Memory utilization of cc1plus blows up compiling tFile.cc under hpux
John David Anglin
dave@hiauly1.hia.nrc.ca
Wed Oct 4 14:30:00 GMT 2000
I was trying to make some progress in understanding why the performance
of cc1plus under hpux 10.20 sometimes is terrible. It has been noted in
the past on this list that various c++ tests in the libio testsuite fail
because of timeouts.
I added "-Q -fmem-report" to try and get some statistics on what was
happening. Initially, I just added -fmem-report and the utilization
figures printed out looked similar to those on a i686 linux system.
However, it was clear from watching the memory usage of cc1plus during
the compilation that the actual memory utilization the final values.
I have attached a log of the compilation of tFile.cc below. Note that
the GC memory allocated balloons to 35762k during the compilation of the
second instance of void t7(). Under i686 linux, there is just one garbage
collection at this point (void t7() {GC 5325k -> 2166k}). Thus, for
some reason the memory required to compile tFile.cc under i686 linux
is much less.
Note also that there are some very strange wall times and percentages printed
(e.g., a total wall time of -1603.-96).
Dave
--
J. David Anglin dave.anglin@nrc.ca
National Research Council of Canada (613) 990-0752 (FAX: 952-6605)
Running ../../../../libio/testsuite/libio.tests/tFile.exp ...
Executing on host: /xxx/gnu/gcc-2.97/objdir/gcc/xgcc -B/xxx/gnu/gcc-2.97/objdir/gcc/ ../../../../libio/testsuite/../tests/tFile.cc -ansi -D_XOPEN_SOURCE=500 -O3 -I.. -I../../../../libio/testsuite/.. -nostdinc++ -I/xxx/gnu/gcc-2.97/libstdc++ -I/xxx/gnu/gcc-2.97/libstdc++/stl -Q -fmem-report -L/xxx/gnu/gcc-2.97/objdir/hppa1.1-hp-hpux10.20//libstdc++ -L/xxx/gnu/gcc-2.97/objdir/hppa1.1-hp-hpux10.20//libiberty -L/xxx/gnu/gcc-2.97/objdir/hppa1.1-hp-hpux10.20//librx -L/xxx/gnu/gcc-2.97/objdir/hppa1.1-hp-hpux10.20//libiberty -g -lstdc++ -liberty -lm -o /xxx/gnu/gcc-2.97/objdir/hppa1.1-hp-hpux10.20/libio/testsuite/tFile (timeout = -1)
ostream *ios::tie() const ostream *ios::tie(ostream *) short int ios::fill() const short int ios::fill(short int) __fmtflags ios::flags() const __fmtflags ios::flags(long unsigned int) int ios::precision() const int ios::precision(int) __fmtflags ios::setf(long unsigned int) __fmtflags ios::setf(long unsigned int, long unsigned int) __fmtflags ios::unsetf(long unsigned int) int ios::width() const int ios::width(int) void ios::_throw_failure() const void ios::clear(int) void ios::set(int) void ios::setstate(int) int ios::good() const int ios::eof() const int ios::fail() const int ios::bad() const int ios::rdstate() const ios::operator void *() const int ios::operator!() const int ios::exceptions() const void ios::exceptions(int) streambuf *ios::rdbuf() const streambuf *ios::rdbuf(streambuf *) void ios::sync_with_stdio() void ios::unset(state_value) ios::Init::Init() void streammarker::set_offset(int) int streammarker::saving() const void *&streambuf::_vtable() _IO_FILE *&strea!
mbuf::xchain() char *streambuf::gptr() const char *streambuf::pptr() const char *streambuf::egptr() const char *streambuf::epptr() const char *streambuf::pbase() const char *streambuf::eback() const char *streambuf::base() const char *streambuf::ebuf() const int streambuf::blen() const void streambuf::xput_char(char) int streambuf::xflags() int streambuf::xflags(int) void streambuf::xsetflags(int) void streambuf::xsetflags(int, int) void streambuf::gbump(int) void streambuf::pbump(int) void streambuf::setp(char *, char *) void streambuf::setg(char *, char *, char *) char *streambuf::shortbuf() int streambuf::in_backup() char *streambuf::Gbase() char *streambuf::eGptr() char *streambuf::Bbase() char *streambuf::Bptr() char *streambuf::eBptr() char *streambuf::Nbase() char *streambuf::eNptr() int streambuf::have_backup() int streambuf::have_markers() int streambuf::put_mode() streampos streambuf::pubseekoff(int, ios::seek_dir, int) streampos streambuf::pubseekpos(int, int) int s!
