A recent patch increased GCC's memory consumption!

Thu Nov 24 12:38:00 GMT 2005

Hi,

I am a friendly script caring about memory consumption in GCC.  Please
contact jh@suse.cz if something is going wrong.

Comparing memory consumption on compilation of combine.i, insn-attrtab.i,
and generate-3.4.ii I got:


comparing combine.c compilation at -O0 level:
    Overall memory needed: 24913k
    Peak memory use before GGC: 9599k
    Peak memory use after GGC: 8946k
    Maximum of released memory in single GGC run: 2736k
    Garbage: 40102k
    Leak: 6709k
    Overhead: 5790k
    GGC runs: 317

comparing combine.c compilation at -O1 level:
    Overall memory needed: 26824k
    Peak memory use before GGC: 17367k
    Peak memory use after GGC: 17184k
    Maximum of released memory in single GGC run: 2372k
    Garbage: 60655k
    Leak: 6892k
    Overhead: 7581k
    GGC runs: 386

comparing combine.c compilation at -O2 level:
    Overall memory needed: 26824k
    Peak memory use before GGC: 17370k
    Peak memory use after GGC: 17184k
    Maximum of released memory in single GGC run: 2453k
    Garbage: 77410k -> 77414k
    Leak: 6977k
    Overhead: 10023k -> 10024k
    GGC runs: 455

comparing combine.c compilation at -O3 level:
  Amount of memory still referenced at the end of compilation increased from 7034k to 7050k, overall 0.23%
    Overall memory needed: 26824k
    Peak memory use before GGC: 18368k
    Peak memory use after GGC: 17999k
    Maximum of released memory in single GGC run: 3509k
    Garbage: 107819k -> 107822k
    Leak: 7034k -> 7050k
    Overhead: 13566k -> 13566k
    GGC runs: 508

comparing insn-attrtab.c compilation at -O0 level:
    Overall memory needed: 80928k
    Peak memory use before GGC: 69473k
    Peak memory use after GGC: 45011k
    Maximum of released memory in single GGC run: 36247k
    Garbage: 146440k
    Leak: 10101k
    Overhead: 19797k
    GGC runs: 251

comparing insn-attrtab.c compilation at -O1 level:
    Overall memory needed: 111720k
    Peak memory use before GGC: 94043k
    Peak memory use after GGC: 83560k
    Maximum of released memory in single GGC run: 32588k
    Garbage: 289921k
    Leak: 10079k
    Overhead: 36686k
    GGC runs: 244

comparing insn-attrtab.c compilation at -O2 level:
    Overall memory needed: 127168k
    Peak memory use before GGC: 113357k
    Peak memory use after GGC: 83476k
    Maximum of released memory in single GGC run: 32177k
    Garbage: 372235k -> 372235k
    Leak: 10180k
    Overhead: 48199k -> 48199k
    GGC runs: 272

comparing insn-attrtab.c compilation at -O3 level:
    Overall memory needed: 127212k
    Peak memory use before GGC: 113381k
    Peak memory use after GGC: 83498k
    Maximum of released memory in single GGC run: 32501k
    Garbage: 372851k -> 372851k
    Leak: 10185k
    Overhead: 48354k -> 48354k
    GGC runs: 276

comparing Gerald's testcase PR8361 compilation at -O0 level:
    Overall memory needed: 118392k
    Peak memory use before GGC: 95141k
    Peak memory use after GGC: 94193k
    Maximum of released memory in single GGC run: 20460k
    Garbage: 223586k
    Leak: 49201k
    Overhead: 36726k
    GGC runs: 369

comparing Gerald's testcase PR8361 compilation at -O1 level:
    Overall memory needed: 105500k
    Peak memory use before GGC: 95270k
    Peak memory use after GGC: 93195k
    Maximum of released memory in single GGC run: 20343k
    Garbage: 566560k
    Leak: 55069k
    Overhead: 67513k
    GGC runs: 515

comparing Gerald's testcase PR8361 compilation at -O2 level:
    Overall memory needed: 106396k -> 106424k
    Peak memory use before GGC: 95270k
    Peak memory use after GGC: 93196k
    Maximum of released memory in single GGC run: 20342k
    Garbage: 651124k -> 651112k
    Leak: 55828k -> 55841k
    Overhead: 78842k -> 78849k
    GGC runs: 585 -> 586

comparing Gerald's testcase PR8361 compilation at -O3 level:
    Overall memory needed: 108816k -> 108812k
    Peak memory use before GGC: 96738k
    Peak memory use after GGC: 94273k
    Maximum of released memory in single GGC run: 20839k
    Garbage: 707211k -> 707224k
    Leak: 57038k -> 57042k
    Overhead: 83856k -> 83859k
    GGC runs: 596 -> 594

