This is the mail archive of the
gcc-regression@gcc.gnu.org
mailing list for the GCC project.
A recent patch increased GCC's memory consumption!
- From: gcctest at suse dot de
- To: jh at suse dot cz, gcc-regression at gcc dot gnu dot org
- Date: Thu, 09 Feb 2006 07:02:54 +0000
- Subject: A recent patch increased GCC's memory consumption!
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: 25320k
Peak memory use before GGC: 9566k
Peak memory use after GGC: 8914k
Maximum of released memory in single GGC run: 2649k
Garbage: 40064k -> 40057k
Leak: 6740k
Overhead: 5738k -> 5738k
GGC runs: 313
comparing combine.c compilation at -O1 level:
Overall memory needed: 26900k
Peak memory use before GGC: 17435k
Peak memory use after GGC: 17256k
Maximum of released memory in single GGC run: 2308k
Garbage: 62427k -> 62424k
Leak: 6890k
Overhead: 7487k -> 7487k
GGC runs: 392
comparing combine.c compilation at -O2 level:
Overall memory needed: 26900k
Peak memory use before GGC: 17438k
Peak memory use after GGC: 17256k
Maximum of released memory in single GGC run: 2399k
Garbage: 82367k -> 82367k
Leak: 6976k
Overhead: 10226k
GGC runs: 461
comparing combine.c compilation at -O3 level:
Overall memory needed: 26900k
Peak memory use before GGC: 18453k
Peak memory use after GGC: 17993k
Maximum of released memory in single GGC run: 3491k -> 3495k
Garbage: 113575k -> 113575k
Leak: 7049k -> 7053k
Overhead: 13950k -> 13950k
GGC runs: 518
comparing insn-attrtab.c compilation at -O0 level:
Amount of memory still referenced at the end of compilation increased from 9892k to 9908k, overall 0.16%
Overall memory needed: 80948k
Peak memory use before GGC: 69508k
Peak memory use after GGC: 45045k
Maximum of released memory in single GGC run: 36220k
Garbage: 146687k -> 146686k
Leak: 9892k -> 9908k
Overhead: 19751k -> 19751k
GGC runs: 247
comparing insn-attrtab.c compilation at -O1 level:
Overall memory needed: 109264k
Peak memory use before GGC: 92087k -> 92085k
Peak memory use after GGC: 81246k -> 81244k
Maximum of released memory in single GGC run: 32383k
Garbage: 296338k -> 296341k
Leak: 10076k
Overhead: 36186k -> 36186k
GGC runs: 245 -> 246
comparing insn-attrtab.c compilation at -O2 level:
Overall memory needed: 110872k
Peak memory use before GGC: 93801k -> 93803k
Peak memory use after GGC: 81341k -> 81343k
Maximum of released memory in single GGC run: 31927k
Garbage: 343618k -> 343615k
Leak: 10059k
Overhead: 44783k -> 44783k
GGC runs: 274 -> 275
comparing insn-attrtab.c compilation at -O3 level:
Overall memory needed: 110836k -> 110840k
Peak memory use before GGC: 93831k -> 93833k
Peak memory use after GGC: 81371k -> 81373k
Maximum of released memory in single GGC run: 32267k
Garbage: 344268k -> 344265k
Leak: 10063k
Overhead: 44982k -> 44982k
GGC runs: 279 -> 280
comparing Gerald's testcase PR8361 compilation at -O0 level:
Overall memory needed: 118424k
Peak memory use before GGC: 95198k
Peak memory use after GGC: 94250k
Maximum of released memory in single GGC run: 20356k
Garbage: 223714k
Leak: 49472k
Overhead: 37113k -> 37113k
GGC runs: 369
comparing Gerald's testcase PR8361 compilation at -O1 level:
Overall memory needed: 108528k
Peak memory use before GGC: 95331k
Peak memory use after GGC: 93251k
Maximum of released memory in single GGC run: 20233k
Garbage: 564794k -> 564802k
Leak: 52294k -> 52294k
Overhead: 63643k -> 63643k
GGC runs: 531
comparing Gerald's testcase PR8361 compilation at -O2 level:
Overall memory needed: 108532k
Peak memory use before GGC: 95331k
Peak memory use after GGC: 93252k
Maximum of released memory in single GGC run: 20232k
Garbage: 687278k -> 687276k
Leak: 53428k
Overhead: 76837k -> 76835k
GGC runs: 618 -> 617
comparing Gerald's testcase PR8361 compilation at -O3 level:
Overall memory needed: 114212k -> 114208k
Peak memory use before GGC: 96797k
Peak memory use after GGC: 94327k
Maximum of released memory in single GGC run: 20729k
Garbage: 753507k -> 753507k
Leak: 54360k
Overhead: 81753k -> 81753k
GGC runs: 629
Head of the ChangeLog is:
--- /usr/src/SpecTests/sandbox-britten-memory/x86_64/mem-result/ChangeLog 2006-02-08 19:27:04.000000000 +0000
+++ /usr/src/SpecTests/sandbox-britten-memory/gcc/gcc/ChangeLog 2006-02-09 05:53:25.000000000 +0000
@@ -1,5 +1,66 @@
2006-02-08 Jeff Law <law@redhat.com>
+ PR tree-optimization/21417
+ * tree-ssa-threadedge.c (thread_across_edge): Reject threading
+ across a backedge if the control statement at the end of the
+ block is data dependent on other statements in the same block.