treambuf::unbuffered() int streambuf::linebuffered() void streambuf::unbuffered(int) void streambuf::linebuffered(int) int streambuf::allocate() void streambuf::allocbuf() int streambuf::in_avail() int streambuf::out_waiting() streamsize streambuf::sputn(const char *, int) streamsize streambuf::padn(char, int) streamsize streambuf::sgetn(char *, int) int streambuf::sputc(int) int streambuf::sbumpc() int streambuf::sgetc() int streambuf::snextc() void streambuf::stossc() int filebuf::is_open() const int filebuf::fd() const int filebuf::is_reading() char *filebuf::cur_ptr() char *filebuf::file_ptr() void ios::init(streambuf *, ostream *) ios::ios(streambuf *, ostream *) ios::~ios() ostream::ostream() int ostream::opfx() void ostream::osfx() ostream &ostream::put(char) ostream &ostream::put(unsigned char) ostream &ostream::put(signed char) ostream &ostream::write(const unsigned char *, int) ostream &ostream::write(const signed char *, int) ostream &ostream::write(const void *, in!
t) ostream &ostream::operator<<(unsigned char) ostream &ostream::operator<<(signed char) ostream &ostream::operator<<(const unsigned char *) ostream &ostream::operator<<(const signed char *) ostream &ostream::operator<<(short int) ostream &ostream::operator<<(short unsigned int) ostream &ostream::operator<<(bool) ostream &ostream::operator<<(float) ostream &ostream::operator<<(long double) ostream &ostream::operator<<(ostream &(*)(ostream &)) ostream &ostream::operator<<(ios &(*)(ios &)) streambuf *ostream::ostreambuf() const istream::istream() istream &istream::get(unsigned char *, int, char) istream &istream::get(unsigned char &) istream &istream::getline(unsigned char *, int, char) istream &istream::get(signed char &) istream &istream::get(signed char *, int, char) istream &istream::getline(signed char *, int, char) istream &istream::read(unsigned char *, int) istream &istream::read(signed char *, int) istream &istream::read(void *, int) int istream::ipfx(int) int istream::!
ipfx0() int istream::ipfx1() void istream::isfx() int istream::get() _G_size_t istream::gcount() istream &istream::putback(char) istream &istream::unget() istream &istream::unget(char) streambuf *istream::istreambuf() const istream &istream::operator>>(unsigned char *) istream &istream::operator>>(signed char *) istream &istream::operator>>(unsigned char &) istream &istream::operator>>(signed char &) istream &istream::operator>>(ios &(*)(ios &)) istream &istream::operator>>(istream &(*)(istream &)) iostream::iostream() istream::~istream() ostream::~ostream() _IO_istream_withassign &_IO_istream_withassign::operator=(_IO_istream_withassign &) _IO_ostream_withassign &_IO_ostream_withassign::operator=(_IO_ostream_withassign &) ios &dec(ios &) ios &hex(ios &) ios &oct(ios &) istream &WS(istream &) filebuf *fstreambase::rdbuf() const int fstreambase::is_open() const void fstreambase::setbuf(char *, int) int fstreambase::filedesc() fstreambase &fstreambase::raw() ifstream::ifstream()!
fstreambase::~fstreambase() ifstream::ifstream(int) ifstream::ifstream(int, char *, int) ifstream::ifstream(const char *, int, int) void ifstream::open(const char *, int, int) ofstream::ofstream() ofstream::ofstream(int) ofstream::ofstream(int, char *, int) ofstream::ofstream(const char *, int, int) void ofstream::open(const char *, int, int) fstream::fstream() iostream::~iostream() fstream::fstream(int) fstream::fstream(const char *, int, int) fstream::fstream(int, char *, int) void fstream::open(const char *, int, int) strstreambuf::strstreambuf() strstreambuf::strstreambuf(int) strstreambuf::strstreambuf(void *(*)(unsigned int), void (*)(void *)) strstreambuf::strstreambuf(char *, int, char *) strstreambuf::strstreambuf(unsigned char *, int, unsigned char *) strstreambuf::strstreambuf(const char *, int) strstreambuf::strstreambuf(const unsigned char *, int) strstreambuf::strstreambuf(signed char *, int, signed char *) strstreambuf::strstreambuf(const signed char *, int) in!
t strstreambuf::frozen() void strstreambuf::freeze(int) strstreambuf *strstreambase::rdbuf() strstreambase::strstreambase() ostrstream::ostrstream() strstreambase::~strstreambase() ostrstream::ostrstream(char *, int, int) _G_ssize_t ostrstream::pcount() char *ostrstream::str() void ostrstream::freeze(int) int ostrstream::frozen() strstream::strstream() strstream::strstream(char *, int, int) _G_ssize_t strstream::pcount() char *strstream::str() void strstream::freeze(int) int strstream::frozen() SFile::SFile() fstream::~fstream() int SFile::size() int SFile::setsize(int) PlotFile::PlotFile() ofstream::~ofstream() PlotFile::PlotFile(int) PlotFile::PlotFile(const char *, int, int) int abs(int) ostream &operator<<(ostream &, record &) ostream &operator<<(ostream &, record &) void t1() void t1() void t2() void t2() void t3() ifstream::~ifstream() void t3() {GC 5747k -> 1542k} {GC 5548k -> 2169k} void t4() void t4() {GC 6258k -> 3421k} {GC 5382k -> 3808k} {GC 5383k -> 3871k} {GC 647!
