This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] Disable CSE skip-blocks


> On 10/29/06, Roger Sayle <roger@eyesopen.com> wrote:
> >Given that we all seem to be in perfect agreement, Steven's patch
> >is OK for mainline.  Hopefully, the more substantial clean-ups and
> >reorganizations will be contributed soon, now that we have consensus.
> 
> I've commited the patch, it is r118219.  I'll start submitting the
> clean-ups soon.

Hi,
it looks like that disabling CSE seems to increase memory consumption on
insn-attrtab by 16%.  Perhaps more allocnos?

Honza

comparing combine.c compilation at -O0 level:
    Overall memory needed: 28347k
    Peak memory use before GGC: 9282k
    Peak memory use after GGC: 8821k
    Maximum of released memory in single GGC run: 2666k
    Garbage: 36829k
    Leak: 6441k
    Overhead: 4856k
    GGC runs: 280

comparing combine.c compilation at -O1 level:
    Overall memory needed: 40195k
    Peak memory use before GGC: 17270k
    Peak memory use after GGC: 17094k
    Maximum of released memory in single GGC run: 2383k
    Garbage: 57448k
    Leak: 6482k
    Overhead: 6196k
    GGC runs: 355

comparing combine.c compilation at -O2 level:
    Overall memory needed: 29774k -> 29778k
    Peak memory use before GGC: 17266k
    Peak memory use after GGC: 17094k
    Maximum of released memory in single GGC run: 2916k -> 2900k
    Garbage: 76563k -> 76340k
    Leak: 6578k
    Overhead: 8786k -> 8741k
    GGC runs: 420

comparing combine.c compilation at -O3 level:
    Overall memory needed: 28882k
    Peak memory use before GGC: 18206k
    Peak memory use after GGC: 17822k
    Maximum of released memory in single GGC run: 4105k
    Garbage: 106666k -> 106171k
    Leak: 6653k
    Overhead: 12394k -> 12295k
    GGC runs: 471 -> 469

comparing insn-attrtab.c compilation at -O0 level:
    Overall memory needed: 88222k
    Peak memory use before GGC: 69766k
    Peak memory use after GGC: 44176k
    Maximum of released memory in single GGC run: 36964k
    Garbage: 129071k
    Leak: 9486k
    Overhead: 16989k
    GGC runs: 217

comparing insn-attrtab.c compilation at -O1 level:
    Overall memory needed: 114146k -> 114150k
    Peak memory use before GGC: 90352k
    Peak memory use after GGC: 83714k
    Maximum of released memory in single GGC run: 31806k
    Garbage: 277742k
    Leak: 9328k
    Overhead: 29771k
    GGC runs: 223

comparing insn-attrtab.c compilation at -O2 level:
  Overall memory allocated via mmap and sbrk increased from 111174k to 120258k, overall 8.17%
    Overall memory needed: 111174k -> 120258k
    Peak memory use before GGC: 92581k
    Peak memory use after GGC: 84694k
    Maximum of released memory in single GGC run: 30383k -> 30380k
    Garbage: 320900k -> 319046k
    Leak: 9330k
    Overhead: 37085k -> 36713k
    GGC runs: 250 -> 247

comparing insn-attrtab.c compilation at -O3 level:
  Overall memory allocated via mmap and sbrk increased from 111206k to 129334k, overall 16.30%
    Overall memory needed: 111206k -> 129334k
    Peak memory use before GGC: 92607k
    Peak memory use after GGC: 84719k
    Maximum of released memory in single GGC run: 30575k -> 30571k
    Garbage: 321528k -> 319694k
    Leak: 9333k
    Overhead: 37278k -> 36910k
    GGC runs: 254 -> 250

comparing Gerald's testcase PR8361 compilation at -O0 level:
    Overall memory needed: 119490k
    Peak memory use before GGC: 92635k
    Peak memory use after GGC: 91717k
    Maximum of released memory in single GGC run: 19299k
    Garbage: 205556k
    Leak: 47662k
    Overhead: 20811k
    GGC runs: 402

