GCC memory consumption increased by recent patch!

Wed Mar 2 22:01:00 GMT 2005

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


comparing combine.c compilation at -O0 level:
    Overall memory needed: 24669k
    Peak memory use before GGC: 9359k
    Peak memory use after GGC: 8673k
    Maximum of released memory in single GGC run: 2865k
    Garbage: 41675k
    Leak: 6395k
    Overhead: 5772k
    GGC runs: 327

comparing combine.c compilation at -O1 level:
    Overall memory needed: 24489k
    Peak memory use before GGC: 8673k
    Peak memory use after GGC: 8093k
    Maximum of released memory in single GGC run: 2026k
    Garbage: 58507k
    Leak: 6756k
    Overhead: 9891k
    GGC runs: 500

comparing combine.c compilation at -O2 level:
  Amount of produced GGC garbage increased from 77998k to 78701k, overall 0.90%
    Overall memory needed: 29145k -> 29141k
    Peak memory use before GGC: 12671k
    Peak memory use after GGC: 12544k
    Maximum of released memory in single GGC run: 2596k
    Garbage: 77998k -> 78701k
    Leak: 6593k
    Overhead: 14177k -> 14236k
    GGC runs: 515 -> 520

comparing combine.c compilation at -O3 level:
  Peak amount of GGC memory allocated before garbage collecting increased from 12680k to 12724k, overall 0.35%
  Amount of produced GGC garbage increased from 105184k to 106399k, overall 1.15%
    Overall memory needed: 30881k -> 30829k
    Peak memory use before GGC: 12680k -> 12724k
    Peak memory use after GGC: 12544k
    Maximum of released memory in single GGC run: 3427k -> 3429k
    Garbage: 105184k -> 106399k
    Leak: 7180k -> 7183k
    Overhead: 18999k -> 19099k
    GGC runs: 581 -> 582

comparing insn-attrtab.c compilation at -O0 level:
    Overall memory needed: 114212k
    Peak memory use before GGC: 74821k
    Peak memory use after GGC: 45494k
    Maximum of released memory in single GGC run: 39413k
    Garbage: 152830k
    Leak: 10984k
    Overhead: 19970k
    GGC runs: 273

comparing insn-attrtab.c compilation at -O1 level:
    Overall memory needed: 124600k
    Peak memory use before GGC: 78641k
    Peak memory use after GGC: 70036k
    Maximum of released memory in single GGC run: 40782k
    Garbage: 369270k
    Leak: 11361k
    Overhead: 71894k
    GGC runs: 398

comparing insn-attrtab.c compilation at -O2 level:
  Amount of produced GGC garbage increased from 483091k to 489784k, overall 1.39%
    Overall memory needed: 147452k -> 148972k
    Peak memory use before GGC: 97678k
    Peak memory use after GGC: 83497k
    Maximum of released memory in single GGC run: 39453k
    Garbage: 483091k -> 489784k
    Leak: 11250k -> 11249k
    Overhead: 86926k -> 87470k
    GGC runs: 341 -> 343

comparing insn-attrtab.c compilation at -O3 level:
  Amount of produced GGC garbage increased from 483905k to 490598k, overall 1.38%
    Overall memory needed: 148932k -> 147516k
    Peak memory use before GGC: 97679k
    Peak memory use after GGC: 83499k
    Maximum of released memory in single GGC run: 39453k
    Garbage: 483905k -> 490598k
    Leak: 11287k -> 11286k
    Overhead: 87057k -> 87601k
    GGC runs: 347 -> 349

comparing Gerald's testcase PR8361 compilation at -O0 level:
    Overall memory needed: 111952k
    Peak memory use before GGC: 87325k
    Peak memory use after GGC: 85900k
    Maximum of released memory in single GGC run: 19511k
    Garbage: 246370k
    Leak: 55513k
    Overhead: 43275k
    GGC runs: 366

comparing Gerald's testcase PR8361 compilation at -O1 level:
    Overall memory needed: 104041k -> 104037k
    Peak memory use before GGC: 86183k
    Peak memory use after GGC: 85119k
    Maximum of released memory in single GGC run: 18950k
    Garbage: 434991k -> 434991k
    Leak: 56864k
    Overhead: 65164k -> 65164k
    GGC runs: 512