+ (record_temporary_equivalences_from_stmts): Remove over-conservative
+ test for threading across backedges.
+
+ * tree-ssa-dce.c (remove_dead_statement): When removing a dead
+ control statement, handle all three special cases in the same
+ manner as they all have the same desired solution. No longer
+ special case edge flag updating when the post dominator is
+ the exit block.
+
+2006-02-08 Roger Sayle <roger@eyesopen.com>
+
+ * tree.def (CHAR_TYPE): Remove.
+ * dbxout.c (dbxout_type): Delete dead handling of CHAR_TYPE.
+ * dwarf2out.c (base_type_die, is_base_type): Likewise.
+ (gen_type_die) <ARRAY_TYPE>: Remove handling of CHAR_TYPE arrays.
+ (gen_string_type_die): Delete unreachable function.
+
+ * tree-pretty-print.c (dump_generic_node): Don't handle CHAR_TYPE.
+ * tree.c (build_int_cst_wide, type_contains_placeholder_1,
+ type_hash_eq, variably_modified_type_p, walk_type_fields): Likewise.
+ * tree.h (NUMERICAL_TYPE_CHECK, INTEGRAL_TYPE_P): Likewise.
+ * builtins.c (type_to_class): Likewise.
+ * fold-const.c (fold_convert, build_range_check,
+ merge_ranges): Likewise.
+ * expr.c (count_type_elements): Likewise.
+ * c-pretty-print.c (pp_c_type_specifier): Likewise.
+ * stor-layout.c (layout_type): Likewise.
+ * ipa-type-escape.c (type_to_consider): Likewise.
+ * gimplify.c (omp_firstprivatize_type_sizes,
+ gimplify_type_sizes): Likewise.
+ * explow.c (promote_mode): Likewise.
+ * tree-sra.c (is_sra_scalar_type): Likewise.
+ * varasm.c (output_constant): Likewise.
+ * tree-inline.c (remap_type_1): Likewise.
+ * convert.c (convert_to_pointer, convert_to_real,
+ convert_to_integer, convert_to_complex): Likewise.
+
+2006-02-08 Daniel Berlin <dberlin@dberlin.org>
+
+ * df-scan.c (df_def_record_1): A set destination may
+ be a parallel regardless of mode.
+
+2006-02-08 Roger Sayle <roger@eyesopen.com>
+
+ PR target/22209
+ * config/mips/mips.h (MIN_UNITS_PER_WORD): Don't define for libgcc.
+ * config/mips/_tilib.c: Remove.
+ * config/fixtfdi.c: New libgcc source file.
+ * config/fixunstfdi.c: New source file.
+ * config/floatditf.c: New source file.
+ * config/floatunditf.c: New souce file.
+ * config/mips/t-iris6 (LIB2FUNCS_EXTRA): Include the new source
+ files above instead of config/mips/_tilib.c.
+ * config/mips/t-linux64 (LIB2FUNCS_EXTRA): Likewise.
+
+2006-02-08 Jeff Law <law@redhat.com>
+
PR tree-optimization/26169
* tree-vrp.c (execute_vrp): Perform any queued SSA updates before
threading jumps.
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.