Head of the ChangeLog is:

--- /usr/src/SpecTests/sandbox-britten-memory/x86_64/mem-result/ChangeLog	2005-11-24 01:09:56.000000000 +0000
+++ /usr/src/SpecTests/sandbox-britten-memory/gcc/gcc/ChangeLog	2005-11-24 11:28:00.000000000 +0000
@@ -1,3 +1,54 @@
+2005-11-24  Richard Guenther  <rguenther@suse.de>
+	Dirk Mueller <dmueller@suse.de>
+
+	PR c++/14024
+	* c-common.h (strict_aliasing_warning): Declare.
+	* c-common.c (strict_aliasing_warning): New function,
+	split out from ...
+	* c-typeck.c (build_c_cast): ... here.
+
+2005-11-24  Paolo Bonzini  <bonzini@gnu.org>
+
+	* optabs.c (expand_binop): Use swap_commutative_operands_with_target
+	to order operands.
+	(swap_commutative_operands_with_target): New.
+
+2005-11-24  Paolo Bonzini  <bonzini@gnu.org>
+
+	* gcse.c (hash_scan_set): Look through REG_EQUAL or REG_EQUIV notes
+	also when doing PRE, rather than only for global CPROP.
+
+2005-11-24  Ben Elliston  <bje@au.ibm.com>
+
+	* machmode.h (CLASS_HAS_WIDER_MODES_P): New macro.
+	* optabs.c (expand_binop): Use CLASS_HAS_WIDER_MODES_P, improve
+	formatting.
+	(expand_twoval_unop): Likewise.
+	(widen_clz): Likewise.
+	(expand_parity): Likewise.
+	(expand_unop): Likewise.
+	(emit_cmp_and_jmp_insn_1): Likewise.
+	(prepare_float_lib_cmp): Likewise.
+
+2005-11-24  Ben Elliston  <bje@au.ibm.com>
+
+	* optabs.c (prepare_cmp_insn): Use SCALAR_FLOAT_MODE_P.
+	Remove unused `class' variable.
+
+2005-11-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
+
+	PR c++/21667
+	* c-typeck.c (build_array_ref): Avoid code duplicate.  Use common
+	C/C++ diagnostic function warn_array_subscript_with_type_char. 
+	* c-common.h (warn_array_subscript_with_type_char): Declare. 
+	* c-common.c (warn_array_subscript_with_type_char): Define.
+
+2005-11-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
+
+	PR c/21668
+	* c-decl.c (grokdeclarator): Don't warn for 'extern const' when
+	compiling at the intersection of C and C++.
+
 2005-11-23  Mark Mitchell  <mark@codesourcery.com>
 
 	* doc/invoke.texi: For man pages, include gcc-vers.texi.
--- /usr/src/SpecTests/sandbox-britten-memory/x86_64/mem-result/ChangeLog.cp	2005-11-22 18:15:51.000000000 +0000
+++ /usr/src/SpecTests/sandbox-britten-memory/gcc/gcc/cp/ChangeLog	2005-11-24 11:28:00.000000000 +0000
@@ -1,3 +1,22 @@
+2005-11-24  Richard Guenther  <rguenther@suse.de>
+	Dirk Mueller <dmueller@suse.de>
+
+	PR c++/14024
+	* typeck.c (build_reinterpret_cast_1): Use
+	strict_aliasing_warning.
+
+2005-11-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
+
+	PR c++/24235
+	* pt.c (check_instantiated_args): Reword diagnostic message about
+	template argument involving local types.
+
+2005-11-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
+
+	PR c++/21667
+	* typeck.c (build_array_ref): Avoid code duplicate.  Use common
+	C/C++ diagnostic function warn_array_subscript_with_type_char. 
+
 2005-11-21  Gabriel Dos Reis  <gdr@integrable-solutions.net>
 
 	PR c++/22238


The results can be reproduced by building a compiler with

--enable-gather-detailed-mem-stats targetting x86-64

and compiling preprocessed combine.c or testcase from PR8632 with:

-fmem-report --param=ggc-min-heapsize=1024 --param=ggc-min-expand=1 -Ox -Q

The memory consumption summary appears in the dump after detailed listing
of the places they are allocated in.  Peak memory consumption is actually
computed by looking for maximal value in {GC XXXX -> YYYY} report.

Your testing script.