3k -> 3928k} {GC 5907k -> 3965k} {GC 6437k -> 3993k} {GC 5610k -> 3121k} void t5() SFile::~SFile() void t5() void t6() PlotFile::~PlotFile() void t6() {GC 5606k -> 2071k} void t7() ostrstream::~ostrstream() istrstream::~istrstream() strstream::~strstream() void t7() {GC 5808k -> 2991k} {GC 29318k -> 19959k} {GC 26766k -> 21618k} {GC 28417k -> 22528k} {GC 29356k -> 23947k} {GC 35762k -> 11055k} {GC 15314k -> 7649k} void t8() void t8() void t9() void t9() void t10() void t10() {GC 10095k -> 3312k} {GC 5363k -> 4269k} {GC 5561k -> 4220k} void t11() void t11() void show_int(long int) void show_int(long int) void t12() void t12() int main(int, char **) int main(int, char **) {GC 5647k -> 3222k} {GC 3553k -> 1235k}
Tree Number Bytes % Total
error_mark 1 16 0.001
identifier_node 2391 149k 14.041
tree_list 2403 75k 7.056
tree_vec 246 15k 1.448
block 505 15k 1.483
void_type 4 512 0.047
integer_type 197 24k 2.314
real_type 5 640 0.059
complex_type 4 512 0.047
vector_type 5 640 0.059
enumeral_type 8 1024 0.094
boolean_type 1 128 0.012
pointer_type 341 42k 4.005
reference_type 89 11k 1.045
method_type 425 53k 4.991
array_type 180 22k 2.114
record_type 84 10k 0.987
function_type 429 53k 5.038
lang_type 2 256 0.023
integer_cst 1378 43k 4.046
string_cst 6 192 0.018
function_decl 787 98k 9.243
label_decl 49 6272 0.575
const_decl 42 5376 0.493
type_decl 242 30k 2.842
var_decl 86 10k 1.010
parm_decl 991 123k 11.639
result_decl 225 28k 2.643
field_decl 94 11k 1.104
namespace_decl 2 256 0.023
component_ref 537 16k 1.577
indirect_ref 491 15k 1.442
array_ref 74 2368 0.217
constructor 105 3360 0.308
compound_expr 36 1152 0.106
modify_expr 255 8160 0.749
init_expr 143 4576 0.420
cond_expr 40 1280 0.117
call_expr 332 10k 0.975
plus_expr 264 8448 0.775
minus_expr 40 1280 0.117
abs_expr 1 32 0.003
bit_ior_expr 14 448 0.041
bit_and_expr 56 1792 0.164
bit_not_expr 4 128 0.012
truth_andif_expr 6 192 0.018
truth_orif_expr 3 96 0.009
lt_expr 3 96 0.009
ge_expr 2 64 0.006
eq_expr 54 1728 0.159
ne_expr 119 3808 0.349
convert_expr 69 2208 0.203
nop_expr 984 30k 2.889
non_lvalue_expr 62 1984 0.182
save_expr 46 1472 0.135
addr_expr 503 15k 1.477
postincrement_expr 3 96 0.009
expr_stmt 383 11k 1.125
compound_stmt 304 9728 0.893
decl_stmt 49 1568 0.144
if_stmt 154 4928 0.452
return_stmt 177 5664 0.520
label_stmt 49 1568 0.144
scope_stmt 904 28k 2.654
stmt_expr 1 32 0.003
binding 833 26k 2.446
overload 232 3712 0.341
subobject 72 2304 0.211
ctor_stmt 70 1120 0.103
Total 18696 1064k
RTX Number Bytes % Total
code_label 1 64 0.146
const_int 2174 16k 39.650
const_double 21 672 1.532
pc 1 4 0.009
reg 18 288 0.657
mem 1006 15k 36.695
symbol_ref 1168 9344 21.302
cc0 1 4 0.009
Total 4390 42k
Log Allocated Used Overhead
2 4096 564 156
3 232k 31k 5336
4 328k 42k 4920
5 2384k 400k 25k
