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: Mon, 22 Sep 2008 14:32:55 +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 empty function compilation at -O0 level:
Overall memory needed: 8184k
Peak memory use before GGC: 1291k
Peak memory use after GGC: 1217k
Maximum of released memory in single GGC run: 134k
Garbage: 218k
Leak: 1221k
Overhead: 136k
GGC runs: 4
Pre-IPA-Garbage: 207k
Pre-IPA-Leak: 1224k
Pre-IPA-Overhead: 135k
Post-IPA-Garbage: 207k
Post-IPA-Leak: 1224k
Post-IPA-Overhead: 135k
comparing empty function compilation at -O0 -g level:
Overall memory needed: 8428k
Peak memory use before GGC: 1319k
Peak memory use after GGC: 1245k
Maximum of released memory in single GGC run: 133k
Garbage: 220k
Leak: 1254k
Overhead: 141k
GGC runs: 4
Pre-IPA-Garbage: 207k
Pre-IPA-Leak: 1224k
Pre-IPA-Overhead: 135k
Post-IPA-Garbage: 207k
Post-IPA-Leak: 1224k
Post-IPA-Overhead: 135k
comparing empty function compilation at -O1 level:
Overall memory needed: 8240k
Peak memory use before GGC: 1291k
Peak memory use after GGC: 1217k
Maximum of released memory in single GGC run: 134k
Garbage: 221k
Leak: 1221k
Overhead: 137k
GGC runs: 4
Pre-IPA-Garbage: 207k
Pre-IPA-Leak: 1224k
Pre-IPA-Overhead: 135k
Post-IPA-Garbage: 207k
Post-IPA-Leak: 1224k
Post-IPA-Overhead: 135k
comparing empty function compilation at -O2 level:
Overall memory needed: 8452k
Peak memory use before GGC: 1291k
Peak memory use after GGC: 1218k
Maximum of released memory in single GGC run: 135k
Garbage: 226k
Leak: 1221k
Overhead: 138k
GGC runs: 4
Pre-IPA-Garbage: 207k
Pre-IPA-Leak: 1224k
Pre-IPA-Overhead: 135k
Post-IPA-Garbage: 207k
Post-IPA-Leak: 1224k
Post-IPA-Overhead: 135k
comparing empty function compilation at -O3 level:
Overall memory needed: 8456k
Peak memory use before GGC: 1291k
Peak memory use after GGC: 1218k
Maximum of released memory in single GGC run: 135k
Garbage: 226k
Leak: 1221k
Overhead: 138k
GGC runs: 4
Pre-IPA-Garbage: 207k
Pre-IPA-Leak: 1224k
Pre-IPA-Overhead: 135k
Post-IPA-Garbage: 207k
Post-IPA-Leak: 1224k
Post-IPA-Overhead: 135k
comparing combine.c compilation at -O0 level:
Amount of produced GGC garbage increased from 39088k to 39162k, overall 0.19%
Overall memory needed: 32044k
Peak memory use before GGC: 18004k
Peak memory use after GGC: 17791k
Maximum of released memory in single GGC run: 1825k -> 1824k
Garbage: 39088k -> 39162k
Leak: 5792k
Overhead: 5267k
GGC runs: 337
Pre-IPA-Garbage: 12430k
Pre-IPA-Leak: 19386k
Pre-IPA-Overhead: 2624k
Post-IPA-Garbage: 12430k
Post-IPA-Leak: 19386k
Post-IPA-Overhead: 2624k
comparing combine.c compilation at -O0 -g level:
Amount of produced GGC garbage increased from 39376k to 39450k, overall 0.19%
Overall memory needed: 34072k
Peak memory use before GGC: 19904k -> 19910k
Peak memory use after GGC: 19646k -> 19647k
Maximum of released memory in single GGC run: 1834k -> 1833k
Garbage: 39376k -> 39450k
Leak: 9082k
Overhead: 6084k -> 6084k
GGC runs: 320
Pre-IPA-Garbage: 12529k
Pre-IPA-Leak: 21659k
Pre-IPA-Overhead: 3114k
Post-IPA-Garbage: 12529k
Post-IPA-Leak: 21659k
Post-IPA-Overhead: 3114k
