This is the mail archive of the
gcc-regression@gcc.gnu.org
mailing list for the GCC project.
GCC memory consumption increased by recent patch!
- From: gcctest at suse dot de
- To: jh at suse dot cz, gcc-regression at gcc dot gnu dot org
- Date: Sat, 01 Jan 2005 07:51:52 +0000
- Subject: GCC memory consumption increased by recent patch!
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: 24389k
Peak memory use before GGC: 9063k
Peak memory use after GGC: 8377k
Maximum of released memory in single GGC run: 2864k
Garbage: 41736k
Leak: 5863k
Overhead: 5545k
GGC runs: 352
comparing combine.c compilation at -O1 level:
Overall memory needed: 25361k
Peak memory use before GGC: 8951k
Peak memory use after GGC: 8450k
Maximum of released memory in single GGC run: 2024k
Garbage: 66726k
Leak: 6258k
Overhead: 10311k
GGC runs: 545
comparing combine.c compilation at -O2 level:
Overall memory needed: 28845k
Peak memory use before GGC: 12383k
Peak memory use after GGC: 12257k
Maximum of released memory in single GGC run: 2533k
Garbage: 80657k
Leak: 6082k
Overhead: 14119k
GGC runs: 546
comparing combine.c compilation at -O3 level:
Overall memory needed: 31345k
Peak memory use before GGC: 12635k
Peak memory use after GGC: 12257k
Maximum of released memory in single GGC run: 3347k
Garbage: 109133k
Leak: 6619k
Overhead: 18983k
GGC runs: 613
comparing insn-attrtab.c compilation at -O0 level:
Overall memory needed: 117268k
Peak memory use before GGC: 77779k
Peak memory use after GGC: 45259k
Maximum of released memory in single GGC run: 42606k
Garbage: 159302k
Leak: 10433k
Overhead: 20572k
GGC runs: 294
comparing insn-attrtab.c compilation at -O1 level:
Overall memory needed: 128492k
Peak memory use before GGC: 83296k
Peak memory use after GGC: 69015k
Maximum of released memory in single GGC run: 40616k
Garbage: 441432k
Leak: 10779k
Overhead: 77164k
GGC runs: 431
comparing insn-attrtab.c compilation at -O2 level:
Overall memory needed: 153116k
Peak memory use before GGC: 99056k
Peak memory use after GGC: 84176k
Maximum of released memory in single GGC run: 41525k
Garbage: 487966k
Leak: 10703k
Overhead: 85264k
GGC runs: 363
comparing insn-attrtab.c compilation at -O3 level:
Overall memory needed: 153128k
Peak memory use before GGC: 99057k
Peak memory use after GGC: 84178k
Maximum of released memory in single GGC run: 41525k
Garbage: 489168k
Leak: 10745k
Overhead: 85424k
GGC runs: 373
comparing Gerald's testcase PR8361 compilation at -O0 level:
Amount of produced GGC garbage increased from 242698k to 246468k, overall 1.55%
Amount of memory still referenced at the end of compilation increased from 54798k to 54981k, overall 0.33%
Overall memory needed: 110796k
Peak memory use before GGC: 86609k
Peak memory use after GGC: 85641k
Maximum of released memory in single GGC run: 19292k
Garbage: 242698k -> 246468k
Leak: 54798k -> 54981k
Overhead: 42333k -> 43128k
GGC runs: 364 -> 369
comparing Gerald's testcase PR8361 compilation at -O1 level:
Overall memory needed: 103701k
Peak memory use before GGC: 85639k
Peak memory use after GGC: 84690k
Maximum of released memory in single GGC run: 18904k
Garbage: 466595k -> 466427k
Leak: 56726k -> 56728k
Overhead: 67009k -> 66947k
GGC runs: 554 -> 553
comparing Gerald's testcase PR8361 compilation at -O2 level:
Overall memory needed: 103757k -> 103761k
Peak memory use before GGC: 85639k
Peak memory use after GGC: 84691k
Maximum of released memory in single GGC run: 18904k
Garbage: 500939k -> 500759k
Leak: 57306k -> 57300k
Overhead: 76509k -> 76409k
GGC runs: 598 -> 597
comparing Gerald's testcase PR8361 compilation at -O3 level:
Overall memory needed: 111449k -> 111441k
Peak memory use before GGC: 92444k
Peak memory use after GGC: 85893k
Maximum of released memory in single GGC run: 19736k
Garbage: 520836k -> 520652k
Leak: 57641k -> 57632k
Overhead: 78275k -> 78248k
GGC runs: 579
Head of changelog is:
--- /usr/src/SpecTests/sandbox-britten-memory/x86_64/mem-result/ChangeLog 2004-12-31 23:58:19.000000000 +0000
+++ /usr/src/SpecTests/sandbox-britten-memory/gcc/gcc/ChangeLog 2005-01-01 06:51:21.000000000 +0000
@@ -1,3 +1,22 @@
+2004-12-31 Richard Henderson <rth@redhat.com>
+
+ PR tree-opt/19042
+ * tree-sra.c (decide_block_copy): Force use_block_copy false
+ for complex values.
+
+2004-12-31 Richard Henderson <rth@redhat.com>
+
+ PR middle-end/17799
+ * function.c (use_register_for_decl): Check DECL_IGNORED_P instead
+ of DECL_ARTIFICIAL.
+ (assign_parms_augmented_arg_list): Set DECL_IGNORED_P.
+ * c-decl.c (build_compound_literal): Likewise.
+ * dwarf2asm.c (dw2_force_const_mem): Likewise.
+ * gimplify.c (create_artificial_label): Likewise.
+ * tree-inline.c (expand_call_inline): Likewise.
+ * var-tracking.c (vt_initialize): Likewise.
+ * tree-outof-ssa.c (create_temp): Copy DECL_IGNORED_P.
+
2004-12-31 Andrew Pinski <pinskia@physics.uc.edu>
PR target/19211
--- /usr/src/SpecTests/sandbox-britten-memory/x86_64/mem-result/ChangeLog.cp 2004-12-31 05:02:47.000000000 +0000
+++ /usr/src/SpecTests/sandbox-britten-memory/gcc/gcc/cp/ChangeLog 2005-01-01 06:51:24.000000000 +0000
@@ -1,3 +1,14 @@
+2004-12-31 Richard Henderson <rth@redhat.com>
+
+ PR middle-end/17799
+ * call.c (make_temporary_var_for_ref_to_temp): Set DECL_IGNORED_P.
+ * class.c (build_vtable): Don't conditionallize setting it
+ based on DWARF2_DEBUG.
+ (layout_class_type): Set DECL_IGNORED_P.
+ * decl2.c (get_guard): Likewise.
+ * rtti.c (get_tinfo_decl, build_lang_decl): Likewise.
+ * tree.c (build_local_temp): Likewise.
+
2004-12-30 Mark Mitchell <mark@codesourcery.com>
* cp-tree.h (cp_declarator): Split "name" field into
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.