6 296k 220k 2664
7 928k 537k 7424
8 4096 1024 28
9 4096 512 28
10 4096 1024 28
Total 4188k 1235k 45k
Execution times (seconds)
garbage collection : 20.27 ( 3%) usr 20.26 (35%) sys 644.72 (-40%) wall
lexical analysis : 0.42 ( 0%) usr 0.14 ( 0%) sys 1.93 ( 0%) wall
parser : 4.87 ( 1%) usr 0.32 ( 1%) sys 17.20 (-1%) wall
expand : 1.94 ( 0%) usr 0.15 ( 0%) sys 7.35 ( 0%) wall
varconst : 0.02 ( 0%) usr 0.01 ( 0%) sys 0.02 ( 0%) wall
integration : 0.59 ( 0%) usr 0.02 ( 0%) sys 2.10 ( 0%) wall
jump : 18.77 ( 3%) usr 3.07 ( 5%) sys 108.24 (-7%) wall
CSE : 5.34 ( 1%) usr 0.07 ( 0%) sys 15.64 (-1%) wall
global CSE : 39.82 ( 7%) usr 3.39 ( 6%) sys 128.55 (-8%) wall
loop analysis : 29.13 ( 5%) usr 1.05 ( 2%) sys 79.13 (-5%) wall
CSE 2 : 298.10 (49%) usr 6.83 (12%) sys 781.11 (-49%) wall
flow analysis : 24.83 ( 4%) usr 3.24 ( 6%) sys 88.25 (-6%) wall
combiner : 8.05 ( 1%) usr 1.67 ( 3%) sys 39.29 (-2%) wall
if-conversion : 4.79 ( 1%) usr 0.35 ( 1%) sys 15.37 (-1%) wall
regmove : 3.89 ( 1%) usr 0.06 ( 0%) sys 7.54 ( 0%) wall
scheduling : 54.90 ( 9%) usr 7.65 (13%) sys 288.24 (-18%) wall
local alloc : 16.06 ( 3%) usr 2.22 ( 4%) sys 114.49 (-7%) wall
global alloc : 9.97 ( 2%) usr 2.38 ( 4%) sys 89.75 (-6%) wall
reload CSE regs : 13.37 ( 2%) usr 0.58 ( 1%) sys 40.54 (-3%) wall
flow 2 : 6.54 ( 1%) usr 0.63 ( 1%) sys 34.31 (-2%) wall
if-conversion 2 : 0.54 ( 0%) usr 0.02 ( 0%) sys 1.38 ( 0%) wall
scheduling 2 : 39.76 ( 7%) usr 2.63 ( 5%) sys 137.87 (-9%) wall
delay branch sched : 1.68 ( 0%) usr 0.16 ( 0%) sys 5.72 ( 0%) wall
reorder blocks : 0.01 ( 0%) usr 0.00 ( 0%) sys 0.14 ( 0%) wall
shorten branches : 0.79 ( 0%) usr 0.30 ( 1%) sys 8.79 (-1%) wall
final : 1.07 ( 0%) usr 0.39 ( 1%) sys 27.04 (-2%) wall
symout : 0.01 ( 0%) usr 0.02 ( 0%) sys 0.29 ( 0%) wall
rest of compilation : 2.10 ( 0%) usr 0.04 ( 0%) sys 5.09 ( 0%) wall
TOTAL : 607.74 57.77 -1603.-96
Tree Number Bytes % Total
error_mark 1 16 0.029
identifier_node 260 16k 30.312
tree_list 44 1408 2.565
void_type 2 256 0.466
integer_type 37 4736 8.627
real_type 3 384 0.700
complex_type 4 512 0.933
vector_type 5 640 1.166
pointer_type 13 1664 3.031
reference_type 1 128 0.233
array_type 12 1536 2.798
record_type 5 640 1.166
function_type 36 4608 8.394
integer_cst 100 3200 5.829
string_cst 2 64 0.117
function_decl 97 12k 22.617
type_decl 37 4736 8.627
var_decl 5 640 1.166
field_decl 5 640 1.166
addr_expr 1 32 0.058
Total 670 53k
RTX Number Bytes % Total
code_label 1 64 0.897
const_int 154 1232 17.265
const_double 21 672 9.417
pc 1 4 0.056
reg 10 160 2.242
mem 154 2464 34.529
symbol_ref 317 2536 35.538
cc0 1 4 0.056
Total 659 7136
Log Allocated Used Overhead
2 4096 108 156
3 8192 5104 184
4 16k 6896 240
5 16k 8416 176
6 20k 16k 180
7 36k 32k 288
Total 104k 69k 1224
More information about the Gcc-bugs
mailing list