comparing combine.c compilation at -O1 level:
Amount of produced GGC garbage increased from 46537k to 46603k, overall 0.14%
Overall memory needed: 30796k -> 30792k
Peak memory use before GGC: 15651k
Peak memory use after GGC: 15482k
Maximum of released memory in single GGC run: 1393k -> 1380k
Garbage: 46537k -> 46603k
Leak: 5775k
Overhead: 6087k -> 6087k
GGC runs: 401
Pre-IPA-Garbage: 13169k
Pre-IPA-Leak: 16872k
Pre-IPA-Overhead: 2532k
Post-IPA-Garbage: 13169k
Post-IPA-Leak: 16872k
Post-IPA-Overhead: 2532k
comparing combine.c compilation at -O2 level:
Amount of produced GGC garbage increased from 60584k to 60652k, overall 0.11%
Overall memory needed: 31224k -> 31480k
Peak memory use before GGC: 15794k
Peak memory use after GGC: 15633k
Maximum of released memory in single GGC run: 1356k -> 1357k
Garbage: 60584k -> 60652k
Leak: 5960k
Overhead: 8146k -> 8146k
GGC runs: 470
Pre-IPA-Garbage: 13336k
Pre-IPA-Leak: 16952k
Pre-IPA-Overhead: 2553k
Post-IPA-Garbage: 13336k
Post-IPA-Leak: 16952k
Post-IPA-Overhead: 2553k
comparing combine.c compilation at -O3 level:
Overall memory needed: 31924k -> 31768k
Peak memory use before GGC: 15827k -> 15828k
Peak memory use after GGC: 15668k -> 15669k
Maximum of released memory in single GGC run: 1794k
Garbage: 76025k -> 76095k
Leak: 5825k
Overhead: 9934k -> 9934k
GGC runs: 506
Pre-IPA-Garbage: 13336k
Pre-IPA-Leak: 16960k
Pre-IPA-Overhead: 2554k
Post-IPA-Garbage: 13336k
Post-IPA-Leak: 16960k
Post-IPA-Overhead: 2554k
comparing insn-attrtab.c compilation at -O0 level:
Amount of produced GGC garbage increased from 103547k to 103735k, overall 0.18%
Overall memory needed: 157640k
Peak memory use before GGC: 65250k -> 65310k
Peak memory use after GGC: 53351k -> 53354k
Maximum of released memory in single GGC run: 16211k
Garbage: 103547k -> 103735k
Leak: 35563k
Overhead: 16157k -> 16157k
GGC runs: 250 -> 251
Pre-IPA-Garbage: 38224k
Pre-IPA-Leak: 55908k
Pre-IPA-Overhead: 8653k
Post-IPA-Garbage: 38224k
Post-IPA-Leak: 55908k
Post-IPA-Overhead: 8653k
comparing insn-attrtab.c compilation at -O0 -g level:
Amount of produced GGC garbage increased from 104027k to 104215k, overall 0.18%
Overall memory needed: 158916k -> 158920k
Peak memory use before GGC: 66524k -> 66584k
Peak memory use after GGC: 54623k -> 54626k
Maximum of released memory in single GGC run: 16212k
Garbage: 104027k -> 104215k
Leak: 37213k
Overhead: 16613k -> 16613k
GGC runs: 245
Pre-IPA-Garbage: 38281k
Pre-IPA-Leak: 57449k
Pre-IPA-Overhead: 8989k
Post-IPA-Garbage: 38281k
Post-IPA-Leak: 57449k
Post-IPA-Overhead: 8989k
comparing insn-attrtab.c compilation at -O1 level:
Peak amount of GGC memory allocated before garbage collecting increased from 50314k to 50384k, overall 0.14%
Peak amount of GGC memory still allocated after garbage collecting increased from 43412k to 43482k, overall 0.16%
Amount of produced GGC garbage increased from 180689k to 180876k, overall 0.10%
Overall memory needed: 134672k -> 134668k
Peak memory use before GGC: 50314k -> 50384k
Peak memory use after GGC: 43412k -> 43482k
Maximum of released memory in single GGC run: 22952k
Garbage: 180689k -> 180876k