comparing Gerald's testcase PR8361 compilation at -O1 level:
    Overall memory needed: 119222k
    Peak memory use before GGC: 97821k
    Peak memory use after GGC: 95611k
    Maximum of released memory in single GGC run: 18569k
    Garbage: 444259k
    Leak: 49995k
    Overhead: 32811k
    GGC runs: 552

comparing Gerald's testcase PR8361 compilation at -O2 level:
    Overall memory needed: 119218k
    Peak memory use before GGC: 97820k
    Peak memory use after GGC: 95611k
    Maximum of released memory in single GGC run: 18569k
    Garbage: 508058k -> 505995k
    Leak: 50699k -> 50700k
    Overhead: 40860k -> 40452k
    GGC runs: 612 -> 609

comparing Gerald's testcase PR8361 compilation at -O3 level:
    Overall memory needed: 118922k
    Peak memory use before GGC: 97868k
    Peak memory use after GGC: 96898k
    Maximum of released memory in single GGC run: 18831k
    Garbage: 527424k -> 525377k
    Leak: 50275k -> 50275k
    Overhead: 41375k -> 40968k
    GGC runs: 626 -> 623

comparing PR rtl-optimization/28071 testcase compilation at -O0 level:
    Overall memory needed: 137934k
    Peak memory use before GGC: 81886k
    Peak memory use after GGC: 58766k
    Maximum of released memory in single GGC run: 45494k
    Garbage: 147250k
    Leak: 7507k
    Overhead: 25296k
    GGC runs: 83

comparing PR rtl-optimization/28071 testcase compilation at -O1 level:
    Overall memory needed: 424386k -> 424294k
    Peak memory use before GGC: 205249k
    Peak memory use after GGC: 201025k
    Maximum of released memory in single GGC run: 101716k
    Garbage: 271709k
    Leak: 47573k
    Overhead: 30825k
    GGC runs: 101

comparing PR rtl-optimization/28071 testcase compilation at -O2 level:
    Overall memory needed: 352226k -> 352134k
    Peak memory use before GGC: 206000k
    Peak memory use after GGC: 201776k
    Maximum of released memory in single GGC run: 108041k
    Garbage: 350431k
    Leak: 48156k
    Overhead: 46270k
    GGC runs: 108

comparing PR rtl-optimization/28071 testcase compilation at -O3 -fno-tree-pre -fno-tree-fre level:
    Overall memory needed: 535282k -> 535278k
    Peak memory use before GGC: 314907k
    Peak memory use after GGC: 293250k
    Maximum of released memory in single GGC run: 163448k
    Garbage: 491201k -> 491202k
    Leak: 65488k
    Overhead: 59087k -> 59087k
    GGC runs: 95

Head of the ChangeLog is:

--- /usr/src/SpecTests/sandbox-britten-memory/x86_64/mem-result/ChangeLog	2006-10-31 01:47:23.000000000 +0000
+++ /usr/src/SpecTests/sandbox-britten-memory/gcc/gcc/ChangeLog	2006-10-31 13:14:53.000000000 +0000
@@ -1,3 +1,19 @@
+2006-10-31  Richard Guenther  <rguenther@suse.de>
+
+	* config/i386/i386.md (asindf2, asinsf2, asinxf2, acosdf2,
+	acossf2, acosxf2, log1psf2, log1pdf2, log1pxf2, ilogbsi2,
+	expsf2, expdf2, expxf2, exp10sf2, exp10df2, exp10xf2,
+	exp2sf2, exp2df2, exp2xf2, expm1df2, expm1sf2, expm1xf2,
+	ldexpdf3, ldexpsf3, ldexpxf3, rintxf2, rintdf2, rintsf2,
+	lround<mode>di2, lround<mode>si2, floorxf2, floordf2, floorsf2,
+	lfloor<mode>di2, lfloor<mode>si2, ceilxf2, ceildf2, ceilsf2,
+	btruncxf2, btruncdf2, btruncsf2): Conditionalize expansion on
+	!optimize_size.
+
+2006-10-31  Steven Bosscher  <steven@gcc.gnu.org>
+
+	* opts.c (decode_options): Disable CSE skip blocks.
+
 2006-10-30  Dirk Mueller  <dmueller@suse.de>
 
 	* c-common.h (strict_aliasing_warning): Fix whitespace.


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.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]