at
../../gcc/ipa-icf.c:562
1453 false returned: 'operand_equal_p failed' in compare_operand at
../../gcc/ipa-icf-gimple.c:376
2917 false returned: 'compare_ao_refs failed (semantic difference)' in
compare_operand at ../../gcc/ipa-icf-gimple.c:356
3362 false returned: 'ctor polymorphic type mismatch' in equals_wpa at
../../gcc/ipa-icf.c:585
3527 false returned: '' in compare_decl at ../../gcc/ipa-icf-gimple.c:157
3535 false returned: '' in compare_variable_decl at
../../gcc/ipa-icf-gimple.c:442
3535 false returned: '' in operand_equal_p at
../../gcc/ipa-icf-gimple.c:307
3563 false returned: 'GIMPLE assignment operands are different' in
compare_gimple_assign at ../../gcc/ipa-icf-gimple.c:719
4604 false returned: 'GIMPLE LHS type mismatch' in compare_gimple_assign
at ../../gcc/ipa-icf-gimple.c:715
6577 false returned: 'parameter type is not compatible' in
compatible_parm_types_p at ../../gcc/ipa-icf.c:512
8441 false returned: 'inline attributes are different' in
compare_referenced_symbol_properties at ../../gcc/ipa-icf.c:350
11259 false returned: 'parameter types are not compatible' in equals_wpa at
../../gcc/ipa-icf.c:639
25276 false returned: 'different tree types' in compatible_types_p at
../../gcc/ipa-icf-gimple.c:206
26073 false returned: 'references to virtual tables cannot be merged' in
compare_referenced_symbol_properties at ../../gcc/ipa-icf.c:373
26693 false returned: 'decl_or_type flags are different' in equals_wpa at
../../gcc/ipa-icf.c:572
61657 false returned: 'call function types are not compatible' in
compare_gimple_call at ../../gcc/ipa-icf-gimple.c:655
70880 false returned: '' in equals_private at ../../gcc/ipa-icf.c:881
104034 false returned: 'types are not compatible' in compatible_types_p at
../../gcc/ipa-icf-gimple.c:212
113477 false returned: 'result types are different' in equals_wpa at
../../gcc/ipa-icf.c:621
410680 false returned: 'THIS pointer ODR type of cdtor mismatch' in
equals_wpa at ../../gcc/ipa-icf.c:673
414222 false returned: 'types are not same for ODR' in
compatible_polymorphic_types_p at ../../gcc/ipa-icf-gimple.c:197
and more code size savings
VM SIZE FILE SIZE
++++++++++++++ GROWING ++++++++++++++
[ = ] 0 .strtab +439Ki +1.0%
[ = ] 0 .symtab +73.4Ki +0.4%
+0.0% +32 .data +32 +0.0%
-------------- SHRINKING --------------
-2.9% -2.67Mi .text -2.67Mi -2.9%
-6.3% -555Ki .eh_frame -555Ki -6.3%
-7.8% -148Ki .eh_frame_hdr -148Ki -7.8%
-0.4% -78.2Ki .rodata -78.2Ki -0.4%
-0.5% -58.8Ki .rela.dyn -58.8Ki -0.5%
-0.4% -15.0Ki .data.rel.ro.local -15.0Ki -0.4%
-1.0% -10.4Ki .data.rel.ro -10.4Ki -1.0%
-0.0% -64 .bss 0 [ = ]
-0.4% -32 .gcc_except_table -32 -0.4%
-+-+-+-+-+-+-+ MIXED +-+-+-+-+-+-+-
+14% +40 [Unmapped] -1.40Ki -42.6%
-2.3% -3.51Mi TOTAL -3.01Mi -1.4%
ICF still take quite some memory:
ipa lto gimple in : 5.82 ( 4%) 2.10 ( 15%) 8.05 ( 5%)
776M ( 12%)
ipa lto gimple out : 1.52 ( 1%) 0.69 ( 5%) 2.23 ( 1%)
0 ( 0%)
ipa icf : 11.64 ( 7%) 0.51 ( 4%) 12.23 ( 7%)
43M ( 1%)
TOTAL : 163.47 13.69 177.49
6537M
So about 100 reduction and ICF got faster. It also improved the chance that
functions read in are actually merged.