Leak: 8074k
Overhead: 25327k -> 25327k
GGC runs: 302
Pre-IPA-Garbage: 43202k
Pre-IPA-Leak: 43434k
Pre-IPA-Overhead: 8052k
Post-IPA-Garbage: 43202k
Post-IPA-Leak: 43434k
Post-IPA-Overhead: 8052k
comparing insn-attrtab.c compilation at -O2 level:
Overall memory needed: 149892k -> 149896k
Peak memory use before GGC: 50318k -> 50361k
Peak memory use after GGC: 44407k -> 44450k
Maximum of released memory in single GGC run: 21566k -> 21565k
Garbage: 212637k -> 212770k
Leak: 8059k
Overhead: 31010k -> 31009k
GGC runs: 335 -> 334
Pre-IPA-Garbage: 43275k
Pre-IPA-Leak: 43440k
Pre-IPA-Overhead: 8061k
Post-IPA-Garbage: 43275k
Post-IPA-Leak: 43440k
Post-IPA-Overhead: 8061k
comparing insn-attrtab.c compilation at -O3 level:
Overall memory needed: 164860k -> 164900k
Peak memory use before GGC: 61946k -> 61989k
Peak memory use after GGC: 58846k -> 58889k
Maximum of released memory in single GGC run: 23864k -> 23713k
Garbage: 242888k -> 243015k
Leak: 8067k
Overhead: 34454k -> 34452k
GGC runs: 341 -> 338
Pre-IPA-Garbage: 43275k
Pre-IPA-Leak: 43440k
Pre-IPA-Overhead: 8061k
Post-IPA-Garbage: 43275k
Post-IPA-Leak: 43440k
Post-IPA-Overhead: 8061k
comparing Gerald's testcase PR8361 compilation at -O0 level:
Overall memory needed: 150608k -> 150600k
Peak memory use before GGC: 82071k -> 82073k
Peak memory use after GGC: 81237k
Maximum of released memory in single GGC run: 15776k
Garbage: 206366k -> 206460k
Leak: 50844k
Overhead: 27261k -> 27261k
GGC runs: 416 -> 417
Pre-IPA-Garbage: 112337k
Pre-IPA-Leak: 87316k
Pre-IPA-Overhead: 14978k
Post-IPA-Garbage: 112337k
Post-IPA-Leak: 87316k
Post-IPA-Overhead: 14978k
comparing Gerald's testcase PR8361 compilation at -O0 -g level:
Overall memory needed: 168112k
Peak memory use before GGC: 95767k
Peak memory use after GGC: 94819k
Maximum of released memory in single GGC run: 16204k
Garbage: 212057k -> 212150k
Leak: 77406k
Overhead: 33925k -> 33925k
GGC runs: 389
Pre-IPA-Garbage: 112959k
Pre-IPA-Leak: 103831k
Pre-IPA-Overhead: 18481k
Post-IPA-Garbage: 112959k
Post-IPA-Leak: 103831k
Post-IPA-Overhead: 18481k
comparing Gerald's testcase PR8361 compilation at -O1 level:
Overall memory needed: 110807k -> 110799k
Peak memory use before GGC: 82820k -> 82822k
Peak memory use after GGC: 81998k
Maximum of released memory in single GGC run: 16084k
Garbage: 285751k -> 285909k
Leak: 47479k
Overhead: 32504k -> 32504k
GGC runs: 508 -> 509
Pre-IPA-Garbage: 163112k
Pre-IPA-Leak: 86583k
Pre-IPA-Overhead: 20472k
Post-IPA-Garbage: 163112k
Post-IPA-Leak: 86583k
Post-IPA-Overhead: 20472k
comparing Gerald's testcase PR8361 compilation at -O2 level:
Overall memory needed: 112303k -> 112315k
Peak memory use before GGC: 84293k -> 84296k
Peak memory use after GGC: 83457k
Maximum of released memory in single GGC run: 16069k
Garbage: 339883k -> 340013k
Leak: 47518k
Overhead: 39048k -> 39048k
GGC runs: 577 -> 576
Pre-IPA-Garbage: 167129k
Pre-IPA-Leak: 86920k
Pre-IPA-Overhead: 20965k
Post-IPA-Garbage: 167129k
Post-IPA-Leak: 86920k
Post-IPA-Overhead: 20965k
comparing Gerald's testcase PR8361 compilation at -O3 level:
Overall memory needed: 112699k