comparing Gerald's testcase PR8361 compilation at -O2 level:
    Overall memory needed: 104253k -> 104257k
    Peak memory use before GGC: 86184k
    Peak memory use after GGC: 85119k
    Maximum of released memory in single GGC run: 18951k
    Garbage: 478256k -> 478569k
    Leak: 57432k
    Overhead: 75179k -> 75203k
    GGC runs: 571 -> 572

comparing Gerald's testcase PR8361 compilation at -O3 level:
  Amount of produced GGC garbage increased from 483490k to 484103k, overall 0.13%
    Overall memory needed: 105353k
    Peak memory use before GGC: 87323k
    Peak memory use after GGC: 86199k
    Maximum of released memory in single GGC run: 19400k
    Garbage: 483490k -> 484103k
    Leak: 57613k -> 57621k
    Overhead: 75879k -> 75948k
    GGC runs: 558 -> 559

Head of changelog is:

--- /usr/src/SpecTests/sandbox-britten-memory/x86_64/mem-result/ChangeLog	2005-03-02 16:49:25.000000000 +0000
+++ /usr/src/SpecTests/sandbox-britten-memory/gcc/gcc/ChangeLog	2005-03-02 21:03:09.000000000 +0000
@@ -1,3 +1,44 @@
+2005-03-02  Geoffrey Keating  <geoffk@apple.com>
+
+	* config/darwin8.h (LIB_SPEC): Switch -lmx and -lSystem.
+
+2005-03-02  Devang Patel  <dpatel@apple.com>
+	
+	Undo PR tree-optimization/18815 fix.
+	
+2005-03-02  Devang Patel  <dpatel@apple.com>
+
+	PR tree-optimization/18815
+	* tree-if-conv.c (combine_blocks): Adjust loop header edges for
+	loops with zero exit edges.
+	
+2005-03-02  Kazu Hirata  <kazu@cs.umass.edu>
+
+	* cfg.c (connect_src, connect_dest, disconnect_src,
+	disconnct_dest): New.
+	(unchecked_make_edge, remove_edge, redirect_edge_succ,
+	redirect_edge_pred): Use the new functions.
+
+	* tree-ssa-dom.c (tree_ssa_dominator_optimize): Call
+	free_dominance_info only when needed.
+
+2005-03-02  David Edelsohn  <edelsohn@gnu.org>
+
+	PR target/20276
+	* config/rs6000/predicates.md (reg_or_cint64_operand): Fix typo.
+	(reg_or_sub_cint64_operand): Same.
+
+2005-03-02  Jeff Law  <law@redhat.com>
+
+	* tree-ssa-dom.c (tree_ssa_dominator_optimize): Fix setting of
+	CFG_ALTERED in the main DOM iteration loop.
+
+2005-03-02  Mark Mitchell  <mark@codesourcery.com>
+
+	PR c++/19916
+	* varasm.c (initializer_constant_valid_p): Allow conversions
+	between OFFSET_TYPEs.  Tidy.
+
 2005-03-02  Hans-Peter Nilsson  <hp@axis.com>
 
 	* config/cris/cris.md ("return"): Remove epilogue delay list
--- /usr/src/SpecTests/sandbox-britten-memory/x86_64/mem-result/ChangeLog.cp	2005-03-02 01:57:35.000000000 +0000
+++ /usr/src/SpecTests/sandbox-britten-memory/gcc/gcc/cp/ChangeLog	2005-03-02 21:03:10.000000000 +0000
@@ -1,3 +1,9 @@
+2005-03-02  Alexandre Oliva  <aoliva@redhat.com>
+
+	* name-lookup.c (push_overloaded_decl): Don't error if the new
+	decl matches the old one.
+	* decl.c (redeclaration_error_message): Likewise.
+
 2005-03-01  Per Bothner  <per@bothner.com>
 
 	* decl.c (finish_function): Use SET_EXPR_LOCATION instead of

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

The results can be reproduced by building 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.

Yours testing script.