The times are not 100% comparable to my previous report since my earlier build
was with profiledbootstrap while now I bootstrapped the normal way.
There are still quite many nonsences. In particular
1) we could enable merging different polymorphic ODR types of this pointers.
Late devitualization is not that important, but we would need to have way
to notice such merges. This is discussed in thread
https://gcc.gnu.org/pipermail/gcc-patches/2020-November/559573.html
2) result types, operand types and ODR types can be hashed in before the
equals_wpa process start to avoid refining the partitions (that should get rid
of good part of ipa-icf time that now seems to be dominated by equal_wpa and
fibonaci heap)
3) we should be able to do something about call operand types compare - it is
way too sensitive.
4) gimple LHS type test is overactive. I see it is needed for conversions but
it is not needed otherwise.
5) we do bad job on hashing SSA_NAMES. In Firefox there seems to be quite few
functions that differs only by permutation of parameters or permutation of
parameters of callee which we do not see as a difference in hash functions
^ permalink raw reply [flat|nested] 16+ messages in thread* [Bug ipa/92535] [10/11 regression] ICF is relatively expensive and became less effective
[not found] <bug-92535-4@http.gcc.gnu.org/bugzilla/>
` (9 preceding siblings ...)
2020-11-19 13:57 ` hubicka at gcc dot gnu.org
@ 2020-11-19 18:10 ` hubicka at gcc dot gnu.org
2021-04-08 12:02 ` rguenth at gcc dot gnu.org
` (4 subsequent siblings)
15 siblings, 0 replies; 16+ messages in thread
From: hubicka at gcc dot gnu.org @ 2020-11-19 18:10 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92535
--- Comment #15 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
Current mainline with
1) fix to COMPONENT_REFs described above
2) improvement of ODR type having for THIS pointers
3) gimple_clobber fix (already approved)
4) compare_ao_refs fix for volatile accesses
5) compare_decs change so it does not test type compatibility of automatic
vars (since these types have little meaning)
1 false returned: '' in compare_decl at ../../gcc/ipa-icf-gimple.c:163
1 false returned: 'replaceable operator flags
are different' in compare_referenced_symbol_properties at
../../gcc/ipa-icf.c:359
4 false returned: '' in compare_bb at ../../gcc/ipa-icf-gimple.c:606
7 false returned: 'Mismatched number of parameters'
in equals_wpa at ../../gcc/ipa-icf.c:651
9 false returned: '' in
compare_variable_decl at ../../gcc/ipa-icf-gimple.c:447
9 false returned: '' in operand_equal_p at ../../gcc/ipa-icf-gimple.c:312
10 false returned: 'INTEGER_CST precision mismatch'
in equals at ../../gcc/ipa-icf.c:1807
11 false returned: 'case
low values are different' in compare_gimple_switch at
../../gcc/ipa-icf-gimple.c:808
23 false returned: '' in operand_equal_p at
../../gcc/ipa-icf-gimple.c:303
23 false
returned: 'one type is not polymorphic' in compatible_polymorphic_types_p at
../../gcc/ipa-icf-gimple.c:208
23 false returned: 'THIS pointer ODR type mismatch' in
equals_wpa at ../../gcc/ipa-icf.c:678
29 false returned: 'ASM
strings are different' in compare_gimple_asm at ../../gcc/ipa-icf-gimple.c:909
46 false returned: 'GIMPLE call operands are different' in
compare_gimple_call at ../../gcc/ipa-icf-gimple.c:681
141 false returned: 'Declaration mismatch' in equals at
../../gcc/ipa-icf.c:1803
169 false returned: '' in compare_decl at ../../gcc/ipa-icf-gimple.c:175