Peak memory use before GGC: 84920k -> 84912k
Peak memory use after GGC: 84071k
Maximum of released memory in single GGC run: 16069k
Garbage: 371223k -> 371366k
Leak: 47521k
Overhead: 42270k -> 42270k
GGC runs: 608
Pre-IPA-Garbage: 167209k
Pre-IPA-Leak: 87592k
Pre-IPA-Overhead: 21020k
Post-IPA-Garbage: 167209k
Post-IPA-Leak: 87592k
Post-IPA-Overhead: 21020k
comparing PR rtl-optimization/28071 testcase compilation at -O0 level:
Overall memory needed: 371176k -> 371204k
Peak memory use before GGC: 81875k -> 81897k
Peak memory use after GGC: 52810k -> 52831k
Maximum of released memory in single GGC run: 38147k -> 38169k
Garbage: 141327k -> 141349k
Leak: 10656k
Overhead: 24891k
GGC runs: 86
Pre-IPA-Garbage: 12562k
Pre-IPA-Leak: 20190k
Pre-IPA-Overhead: 2242k
Post-IPA-Garbage: 12562k
Post-IPA-Leak: 20190k
Post-IPA-Overhead: 2242k
comparing PR rtl-optimization/28071 testcase compilation at -O0 -g level:
Overall memory needed: 372056k -> 372104k
Peak memory use before GGC: 82572k -> 82594k
Peak memory use after GGC: 53506k -> 53528k
Maximum of released memory in single GGC run: 38132k -> 38154k
Garbage: 141428k -> 141450k
Leak: 12698k
Overhead: 25474k
GGC runs: 92
Pre-IPA-Garbage: 12570k
Pre-IPA-Leak: 20439k
Pre-IPA-Overhead: 2296k
Post-IPA-Garbage: 12570k
Post-IPA-Leak: 20439k
Post-IPA-Overhead: 2296k
comparing PR rtl-optimization/28071 testcase compilation at -O1 level:
Overall memory needed: 230260k
Peak memory use before GGC: 73624k
Peak memory use after GGC: 66143k
Maximum of released memory in single GGC run: 34721k
Garbage: 219596k -> 219628k
Leak: 11158k
Overhead: 30799k
GGC runs: 97
Pre-IPA-Garbage: 48349k
Pre-IPA-Leak: 63525k
Pre-IPA-Overhead: 8965k
Post-IPA-Garbage: 48349k
Post-IPA-Leak: 63525k
Post-IPA-Overhead: 8965k
comparing PR rtl-optimization/28071 testcase compilation at -O2 level:
Overall memory needed: 362224k -> 362140k
Peak memory use before GGC: 73624k
Peak memory use after GGC: 66143k
Maximum of released memory in single GGC run: 36043k
Garbage: 247851k -> 247883k
Leak: 10980k
Overhead: 36877k
GGC runs: 106
Pre-IPA-Garbage: 109693k
Pre-IPA-Leak: 73929k
Pre-IPA-Overhead: 15408k
Post-IPA-Garbage: 109693k
Post-IPA-Leak: 73929k
Post-IPA-Overhead: 15408k
comparing PR rtl-optimization/28071 testcase compilation at -O3 -fno-tree-pre -fno-tree-fre level:
Overall memory needed: 1024552k -> 1024116k
Peak memory use before GGC: 139561k
Peak memory use after GGC: 126838k
Maximum of released memory in single GGC run: 62382k -> 62381k
Garbage: 365693k -> 365724k
Leak: 6759k
Overhead: 45401k
GGC runs: 104 -> 105
Pre-IPA-Garbage: 109693k
Pre-IPA-Leak: 73929k
Pre-IPA-Overhead: 15408k
Post-IPA-Garbage: 109693k
Post-IPA-Leak: 73929k
Post-IPA-Overhead: 15408k
Head of the ChangeLog is:
--- /usr/src/SpecTests/sandbox-britten-memory/x86_64/mem-result/ChangeLog 2008-09-21 18:28:45.000000000 +0000
+++ /usr/src/SpecTests/sandbox-britten-memory/gcc/gcc/ChangeLog 2008-09-22 11:03:42.000000000 +0000
@@ -1,3 +1,78 @@
+2008-09-22 Ira Rosen <irar@il.ibm.com>
+
+ PR tree-optimization/37482
+ * tree-vectorizer.h (struct _slp_instance): Add new field.