182 false returned: 'DECL_CXX_DESTRUCTOR mismatch' in equals_wpa at
../../gcc/ipa-icf.c:565
246 false returned: 'final flag mismatch' in
compare_referenced_symbol_properties at ../../gcc/ipa-icf.c:401
391 false returned: '' in compare_phi_node at ../../gcc/ipa-icf.c:1581
391 false returned: 'PHI node comparison returns false' in equals_private
at ../../gcc/ipa-icf.c:922
571 false returned: 'compare_ao_refs failed (dependence clique
difference)' in compare_operand at ../../gcc/ipa-icf-gimple.c:373
676 false returned: 'compare_ao_refs failed (semantic difference)' in
compare_operand at ../../gcc/ipa-icf-gimple.c:361
676 false returned: 'METHOD_TYPE and FUNCTION_TYPE mismatch' in
equals_wpa at ../../gcc/ipa-icf.c:674
707 false returned: 'operand_equal_p failed' in compare_operand at
../../gcc/ipa-icf-gimple.c:381
957 false returned: 'different references' in compare_symbol_references
at ../../gcc/ipa-icf.c:465
961 false returned: 'size mismatch' in equals_wpa at
../../gcc/ipa-icf.c:1652
1125 false returned: 'variables types are different' in equals at
../../gcc/ipa-icf.c:1698
1171 false returned: 'DECL_CXX_CONSTRUCTOR mismatch' in equals_wpa at
../../gcc/ipa-icf.c:562
1321 false returned: 'GIMPLE assignment operands are different' in
compare_gimple_assign at ../../gcc/ipa-icf-gimple.c:724
2209 false returned: 'different decl attributes' in equals_wpa at
../../gcc/ipa-icf.c:663
4643 false returned: 'GIMPLE LHS type mismatch' in compare_gimple_assign
at ../../gcc/ipa-icf-gimple.c:720
8400 false returned: 'parameter type is not compatible' in
compatible_parm_types_p at ../../gcc/ipa-icf.c:512
10187 false returned: 'inline attributes are different' in
compare_referenced_symbol_properties at ../../gcc/ipa-icf.c:350
16210 false returned: 'parameter types are not compatible' in equals_wpa at
../../gcc/ipa-icf.c:640
26071 false returned: 'references to virtual tables cannot be merged' in
compare_referenced_symbol_properties at ../../gcc/ipa-icf.c:373
28810 false returned: 'decl_or_type flags are different' in equals_wpa at
../../gcc/ipa-icf.c:572
32971 false returned: 'different tree types' in compatible_types_p at
../../gcc/ipa-icf-gimple.c:219
61665 false returned: 'call function types are not compatible' in
compare_gimple_call at ../../gcc/ipa-icf-gimple.c:660
96652 false returned: '' in equals_private at ../../gcc/ipa-icf.c:887
101194 false returned: 'types are not compatible' in compatible_types_p at
../../gcc/ipa-icf-gimple.c:225
119996 false returned: 'result types are different' in equals_wpa at
../../gcc/ipa-icf.c:622
many of call function types are not compatible are indirect calls with
mismatched number of parameters. This is because we hash SSA_NAME as nothing.
That is easy to fix.
^ permalink raw reply [flat|nested] 16+ messages in thread* [Bug ipa/92535] [10/11 regression] ICF is relatively expensive and became less effective
[not found] <bug-92535-4@http.gcc.gnu.org/bugzilla/>
` (10 preceding siblings ...)
2020-11-19 18:10 ` hubicka at gcc dot gnu.org
@ 2021-04-08 12:02 ` rguenth at gcc dot gnu.org
2021-04-15 9:38 ` [Bug ipa/92535] [10 " hubicka at gcc dot gnu.org
` (3 subsequent siblings)
15 siblings, 0 replies; 16+ messages in thread
From: rguenth at gcc dot gnu.org @ 2021-04-08 12:02 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92535
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Target Milestone|10.3 |10.4
--- Comment #16 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 10.3 is being released, retargeting bugs to GCC 10.4.