+ (SLP_INSTANCE_FIRST_LOAD_STMT): New.
+ (get_earlier_stmt): New function.
+ * tree-vect-analyze.c (vect_find_first_load_in_slp_instance): New
+ function.
+ (vect_analyze_slp_instance): Set SLP_INSTANCE_FIRST_LOAD_STMT.
+ * tree-vect-transform.c (vect_finish_stmt_generation): Remove the
+ asserts that GSI points to the scalar statement being vectorized.
+ Set new statement location according to GSI.
+ (vect_schedule_slp_instance): Use GSI of
+ SLP_INSTANCE_FIRST_LOAD_STMT when vectorizing loads.
+
+2008-09-21 Jan Hubicka <jh@suse.cz>
+
+ * ipa-cp.c (ipcp_estimate_growth): Check recursive calls.
+ (ipcp_insert_stage): Update dead_nodes bitmap.
+
+2008-09-22 Danny Smith <dannysmith@users.sourceforge.net>
+
+ PR target/37528
+ * config/i386/t-cygming (SHLIB_LC): Remove.
+ (SHLIB_LINK): Don't add static objects to SHLIB_IMPLIB
+ * config/i386/t-cygwin (SHLIB_LC): Specify all required
+ libraries.
+
+2008-09-22 Hans-Peter Nilsson <hp@axis.com>
+
+ PR middle-end/37170
+ PR middle-end/37280
+ * final.c (mark_symbol_ref_as_used): New helper function.
+ (output_operand): Instead of just looking inside MEMs for
+ SYMBOL_REFs, use new helper function and for_each_rtx.
+ * varasm.c (assemble_external): Move #ifndef ASM_OUTPUT_EXTERNAL
+ to after weak-handling. Don't mark decls with TREE_STATIC as weak.
+ Make head comment more general.
+ * config/darwin.c (machopic_output_indirection): Handle weak
+ references here, like in assemble_external.
+
+2008-09-21 Eric Botcazou <ebotcazou@adacore.com>
+
+ * config/sparc/sparc-protos.h (gen_compare_operator): Declare.
+ (sparc_emit_float_lib_cmp): Change return type.
+ * config/sparc/sparc.c (gen_compare_reg): Add comment about TFmode.
+ (gen_compare_operator): New function.
+ (sparc_emit_float_lib_cmp): Return the new operator to be used in
+ the comparison sequence. Minor tweaks.
+ * config/sparc/sparc.md (seq, sne, sgt, slt, sge, sle): Assert
+ that the final operator and the result of sparc_emit_float_lib_cmp
+ match for software TFmode; use emit_insn in lieu of emit_jump_insn.
+ (beq, bne, bgt, blt, bge, ble, bunordered, bordered, bungt, bunlt,
+ buneq, bunge, bunle, bltgt): Assert that the final operator and the
+ result of sparc_emit_float_lib_cmp match for software TFmode.
+ (movqicc, movhicc, movsicc, movdicc): Merge into...
+ (mov<I:mode>cc): ...this.
+ (movsfcc, movdfcc, movtfcc): Merge into...
+ (mov<F:mode>cc): ...this.
+ (movqi_cc_sp64, movhi_cc_sp64, movsi_cc_sp64, movdi_cc_sp64): Merge
+ into...
+ (mov<I:mode>_cc_v9): ...this.
+ (movdi_cc_sp64_trunc): Delete.
+ (movqi_cc_reg_sp64, movhi_cc_reg_sp64, movsi_cc_reg_sp64,
+ movdi_cc_reg_sp64): Merge into...
+ (mov<I:mode>_cc_reg_sp64): ...this.
+ (movsf_cc_sp64): Rename into...
+ (movsf_cc_v9): ...this.
+ (movdf_cc_sp64): Rename into...
+ (movdf_cc_v9): ...this.
+ (movtf_cc_hq_sp64): Rename into...
+ (movtf_cc_hq_v9): ...this.
+ (movtf_cc_sp64): Rename into...
+ (movtf_cc_v9): ...this. Adjust for renaming of movdf_cc_sp64.
+
2008-09-21 Diego Novillo <dnovillo@google.com>
* doc/gccint.texi: Include generic.texi and gimple.texi.
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.