^ permalink raw reply [flat|nested] 16+ messages in thread* [Bug ipa/92535] [10 regression] ICF is relatively expensive and became less effective
[not found] <bug-92535-4@http.gcc.gnu.org/bugzilla/>
` (11 preceding siblings ...)
2021-04-08 12:02 ` rguenth at gcc dot gnu.org
@ 2021-04-15 9:38 ` hubicka at gcc dot gnu.org
2021-05-04 12:31 ` rguenth at gcc dot gnu.org
` (2 subsequent siblings)
15 siblings, 0 replies; 16+ messages in thread
From: hubicka at gcc dot gnu.org @ 2021-04-15 9:38 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92535
Jan Hubicka <hubicka at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|ASSIGNED |NEW
Summary|[10/11 regression] ICF is |[10 regression] ICF is
|relatively expensive and |relatively expensive and
|became less effective |became less effective
--- Comment #17 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
For GCC 11 we now get faster build times with ICF than without on cc1plus,
Firefox and clang LTO build. So I think we can consider it no longer
regression while ICF can always be improved (and I have some changes queues for
next stage1).
I have no plan to backport this to gcc10, so unasigning.
^ permalink raw reply [flat|nested] 16+ messages in thread* [Bug ipa/92535] [10 regression] ICF is relatively expensive and became less effective
[not found] <bug-92535-4@http.gcc.gnu.org/bugzilla/>
` (12 preceding siblings ...)
2021-04-15 9:38 ` [Bug ipa/92535] [10 " hubicka at gcc dot gnu.org
@ 2021-05-04 12:31 ` rguenth at gcc dot gnu.org
2022-06-28 10:38 ` jakub at gcc dot gnu.org
2023-07-07 8:41 ` rguenth at gcc dot gnu.org
15 siblings, 0 replies; 16+ messages in thread
From: rguenth at gcc dot gnu.org @ 2021-05-04 12:31 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92535
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |ASSIGNED
^ permalink raw reply [flat|nested] 16+ messages in thread* [Bug ipa/92535] [10 regression] ICF is relatively expensive and became less effective
[not found] <bug-92535-4@http.gcc.gnu.org/bugzilla/>
` (13 preceding siblings ...)
2021-05-04 12:31 ` rguenth at gcc dot gnu.org
@ 2022-06-28 10:38 ` jakub at gcc dot gnu.org
2023-07-07 8:41 ` rguenth at gcc dot gnu.org
15 siblings, 0 replies; 16+ messages in thread
From: jakub at gcc dot gnu.org @ 2022-06-28 10:38 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92535
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Target Milestone|10.4 |10.5
--- Comment #18 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
GCC 10.4 is being released, retargeting bugs to GCC 10.5.
^ permalink raw reply [flat|nested] 16+ messages in thread* [Bug ipa/92535] [10 regression] ICF is relatively expensive and became less effective
[not found] <bug-92535-4@http.gcc.gnu.org/bugzilla/>
` (14 preceding siblings ...)
2022-06-28 10:38 ` jakub at gcc dot gnu.org
@ 2023-07-07 8:41 ` rguenth at gcc dot gnu.org
15 siblings, 0 replies; 16+ messages in thread
From: rguenth at gcc dot gnu.org @ 2023-07-07 8:41 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92535
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Target Milestone|10.5 |11.0
Known to fail| |10.5.0
Status|ASSIGNED |RESOLVED
Resolution|--- |FIXED
--- Comment #19 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed in GCC 11 then.
--- Comment #20 from Richard Biener <rguenth at gcc dot gnu.org> ---
.
^ permalink raw reply [flat|nested] 16+ messages in thread